commit e26c7b17f12fbdd8208975e2914add2209d1a24d Author: ModelHub XC Date: Sun Aug 9 23:27:32 2026 +0800 初始化项目，由ModelHub XC社区提供模型 Model: Neelectric/Llama-3.1-8B-Instruct_SFT_sciencev00.18 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..1ee0efb --- /dev/null +++ b/README.md @@ -0,0 +1,59 @@ +--- +base_model: meta-llama/Llama-3.1-8B-Instruct +datasets: Neelectric/Replay_0.02.MoT_science.wildguardmix.Llama3_4096toks +library_name: transformers +model_name: Llama-3.1-8B-Instruct_SFT_sciencev00.18 +tags: +- generated_from_trainer +- sft +- trl +- open-r1 +licence: license +--- + +# Model Card for Llama-3.1-8B-Instruct_SFT_sciencev00.18 + +This model is a fine-tuned version of [meta-llama/Llama-3.1-8B-Instruct](https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct) on the [Neelectric/Replay_0.02.MoT_science.wildguardmix.Llama3_4096toks](https://huggingface.co/datasets/Neelectric/Replay_0.02.MoT_science.wildguardmix.Llama3_4096toks) dataset. +It has been trained using [TRL](https://github.com/huggingface/trl). + +## Quick start + +```python +from transformers import pipeline + +question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?" +generator = pipeline("text-generation", model="Neelectric/Llama-3.1-8B-Instruct_SFT_sciencev00.18", device="cuda") +output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0] +print(output["generated_text"]) +``` + +## Training procedure + +[Visualize in Weights & Biases](https://wandb.ai/neelectric/open-r1_science/runs/60gxfqmy) + + +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..957dd92 --- /dev/null +++ b/all_results.json @@ -0,0 +1,8 @@ +{ + "total_flos": 2.0853122720147177e+19, + "train_loss": 0.9531466799229548, + "train_runtime": 21268.4838, + "train_samples": 145693, + "train_samples_per_second": 13.7, + "train_steps_per_second": 0.856 +} \ 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..2c14ad0 --- /dev/null +++ b/model-00001-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77a6e5df19507e053ad39a8ba35c33bd709d5d5e6ad4b360101836ba0c2e9d26 +size 4976698672 diff --git a/model-00002-of-00004.safetensors b/model-00002-of-00004.safetensors new file mode 100644 index 0000000..e49628b --- /dev/null +++ b/model-00002-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7746ecd1fbe0b58e562a038263ea58ffa7217812ffbdb8003a6927f34e2f1857 +size 4999802720 diff --git a/model-00003-of-00004.safetensors b/model-00003-of-00004.safetensors new file mode 100644 index 0000000..f3b0c08 --- /dev/null +++ b/model-00003-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:29c527d7bb63b06e9c629fbb404325af2719cf85902c6e03b6e2cd1c2d5cf9a6 +size 4915916176 diff --git a/model-00004-of-00004.safetensors b/model-00004-of-00004.safetensors new file mode 100644 index 0000000..1de78b6 --- /dev/null +++ b/model-00004-of-00004.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:15e30387b9bd147b4d7d708803de55f394ea585f2146c1c5edd45991dda6f93a +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..957dd92 --- /dev/null +++ b/train_results.json @@ -0,0 +1,8 @@ +{ + "total_flos": 2.0853122720147177e+19, + "train_loss": 0.9531466799229548, + "train_runtime": 21268.4838, + "train_samples": 145693, + "train_samples_per_second": 13.7, + "train_steps_per_second": 0.856 +} \ No newline at end of file diff --git a/trainer_state.json b/trainer_state.json new file mode 100644 index 0000000..2d1dc32 --- /dev/null +++ b/trainer_state.json @@ -0,0 +1,163951 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 2.0, + "eval_steps": 500, + "global_step": 18212, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.00010981770261366133, + "grad_norm": 7.300217151641846, + "learning_rate": 0.0, + "loss": 1.4841, + "mean_token_accuracy": 0.6132287979125977, + "num_tokens": 25099.0, + "step": 1 + }, + { + "epoch": 0.00021963540522732265, + "grad_norm": 6.589252948760986, + "learning_rate": 5.48847420417124e-10, + "loss": 1.4857, + "mean_token_accuracy": 0.6123285293579102, + "num_tokens": 54691.0, + "step": 2 + }, + { + "epoch": 0.000329453107840984, + "grad_norm": 6.737201690673828, + "learning_rate": 1.097694840834248e-09, + "loss": 1.4471, + "mean_token_accuracy": 0.6171982288360596, + "num_tokens": 81754.0, + "step": 3 + }, + { + "epoch": 0.0004392708104546453, + "grad_norm": 7.583634376525879, + "learning_rate": 1.646542261251372e-09, + "loss": 1.5596, + "mean_token_accuracy": 0.6015388369560242, + "num_tokens": 106869.0, + "step": 4 + }, + { + "epoch": 0.0005490885130683066, + "grad_norm": 7.94742488861084, + "learning_rate": 2.195389681668496e-09, + "loss": 1.5558, + "mean_token_accuracy": 0.6066004633903503, + "num_tokens": 128306.0, + "step": 5 + }, + { + "epoch": 0.000658906215681968, + "grad_norm": 7.764647006988525, + "learning_rate": 2.7442371020856205e-09, + "loss": 1.4749, + "mean_token_accuracy": 0.6179677844047546, + "num_tokens": 150830.0, + "step": 6 + }, + { + "epoch": 0.0007687239182956292, + "grad_norm": 7.081644535064697, + "learning_rate": 3.293084522502744e-09, + "loss": 1.4695, + "mean_token_accuracy": 0.6151657104492188, + "num_tokens": 177627.0, + "step": 7 + }, + { + "epoch": 0.0008785416209092906, + "grad_norm": 7.94363260269165, + "learning_rate": 3.841931942919868e-09, + "loss": 1.5503, + "mean_token_accuracy": 0.595615565776825, + "num_tokens": 200623.0, + "step": 8 + }, + { + "epoch": 0.000988359323522952, + "grad_norm": 6.677658557891846, + "learning_rate": 4.390779363336992e-09, + "loss": 1.3833, + "mean_token_accuracy": 0.6345776915550232, + "num_tokens": 227816.0, + "step": 9 + }, + { + "epoch": 0.0010981770261366132, + "grad_norm": 8.321730613708496, + "learning_rate": 4.939626783754116e-09, + "loss": 1.6381, + "mean_token_accuracy": 0.5835816860198975, + "num_tokens": 249911.0, + "step": 10 + }, + { + "epoch": 0.0012079947287502745, + "grad_norm": 6.200034141540527, + "learning_rate": 5.488474204171241e-09, + "loss": 1.5233, + "mean_token_accuracy": 0.6060580611228943, + "num_tokens": 285654.0, + "step": 11 + }, + { + "epoch": 0.001317812431363936, + "grad_norm": 7.45287561416626, + "learning_rate": 6.0373216245883644e-09, + "loss": 1.5377, + "mean_token_accuracy": 0.6079137325286865, + "num_tokens": 309084.0, + "step": 12 + }, + { + "epoch": 0.0014276301339775973, + "grad_norm": 6.8971357345581055, + "learning_rate": 6.586169045005488e-09, + "loss": 1.5, + "mean_token_accuracy": 0.612810492515564, + "num_tokens": 335790.0, + "step": 13 + }, + { + "epoch": 0.0015374478365912585, + "grad_norm": 7.3247857093811035, + "learning_rate": 7.135016465422612e-09, + "loss": 1.5266, + "mean_token_accuracy": 0.6071995496749878, + "num_tokens": 361392.0, + "step": 14 + }, + { + "epoch": 0.0016472655392049198, + "grad_norm": 6.274947643280029, + "learning_rate": 7.683863885839736e-09, + "loss": 1.5064, + "mean_token_accuracy": 0.6064631938934326, + "num_tokens": 396468.0, + "step": 15 + }, + { + "epoch": 0.0017570832418185812, + "grad_norm": 7.490630626678467, + "learning_rate": 8.23271130625686e-09, + "loss": 1.4102, + "mean_token_accuracy": 0.6237772703170776, + "num_tokens": 420520.0, + "step": 16 + }, + { + "epoch": 0.0018669009444322424, + "grad_norm": 7.133952617645264, + "learning_rate": 8.781558726673984e-09, + "loss": 1.4654, + "mean_token_accuracy": 0.60985267162323, + "num_tokens": 445405.0, + "step": 17 + }, + { + "epoch": 0.001976718647045904, + "grad_norm": 8.651247024536133, + "learning_rate": 9.330406147091108e-09, + "loss": 1.5173, + "mean_token_accuracy": 0.6092146635055542, + "num_tokens": 465540.0, + "step": 18 + }, + { + "epoch": 0.002086536349659565, + "grad_norm": 7.914252758026123, + "learning_rate": 9.879253567508231e-09, + "loss": 1.5013, + "mean_token_accuracy": 0.6079566478729248, + "num_tokens": 487106.0, + "step": 19 + }, + { + "epoch": 0.0021963540522732263, + "grad_norm": 7.974539279937744, + "learning_rate": 1.0428100987925357e-08, + "loss": 1.5382, + "mean_token_accuracy": 0.6065188050270081, + "num_tokens": 509913.0, + "step": 20 + }, + { + "epoch": 0.002306171754886888, + "grad_norm": 7.789461612701416, + "learning_rate": 1.0976948408342482e-08, + "loss": 1.6002, + "mean_token_accuracy": 0.5907433032989502, + "num_tokens": 533075.0, + "step": 21 + }, + { + "epoch": 0.002415989457500549, + "grad_norm": 7.048075199127197, + "learning_rate": 1.1525795828759604e-08, + "loss": 1.4033, + "mean_token_accuracy": 0.6364850997924805, + "num_tokens": 558198.0, + "step": 22 + }, + { + "epoch": 0.0025258071601142102, + "grad_norm": 6.6210808753967285, + "learning_rate": 1.2074643249176729e-08, + "loss": 1.5045, + "mean_token_accuracy": 0.6114836931228638, + "num_tokens": 586813.0, + "step": 23 + }, + { + "epoch": 0.002635624862727872, + "grad_norm": 7.112679481506348, + "learning_rate": 1.2623490669593852e-08, + "loss": 1.5105, + "mean_token_accuracy": 0.6043601036071777, + "num_tokens": 612556.0, + "step": 24 + }, + { + "epoch": 0.002745442565341533, + "grad_norm": 6.355383396148682, + "learning_rate": 1.3172338090010976e-08, + "loss": 1.3612, + "mean_token_accuracy": 0.6446248292922974, + "num_tokens": 641057.0, + "step": 25 + }, + { + "epoch": 0.0028552602679551946, + "grad_norm": 6.30284309387207, + "learning_rate": 1.3721185510428101e-08, + "loss": 1.5176, + "mean_token_accuracy": 0.6030095815658569, + "num_tokens": 673639.0, + "step": 26 + }, + { + "epoch": 0.0029650779705688557, + "grad_norm": 6.257452964782715, + "learning_rate": 1.4270032930845225e-08, + "loss": 1.5201, + "mean_token_accuracy": 0.6015155911445618, + "num_tokens": 705135.0, + "step": 27 + }, + { + "epoch": 0.003074895673182517, + "grad_norm": 7.045098304748535, + "learning_rate": 1.4818880351262348e-08, + "loss": 1.5011, + "mean_token_accuracy": 0.6112154126167297, + "num_tokens": 730901.0, + "step": 28 + }, + { + "epoch": 0.0031847133757961785, + "grad_norm": 8.116623878479004, + "learning_rate": 1.536772777167947e-08, + "loss": 1.5567, + "mean_token_accuracy": 0.6028671264648438, + "num_tokens": 749685.0, + "step": 29 + }, + { + "epoch": 0.0032945310784098397, + "grad_norm": 7.39232063293457, + "learning_rate": 1.5916575192096597e-08, + "loss": 1.5084, + "mean_token_accuracy": 0.6177272796630859, + "num_tokens": 773213.0, + "step": 30 + }, + { + "epoch": 0.003404348781023501, + "grad_norm": 6.968527317047119, + "learning_rate": 1.646542261251372e-08, + "loss": 1.5111, + "mean_token_accuracy": 0.6118353605270386, + "num_tokens": 801231.0, + "step": 31 + }, + { + "epoch": 0.0035141664836371624, + "grad_norm": 6.88621711730957, + "learning_rate": 1.7014270032930844e-08, + "loss": 1.4985, + "mean_token_accuracy": 0.6087011098861694, + "num_tokens": 830231.0, + "step": 32 + }, + { + "epoch": 0.0036239841862508236, + "grad_norm": 7.369091987609863, + "learning_rate": 1.756311745334797e-08, + "loss": 1.452, + "mean_token_accuracy": 0.6208947896957397, + "num_tokens": 855188.0, + "step": 33 + }, + { + "epoch": 0.0037338018888644848, + "grad_norm": 6.989710330963135, + "learning_rate": 1.811196487376509e-08, + "loss": 1.4433, + "mean_token_accuracy": 0.618872344493866, + "num_tokens": 881202.0, + "step": 34 + }, + { + "epoch": 0.0038436195914781464, + "grad_norm": 6.39633846282959, + "learning_rate": 1.8660812294182216e-08, + "loss": 1.4574, + "mean_token_accuracy": 0.615617036819458, + "num_tokens": 911522.0, + "step": 35 + }, + { + "epoch": 0.003953437294091808, + "grad_norm": 6.205531597137451, + "learning_rate": 1.920965971459934e-08, + "loss": 1.4333, + "mean_token_accuracy": 0.6314854025840759, + "num_tokens": 940955.0, + "step": 36 + }, + { + "epoch": 0.004063254996705469, + "grad_norm": 6.841545581817627, + "learning_rate": 1.9758507135016463e-08, + "loss": 1.5082, + "mean_token_accuracy": 0.6079716086387634, + "num_tokens": 969029.0, + "step": 37 + }, + { + "epoch": 0.00417307269931913, + "grad_norm": 8.310525894165039, + "learning_rate": 2.030735455543359e-08, + "loss": 1.5401, + "mean_token_accuracy": 0.5996425151824951, + "num_tokens": 990256.0, + "step": 38 + }, + { + "epoch": 0.004282890401932792, + "grad_norm": 7.998595714569092, + "learning_rate": 2.0856201975850713e-08, + "loss": 1.502, + "mean_token_accuracy": 0.616737425327301, + "num_tokens": 1012299.0, + "step": 39 + }, + { + "epoch": 0.004392708104546453, + "grad_norm": 7.245059013366699, + "learning_rate": 2.1405049396267835e-08, + "loss": 1.4845, + "mean_token_accuracy": 0.6128872632980347, + "num_tokens": 1038393.0, + "step": 40 + }, + { + "epoch": 0.004502525807160114, + "grad_norm": 6.464139461517334, + "learning_rate": 2.1953896816684964e-08, + "loss": 1.5878, + "mean_token_accuracy": 0.5887458324432373, + "num_tokens": 1070593.0, + "step": 41 + }, + { + "epoch": 0.004612343509773776, + "grad_norm": 8.003179550170898, + "learning_rate": 2.2502744237102085e-08, + "loss": 1.5, + "mean_token_accuracy": 0.6205024719238281, + "num_tokens": 1092494.0, + "step": 42 + }, + { + "epoch": 0.0047221612123874365, + "grad_norm": 7.175567150115967, + "learning_rate": 2.3051591657519207e-08, + "loss": 1.3971, + "mean_token_accuracy": 0.6405119895935059, + "num_tokens": 1116768.0, + "step": 43 + }, + { + "epoch": 0.004831978915001098, + "grad_norm": 8.04098129272461, + "learning_rate": 2.3600439077936336e-08, + "loss": 1.4997, + "mean_token_accuracy": 0.6148420572280884, + "num_tokens": 1138530.0, + "step": 44 + }, + { + "epoch": 0.00494179661761476, + "grad_norm": 8.159924507141113, + "learning_rate": 2.4149286498353458e-08, + "loss": 1.5912, + "mean_token_accuracy": 0.5890048146247864, + "num_tokens": 1159960.0, + "step": 45 + }, + { + "epoch": 0.0050516143202284204, + "grad_norm": 7.005256175994873, + "learning_rate": 2.469813391877058e-08, + "loss": 1.4909, + "mean_token_accuracy": 0.607750654220581, + "num_tokens": 1187179.0, + "step": 46 + }, + { + "epoch": 0.005161432022842082, + "grad_norm": 7.529395580291748, + "learning_rate": 2.5246981339187705e-08, + "loss": 1.3737, + "mean_token_accuracy": 0.637641429901123, + "num_tokens": 1209353.0, + "step": 47 + }, + { + "epoch": 0.005271249725455744, + "grad_norm": 7.030339241027832, + "learning_rate": 2.579582875960483e-08, + "loss": 1.5163, + "mean_token_accuracy": 0.6043077111244202, + "num_tokens": 1237163.0, + "step": 48 + }, + { + "epoch": 0.005381067428069404, + "grad_norm": 6.289010524749756, + "learning_rate": 2.6344676180021952e-08, + "loss": 1.4034, + "mean_token_accuracy": 0.6295833587646484, + "num_tokens": 1267359.0, + "step": 49 + }, + { + "epoch": 0.005490885130683066, + "grad_norm": 7.012529373168945, + "learning_rate": 2.6893523600439077e-08, + "loss": 1.4184, + "mean_token_accuracy": 0.6309919357299805, + "num_tokens": 1294502.0, + "step": 50 + }, + { + "epoch": 0.0056007028332967276, + "grad_norm": 8.415987968444824, + "learning_rate": 2.7442371020856202e-08, + "loss": 1.4829, + "mean_token_accuracy": 0.6100214719772339, + "num_tokens": 1313738.0, + "step": 51 + }, + { + "epoch": 0.005710520535910389, + "grad_norm": 7.258707046508789, + "learning_rate": 2.7991218441273324e-08, + "loss": 1.4357, + "mean_token_accuracy": 0.6211686730384827, + "num_tokens": 1337335.0, + "step": 52 + }, + { + "epoch": 0.00582033823852405, + "grad_norm": 7.5146403312683105, + "learning_rate": 2.854006586169045e-08, + "loss": 1.4699, + "mean_token_accuracy": 0.6220319271087646, + "num_tokens": 1359920.0, + "step": 53 + }, + { + "epoch": 0.0059301559411377115, + "grad_norm": 7.393195629119873, + "learning_rate": 2.9088913282107574e-08, + "loss": 1.5112, + "mean_token_accuracy": 0.6088619232177734, + "num_tokens": 1384313.0, + "step": 54 + }, + { + "epoch": 0.006039973643751373, + "grad_norm": 6.48404598236084, + "learning_rate": 2.9637760702524696e-08, + "loss": 1.4629, + "mean_token_accuracy": 0.61573725938797, + "num_tokens": 1415664.0, + "step": 55 + }, + { + "epoch": 0.006149791346365034, + "grad_norm": 5.997020244598389, + "learning_rate": 3.018660812294182e-08, + "loss": 1.5736, + "mean_token_accuracy": 0.5959450602531433, + "num_tokens": 1454133.0, + "step": 56 + }, + { + "epoch": 0.006259609048978695, + "grad_norm": 7.868096351623535, + "learning_rate": 3.073545554335894e-08, + "loss": 1.5338, + "mean_token_accuracy": 0.5999817848205566, + "num_tokens": 1476023.0, + "step": 57 + }, + { + "epoch": 0.006369426751592357, + "grad_norm": 7.971103191375732, + "learning_rate": 3.128430296377607e-08, + "loss": 1.4886, + "mean_token_accuracy": 0.6128200888633728, + "num_tokens": 1497744.0, + "step": 58 + }, + { + "epoch": 0.006479244454206018, + "grad_norm": 7.506596088409424, + "learning_rate": 3.1833150384193193e-08, + "loss": 1.4892, + "mean_token_accuracy": 0.6171019077301025, + "num_tokens": 1522143.0, + "step": 59 + }, + { + "epoch": 0.006589062156819679, + "grad_norm": 6.55207633972168, + "learning_rate": 3.2381997804610315e-08, + "loss": 1.4589, + "mean_token_accuracy": 0.6231217980384827, + "num_tokens": 1551267.0, + "step": 60 + }, + { + "epoch": 0.006698879859433341, + "grad_norm": 6.870192050933838, + "learning_rate": 3.293084522502744e-08, + "loss": 1.4876, + "mean_token_accuracy": 0.6067413091659546, + "num_tokens": 1578878.0, + "step": 61 + }, + { + "epoch": 0.006808697562047002, + "grad_norm": 6.896904468536377, + "learning_rate": 3.3479692645444566e-08, + "loss": 1.5111, + "mean_token_accuracy": 0.6176468729972839, + "num_tokens": 1606620.0, + "step": 62 + }, + { + "epoch": 0.006918515264660663, + "grad_norm": 8.743321418762207, + "learning_rate": 3.402854006586169e-08, + "loss": 1.464, + "mean_token_accuracy": 0.6236720681190491, + "num_tokens": 1626148.0, + "step": 63 + }, + { + "epoch": 0.007028332967274325, + "grad_norm": 7.8488569259643555, + "learning_rate": 3.4577387486278816e-08, + "loss": 1.5113, + "mean_token_accuracy": 0.6107940077781677, + "num_tokens": 1647385.0, + "step": 64 + }, + { + "epoch": 0.007138150669887986, + "grad_norm": 7.486027717590332, + "learning_rate": 3.512623490669594e-08, + "loss": 1.5884, + "mean_token_accuracy": 0.6027314066886902, + "num_tokens": 1671511.0, + "step": 65 + }, + { + "epoch": 0.007247968372501647, + "grad_norm": 7.019283771514893, + "learning_rate": 3.567508232711306e-08, + "loss": 1.4844, + "mean_token_accuracy": 0.6039084196090698, + "num_tokens": 1695449.0, + "step": 66 + }, + { + "epoch": 0.007357786075115309, + "grad_norm": 6.765707969665527, + "learning_rate": 3.622392974753018e-08, + "loss": 1.493, + "mean_token_accuracy": 0.611631453037262, + "num_tokens": 1721678.0, + "step": 67 + }, + { + "epoch": 0.0074676037777289695, + "grad_norm": 7.713114261627197, + "learning_rate": 3.677277716794731e-08, + "loss": 1.5554, + "mean_token_accuracy": 0.6026549339294434, + "num_tokens": 1744979.0, + "step": 68 + }, + { + "epoch": 0.007577421480342631, + "grad_norm": 6.841897964477539, + "learning_rate": 3.732162458836443e-08, + "loss": 1.512, + "mean_token_accuracy": 0.6029990315437317, + "num_tokens": 1772294.0, + "step": 69 + }, + { + "epoch": 0.007687239182956293, + "grad_norm": 6.1020188331604, + "learning_rate": 3.787047200878156e-08, + "loss": 1.3884, + "mean_token_accuracy": 0.6362130641937256, + "num_tokens": 1803834.0, + "step": 70 + }, + { + "epoch": 0.007797056885569954, + "grad_norm": 7.298198223114014, + "learning_rate": 3.841931942919868e-08, + "loss": 1.5748, + "mean_token_accuracy": 0.603700578212738, + "num_tokens": 1826961.0, + "step": 71 + }, + { + "epoch": 0.007906874588183616, + "grad_norm": 7.283111095428467, + "learning_rate": 3.8968166849615804e-08, + "loss": 1.5969, + "mean_token_accuracy": 0.5894321799278259, + "num_tokens": 1852992.0, + "step": 72 + }, + { + "epoch": 0.008016692290797276, + "grad_norm": 7.222440719604492, + "learning_rate": 3.9517014270032926e-08, + "loss": 1.5592, + "mean_token_accuracy": 0.5943928956985474, + "num_tokens": 1877299.0, + "step": 73 + }, + { + "epoch": 0.008126509993410937, + "grad_norm": 7.264228820800781, + "learning_rate": 4.006586169045005e-08, + "loss": 1.4357, + "mean_token_accuracy": 0.6176801919937134, + "num_tokens": 1900896.0, + "step": 74 + }, + { + "epoch": 0.008236327696024599, + "grad_norm": 7.492104530334473, + "learning_rate": 4.061470911086718e-08, + "loss": 1.5373, + "mean_token_accuracy": 0.599504292011261, + "num_tokens": 1923790.0, + "step": 75 + }, + { + "epoch": 0.00834614539863826, + "grad_norm": 6.757571697235107, + "learning_rate": 4.1163556531284305e-08, + "loss": 1.4626, + "mean_token_accuracy": 0.6147168278694153, + "num_tokens": 1951558.0, + "step": 76 + }, + { + "epoch": 0.008455963101251922, + "grad_norm": 6.347597122192383, + "learning_rate": 4.1712403951701427e-08, + "loss": 1.4707, + "mean_token_accuracy": 0.6251159906387329, + "num_tokens": 1979887.0, + "step": 77 + }, + { + "epoch": 0.008565780803865584, + "grad_norm": 6.938321590423584, + "learning_rate": 4.226125137211855e-08, + "loss": 1.4736, + "mean_token_accuracy": 0.6187638640403748, + "num_tokens": 2004159.0, + "step": 78 + }, + { + "epoch": 0.008675598506479244, + "grad_norm": 6.393320560455322, + "learning_rate": 4.281009879253567e-08, + "loss": 1.5003, + "mean_token_accuracy": 0.6025761365890503, + "num_tokens": 2034234.0, + "step": 79 + }, + { + "epoch": 0.008785416209092905, + "grad_norm": 7.053067207336426, + "learning_rate": 4.335894621295279e-08, + "loss": 1.4296, + "mean_token_accuracy": 0.6265895366668701, + "num_tokens": 2057969.0, + "step": 80 + }, + { + "epoch": 0.008895233911706567, + "grad_norm": 6.3645734786987305, + "learning_rate": 4.390779363336993e-08, + "loss": 1.5251, + "mean_token_accuracy": 0.6056689023971558, + "num_tokens": 2088734.0, + "step": 81 + }, + { + "epoch": 0.009005051614320228, + "grad_norm": 7.069924831390381, + "learning_rate": 4.445664105378705e-08, + "loss": 1.5536, + "mean_token_accuracy": 0.5905734896659851, + "num_tokens": 2113722.0, + "step": 82 + }, + { + "epoch": 0.00911486931693389, + "grad_norm": 7.899157524108887, + "learning_rate": 4.500548847420417e-08, + "loss": 1.4538, + "mean_token_accuracy": 0.62293541431427, + "num_tokens": 2134293.0, + "step": 83 + }, + { + "epoch": 0.009224687019547552, + "grad_norm": 6.404735565185547, + "learning_rate": 4.555433589462129e-08, + "loss": 1.4544, + "mean_token_accuracy": 0.614185094833374, + "num_tokens": 2160459.0, + "step": 84 + }, + { + "epoch": 0.009334504722161213, + "grad_norm": 7.01681661605835, + "learning_rate": 4.6103183315038415e-08, + "loss": 1.5554, + "mean_token_accuracy": 0.5920899510383606, + "num_tokens": 2186118.0, + "step": 85 + }, + { + "epoch": 0.009444322424774873, + "grad_norm": 7.677351951599121, + "learning_rate": 4.6652030735455537e-08, + "loss": 1.462, + "mean_token_accuracy": 0.6207913756370544, + "num_tokens": 2207104.0, + "step": 86 + }, + { + "epoch": 0.009554140127388535, + "grad_norm": 6.574330806732178, + "learning_rate": 4.720087815587267e-08, + "loss": 1.4778, + "mean_token_accuracy": 0.6093591451644897, + "num_tokens": 2234836.0, + "step": 87 + }, + { + "epoch": 0.009663957830002196, + "grad_norm": 6.592723369598389, + "learning_rate": 4.7749725576289793e-08, + "loss": 1.4704, + "mean_token_accuracy": 0.6076290607452393, + "num_tokens": 2261225.0, + "step": 88 + }, + { + "epoch": 0.009773775532615858, + "grad_norm": 7.745190620422363, + "learning_rate": 4.8298572996706915e-08, + "loss": 1.419, + "mean_token_accuracy": 0.6234237551689148, + "num_tokens": 2281681.0, + "step": 89 + }, + { + "epoch": 0.00988359323522952, + "grad_norm": 5.853593349456787, + "learning_rate": 4.884742041712404e-08, + "loss": 1.3856, + "mean_token_accuracy": 0.633669376373291, + "num_tokens": 2309735.0, + "step": 90 + }, + { + "epoch": 0.009993410937843181, + "grad_norm": 6.382895469665527, + "learning_rate": 4.939626783754116e-08, + "loss": 1.4658, + "mean_token_accuracy": 0.6202864646911621, + "num_tokens": 2336566.0, + "step": 91 + }, + { + "epoch": 0.010103228640456841, + "grad_norm": 5.742077350616455, + "learning_rate": 4.994511525795828e-08, + "loss": 1.3999, + "mean_token_accuracy": 0.6311905980110168, + "num_tokens": 2364271.0, + "step": 92 + }, + { + "epoch": 0.010213046343070502, + "grad_norm": 5.952073574066162, + "learning_rate": 5.049396267837541e-08, + "loss": 1.3968, + "mean_token_accuracy": 0.6322518587112427, + "num_tokens": 2392828.0, + "step": 93 + }, + { + "epoch": 0.010322864045684164, + "grad_norm": 6.322819232940674, + "learning_rate": 5.104281009879254e-08, + "loss": 1.4609, + "mean_token_accuracy": 0.6104878783226013, + "num_tokens": 2419525.0, + "step": 94 + }, + { + "epoch": 0.010432681748297826, + "grad_norm": 6.462299346923828, + "learning_rate": 5.159165751920966e-08, + "loss": 1.4361, + "mean_token_accuracy": 0.6178503632545471, + "num_tokens": 2447368.0, + "step": 95 + }, + { + "epoch": 0.010542499450911487, + "grad_norm": 7.145196914672852, + "learning_rate": 5.214050493962678e-08, + "loss": 1.4817, + "mean_token_accuracy": 0.6092472672462463, + "num_tokens": 2470202.0, + "step": 96 + }, + { + "epoch": 0.010652317153525149, + "grad_norm": 6.037055492401123, + "learning_rate": 5.2689352360043903e-08, + "loss": 1.4633, + "mean_token_accuracy": 0.6104044914245605, + "num_tokens": 2499636.0, + "step": 97 + }, + { + "epoch": 0.010762134856138809, + "grad_norm": 7.089334964752197, + "learning_rate": 5.3238199780461025e-08, + "loss": 1.5008, + "mean_token_accuracy": 0.6112942695617676, + "num_tokens": 2521522.0, + "step": 98 + }, + { + "epoch": 0.01087195255875247, + "grad_norm": 6.957789421081543, + "learning_rate": 5.3787047200878154e-08, + "loss": 1.5021, + "mean_token_accuracy": 0.6071734428405762, + "num_tokens": 2545343.0, + "step": 99 + }, + { + "epoch": 0.010981770261366132, + "grad_norm": 6.612650394439697, + "learning_rate": 5.433589462129528e-08, + "loss": 1.5006, + "mean_token_accuracy": 0.603764533996582, + "num_tokens": 2570875.0, + "step": 100 + }, + { + "epoch": 0.011091587963979794, + "grad_norm": 6.62785530090332, + "learning_rate": 5.4884742041712404e-08, + "loss": 1.4808, + "mean_token_accuracy": 0.6163198351860046, + "num_tokens": 2596653.0, + "step": 101 + }, + { + "epoch": 0.011201405666593455, + "grad_norm": 6.656164646148682, + "learning_rate": 5.5433589462129526e-08, + "loss": 1.4754, + "mean_token_accuracy": 0.614667534828186, + "num_tokens": 2621831.0, + "step": 102 + }, + { + "epoch": 0.011311223369207117, + "grad_norm": 7.292792320251465, + "learning_rate": 5.598243688254665e-08, + "loss": 1.5902, + "mean_token_accuracy": 0.5945414304733276, + "num_tokens": 2644785.0, + "step": 103 + }, + { + "epoch": 0.011421041071820778, + "grad_norm": 8.125263214111328, + "learning_rate": 5.653128430296377e-08, + "loss": 1.5069, + "mean_token_accuracy": 0.6075853109359741, + "num_tokens": 2663936.0, + "step": 104 + }, + { + "epoch": 0.011530858774434438, + "grad_norm": 6.995969295501709, + "learning_rate": 5.70801317233809e-08, + "loss": 1.5256, + "mean_token_accuracy": 0.6036872863769531, + "num_tokens": 2688344.0, + "step": 105 + }, + { + "epoch": 0.0116406764770481, + "grad_norm": 6.630096912384033, + "learning_rate": 5.762897914379802e-08, + "loss": 1.3863, + "mean_token_accuracy": 0.628242015838623, + "num_tokens": 2712702.0, + "step": 106 + }, + { + "epoch": 0.011750494179661761, + "grad_norm": 7.639283180236816, + "learning_rate": 5.817782656421515e-08, + "loss": 1.3793, + "mean_token_accuracy": 0.6352887153625488, + "num_tokens": 2732437.0, + "step": 107 + }, + { + "epoch": 0.011860311882275423, + "grad_norm": 8.158872604370117, + "learning_rate": 5.872667398463227e-08, + "loss": 1.4228, + "mean_token_accuracy": 0.6309449076652527, + "num_tokens": 2751200.0, + "step": 108 + }, + { + "epoch": 0.011970129584889085, + "grad_norm": 6.876063823699951, + "learning_rate": 5.927552140504939e-08, + "loss": 1.4303, + "mean_token_accuracy": 0.6215893030166626, + "num_tokens": 2776168.0, + "step": 109 + }, + { + "epoch": 0.012079947287502746, + "grad_norm": 7.522066116333008, + "learning_rate": 5.982436882546651e-08, + "loss": 1.4829, + "mean_token_accuracy": 0.6148393154144287, + "num_tokens": 2798273.0, + "step": 110 + }, + { + "epoch": 0.012189764990116406, + "grad_norm": 6.509666919708252, + "learning_rate": 6.037321624588364e-08, + "loss": 1.5074, + "mean_token_accuracy": 0.6085059642791748, + "num_tokens": 2827158.0, + "step": 111 + }, + { + "epoch": 0.012299582692730068, + "grad_norm": 6.436208248138428, + "learning_rate": 6.092206366630077e-08, + "loss": 1.4382, + "mean_token_accuracy": 0.6149194240570068, + "num_tokens": 2854571.0, + "step": 112 + }, + { + "epoch": 0.01240940039534373, + "grad_norm": 7.045963287353516, + "learning_rate": 6.147091108671789e-08, + "loss": 1.4127, + "mean_token_accuracy": 0.6222429275512695, + "num_tokens": 2877878.0, + "step": 113 + }, + { + "epoch": 0.01251921809795739, + "grad_norm": 6.991760730743408, + "learning_rate": 6.201975850713501e-08, + "loss": 1.3674, + "mean_token_accuracy": 0.6376074552536011, + "num_tokens": 2901543.0, + "step": 114 + }, + { + "epoch": 0.012629035800571052, + "grad_norm": 6.352198123931885, + "learning_rate": 6.256860592755214e-08, + "loss": 1.433, + "mean_token_accuracy": 0.6203060150146484, + "num_tokens": 2927917.0, + "step": 115 + }, + { + "epoch": 0.012738853503184714, + "grad_norm": 6.198991775512695, + "learning_rate": 6.311745334796927e-08, + "loss": 1.3775, + "mean_token_accuracy": 0.6289072632789612, + "num_tokens": 2956723.0, + "step": 116 + }, + { + "epoch": 0.012848671205798374, + "grad_norm": 7.4495954513549805, + "learning_rate": 6.366630076838639e-08, + "loss": 1.4269, + "mean_token_accuracy": 0.616727888584137, + "num_tokens": 2978425.0, + "step": 117 + }, + { + "epoch": 0.012958488908412035, + "grad_norm": 6.915719032287598, + "learning_rate": 6.421514818880352e-08, + "loss": 1.4557, + "mean_token_accuracy": 0.6091991662979126, + "num_tokens": 3002328.0, + "step": 118 + }, + { + "epoch": 0.013068306611025697, + "grad_norm": 7.106113433837891, + "learning_rate": 6.476399560922063e-08, + "loss": 1.4118, + "mean_token_accuracy": 0.6290383338928223, + "num_tokens": 3024949.0, + "step": 119 + }, + { + "epoch": 0.013178124313639359, + "grad_norm": 6.216148853302002, + "learning_rate": 6.531284302963776e-08, + "loss": 1.453, + "mean_token_accuracy": 0.6123758554458618, + "num_tokens": 3053527.0, + "step": 120 + }, + { + "epoch": 0.01328794201625302, + "grad_norm": 5.501377105712891, + "learning_rate": 6.586169045005487e-08, + "loss": 1.2736, + "mean_token_accuracy": 0.6522762775421143, + "num_tokens": 3080578.0, + "step": 121 + }, + { + "epoch": 0.013397759718866682, + "grad_norm": 5.591029167175293, + "learning_rate": 6.6410537870472e-08, + "loss": 1.4868, + "mean_token_accuracy": 0.605283260345459, + "num_tokens": 3110672.0, + "step": 122 + }, + { + "epoch": 0.013507577421480343, + "grad_norm": 6.012486457824707, + "learning_rate": 6.695938529088913e-08, + "loss": 1.3919, + "mean_token_accuracy": 0.6284012198448181, + "num_tokens": 3135619.0, + "step": 123 + }, + { + "epoch": 0.013617395124094003, + "grad_norm": 6.087145805358887, + "learning_rate": 6.750823271130625e-08, + "loss": 1.3764, + "mean_token_accuracy": 0.6294901967048645, + "num_tokens": 3159336.0, + "step": 124 + }, + { + "epoch": 0.013727212826707665, + "grad_norm": 5.536701679229736, + "learning_rate": 6.805708013172338e-08, + "loss": 1.4075, + "mean_token_accuracy": 0.6168534159660339, + "num_tokens": 3187783.0, + "step": 125 + }, + { + "epoch": 0.013837030529321327, + "grad_norm": 5.2248125076293945, + "learning_rate": 6.86059275521405e-08, + "loss": 1.358, + "mean_token_accuracy": 0.6304128766059875, + "num_tokens": 3214851.0, + "step": 126 + }, + { + "epoch": 0.013946848231934988, + "grad_norm": 5.933658599853516, + "learning_rate": 6.915477497255763e-08, + "loss": 1.316, + "mean_token_accuracy": 0.6463679075241089, + "num_tokens": 3236610.0, + "step": 127 + }, + { + "epoch": 0.01405666593454865, + "grad_norm": 5.61740255355835, + "learning_rate": 6.970362239297475e-08, + "loss": 1.4131, + "mean_token_accuracy": 0.6190099716186523, + "num_tokens": 3261584.0, + "step": 128 + }, + { + "epoch": 0.014166483637162311, + "grad_norm": 5.64325475692749, + "learning_rate": 7.025246981339188e-08, + "loss": 1.4832, + "mean_token_accuracy": 0.6059250235557556, + "num_tokens": 3285343.0, + "step": 129 + }, + { + "epoch": 0.014276301339775971, + "grad_norm": 5.317951679229736, + "learning_rate": 7.0801317233809e-08, + "loss": 1.3389, + "mean_token_accuracy": 0.6471266746520996, + "num_tokens": 3311747.0, + "step": 130 + }, + { + "epoch": 0.014386119042389633, + "grad_norm": 4.80671501159668, + "learning_rate": 7.135016465422612e-08, + "loss": 1.3959, + "mean_token_accuracy": 0.6288163065910339, + "num_tokens": 3343075.0, + "step": 131 + }, + { + "epoch": 0.014495936745003294, + "grad_norm": 5.304012298583984, + "learning_rate": 7.189901207464325e-08, + "loss": 1.4647, + "mean_token_accuracy": 0.6099487543106079, + "num_tokens": 3369663.0, + "step": 132 + }, + { + "epoch": 0.014605754447616956, + "grad_norm": 5.3747100830078125, + "learning_rate": 7.244785949506036e-08, + "loss": 1.4435, + "mean_token_accuracy": 0.6002536416053772, + "num_tokens": 3393521.0, + "step": 133 + }, + { + "epoch": 0.014715572150230618, + "grad_norm": 4.572581768035889, + "learning_rate": 7.299670691547749e-08, + "loss": 1.4284, + "mean_token_accuracy": 0.6164723038673401, + "num_tokens": 3427249.0, + "step": 134 + }, + { + "epoch": 0.01482538985284428, + "grad_norm": 5.139473915100098, + "learning_rate": 7.354555433589462e-08, + "loss": 1.3729, + "mean_token_accuracy": 0.6253637671470642, + "num_tokens": 3453888.0, + "step": 135 + }, + { + "epoch": 0.014935207555457939, + "grad_norm": 5.578888416290283, + "learning_rate": 7.409440175631174e-08, + "loss": 1.4252, + "mean_token_accuracy": 0.6098592281341553, + "num_tokens": 3477031.0, + "step": 136 + }, + { + "epoch": 0.0150450252580716, + "grad_norm": 5.437226295471191, + "learning_rate": 7.464324917672886e-08, + "loss": 1.4551, + "mean_token_accuracy": 0.6097795963287354, + "num_tokens": 3501161.0, + "step": 137 + }, + { + "epoch": 0.015154842960685262, + "grad_norm": 5.0031514167785645, + "learning_rate": 7.519209659714599e-08, + "loss": 1.4162, + "mean_token_accuracy": 0.6102628111839294, + "num_tokens": 3528654.0, + "step": 138 + }, + { + "epoch": 0.015264660663298924, + "grad_norm": 4.552621841430664, + "learning_rate": 7.574094401756312e-08, + "loss": 1.4193, + "mean_token_accuracy": 0.6214172840118408, + "num_tokens": 3559002.0, + "step": 139 + }, + { + "epoch": 0.015374478365912585, + "grad_norm": 4.93706750869751, + "learning_rate": 7.628979143798024e-08, + "loss": 1.3571, + "mean_token_accuracy": 0.6340728998184204, + "num_tokens": 3584776.0, + "step": 140 + }, + { + "epoch": 0.015484296068526247, + "grad_norm": 5.215694427490234, + "learning_rate": 7.683863885839736e-08, + "loss": 1.3169, + "mean_token_accuracy": 0.6371525526046753, + "num_tokens": 3607232.0, + "step": 141 + }, + { + "epoch": 0.015594113771139909, + "grad_norm": 5.364761829376221, + "learning_rate": 7.738748627881449e-08, + "loss": 1.3276, + "mean_token_accuracy": 0.6309010982513428, + "num_tokens": 3630150.0, + "step": 142 + }, + { + "epoch": 0.01570393147375357, + "grad_norm": 5.227049827575684, + "learning_rate": 7.793633369923161e-08, + "loss": 1.3628, + "mean_token_accuracy": 0.6412457823753357, + "num_tokens": 3653505.0, + "step": 143 + }, + { + "epoch": 0.015813749176367232, + "grad_norm": 5.063607692718506, + "learning_rate": 7.848518111964874e-08, + "loss": 1.3016, + "mean_token_accuracy": 0.6348247528076172, + "num_tokens": 3677012.0, + "step": 144 + }, + { + "epoch": 0.01592356687898089, + "grad_norm": 5.130465030670166, + "learning_rate": 7.903402854006585e-08, + "loss": 1.403, + "mean_token_accuracy": 0.6204181909561157, + "num_tokens": 3700678.0, + "step": 145 + }, + { + "epoch": 0.01603338458159455, + "grad_norm": 5.620448112487793, + "learning_rate": 7.958287596048298e-08, + "loss": 1.4617, + "mean_token_accuracy": 0.6030761003494263, + "num_tokens": 3720859.0, + "step": 146 + }, + { + "epoch": 0.016143202284208215, + "grad_norm": 5.433720111846924, + "learning_rate": 8.01317233809001e-08, + "loss": 1.34, + "mean_token_accuracy": 0.6363637447357178, + "num_tokens": 3741905.0, + "step": 147 + }, + { + "epoch": 0.016253019986821875, + "grad_norm": 4.206226825714111, + "learning_rate": 8.068057080131722e-08, + "loss": 1.3589, + "mean_token_accuracy": 0.6257153749465942, + "num_tokens": 3773320.0, + "step": 148 + }, + { + "epoch": 0.016362837689435538, + "grad_norm": 5.190129280090332, + "learning_rate": 8.122941822173437e-08, + "loss": 1.3414, + "mean_token_accuracy": 0.638034462928772, + "num_tokens": 3795531.0, + "step": 149 + }, + { + "epoch": 0.016472655392049198, + "grad_norm": 4.571291446685791, + "learning_rate": 8.177826564215148e-08, + "loss": 1.3649, + "mean_token_accuracy": 0.6332225799560547, + "num_tokens": 3822919.0, + "step": 150 + }, + { + "epoch": 0.01658247309466286, + "grad_norm": 4.587471961975098, + "learning_rate": 8.232711306256861e-08, + "loss": 1.4303, + "mean_token_accuracy": 0.606686532497406, + "num_tokens": 3851659.0, + "step": 151 + }, + { + "epoch": 0.01669229079727652, + "grad_norm": 4.750936508178711, + "learning_rate": 8.287596048298572e-08, + "loss": 1.2879, + "mean_token_accuracy": 0.6490489840507507, + "num_tokens": 3875997.0, + "step": 152 + }, + { + "epoch": 0.01680210849989018, + "grad_norm": 4.814505577087402, + "learning_rate": 8.342480790340285e-08, + "loss": 1.3431, + "mean_token_accuracy": 0.637825608253479, + "num_tokens": 3900820.0, + "step": 153 + }, + { + "epoch": 0.016911926202503844, + "grad_norm": 4.500019550323486, + "learning_rate": 8.397365532381998e-08, + "loss": 1.3539, + "mean_token_accuracy": 0.6319369673728943, + "num_tokens": 3928146.0, + "step": 154 + }, + { + "epoch": 0.017021743905117504, + "grad_norm": 4.291878700256348, + "learning_rate": 8.45225027442371e-08, + "loss": 1.4003, + "mean_token_accuracy": 0.6195961833000183, + "num_tokens": 3962280.0, + "step": 155 + }, + { + "epoch": 0.017131561607731167, + "grad_norm": 4.461132049560547, + "learning_rate": 8.507135016465423e-08, + "loss": 1.2998, + "mean_token_accuracy": 0.64091956615448, + "num_tokens": 3987367.0, + "step": 156 + }, + { + "epoch": 0.017241379310344827, + "grad_norm": 4.63085412979126, + "learning_rate": 8.562019758507134e-08, + "loss": 1.3898, + "mean_token_accuracy": 0.625842273235321, + "num_tokens": 4015166.0, + "step": 157 + }, + { + "epoch": 0.017351197012958487, + "grad_norm": 4.65819787979126, + "learning_rate": 8.616904500548847e-08, + "loss": 1.3191, + "mean_token_accuracy": 0.636986494064331, + "num_tokens": 4038617.0, + "step": 158 + }, + { + "epoch": 0.01746101471557215, + "grad_norm": 4.755382537841797, + "learning_rate": 8.671789242590558e-08, + "loss": 1.3801, + "mean_token_accuracy": 0.6230050921440125, + "num_tokens": 4063597.0, + "step": 159 + }, + { + "epoch": 0.01757083241818581, + "grad_norm": 5.159773349761963, + "learning_rate": 8.726673984632271e-08, + "loss": 1.4044, + "mean_token_accuracy": 0.6204932928085327, + "num_tokens": 4086804.0, + "step": 160 + }, + { + "epoch": 0.017680650120799474, + "grad_norm": 4.203880786895752, + "learning_rate": 8.781558726673985e-08, + "loss": 1.2929, + "mean_token_accuracy": 0.6490458250045776, + "num_tokens": 4115439.0, + "step": 161 + }, + { + "epoch": 0.017790467823413134, + "grad_norm": 4.461368083953857, + "learning_rate": 8.836443468715697e-08, + "loss": 1.3838, + "mean_token_accuracy": 0.6246867179870605, + "num_tokens": 4141506.0, + "step": 162 + }, + { + "epoch": 0.017900285526026797, + "grad_norm": 4.646600723266602, + "learning_rate": 8.89132821075741e-08, + "loss": 1.3746, + "mean_token_accuracy": 0.6217072010040283, + "num_tokens": 4166039.0, + "step": 163 + }, + { + "epoch": 0.018010103228640457, + "grad_norm": 4.4872260093688965, + "learning_rate": 8.946212952799121e-08, + "loss": 1.3204, + "mean_token_accuracy": 0.6467245221138, + "num_tokens": 4192929.0, + "step": 164 + }, + { + "epoch": 0.018119920931254117, + "grad_norm": 4.108896255493164, + "learning_rate": 9.001097694840834e-08, + "loss": 1.3213, + "mean_token_accuracy": 0.6402141451835632, + "num_tokens": 4222519.0, + "step": 165 + }, + { + "epoch": 0.01822973863386778, + "grad_norm": 4.107428073883057, + "learning_rate": 9.055982436882546e-08, + "loss": 1.4121, + "mean_token_accuracy": 0.6166970133781433, + "num_tokens": 4252293.0, + "step": 166 + }, + { + "epoch": 0.01833955633648144, + "grad_norm": 4.591229438781738, + "learning_rate": 9.110867178924259e-08, + "loss": 1.4259, + "mean_token_accuracy": 0.6128690242767334, + "num_tokens": 4280243.0, + "step": 167 + }, + { + "epoch": 0.018449374039095103, + "grad_norm": 5.446043014526367, + "learning_rate": 9.165751920965971e-08, + "loss": 1.2802, + "mean_token_accuracy": 0.6400320529937744, + "num_tokens": 4300041.0, + "step": 168 + }, + { + "epoch": 0.018559191741708763, + "grad_norm": 4.732199192047119, + "learning_rate": 9.220636663007683e-08, + "loss": 1.362, + "mean_token_accuracy": 0.6233434677124023, + "num_tokens": 4321804.0, + "step": 169 + }, + { + "epoch": 0.018669009444322426, + "grad_norm": 6.036659240722656, + "learning_rate": 9.275521405049396e-08, + "loss": 1.2659, + "mean_token_accuracy": 0.6505096554756165, + "num_tokens": 4339295.0, + "step": 170 + }, + { + "epoch": 0.018778827146936086, + "grad_norm": 5.400204658508301, + "learning_rate": 9.330406147091107e-08, + "loss": 1.334, + "mean_token_accuracy": 0.6355339288711548, + "num_tokens": 4358001.0, + "step": 171 + }, + { + "epoch": 0.018888644849549746, + "grad_norm": 4.491601467132568, + "learning_rate": 9.38529088913282e-08, + "loss": 1.3719, + "mean_token_accuracy": 0.6185905337333679, + "num_tokens": 4387058.0, + "step": 172 + }, + { + "epoch": 0.01899846255216341, + "grad_norm": 4.401106834411621, + "learning_rate": 9.440175631174534e-08, + "loss": 1.4005, + "mean_token_accuracy": 0.6167795658111572, + "num_tokens": 4412168.0, + "step": 173 + }, + { + "epoch": 0.01910828025477707, + "grad_norm": 4.069396018981934, + "learning_rate": 9.495060373216246e-08, + "loss": 1.2206, + "mean_token_accuracy": 0.6573925018310547, + "num_tokens": 4438041.0, + "step": 174 + }, + { + "epoch": 0.019218097957390733, + "grad_norm": 4.674571990966797, + "learning_rate": 9.549945115257959e-08, + "loss": 1.3167, + "mean_token_accuracy": 0.6402021646499634, + "num_tokens": 4457967.0, + "step": 175 + }, + { + "epoch": 0.019327915660004392, + "grad_norm": 3.8826167583465576, + "learning_rate": 9.60482985729967e-08, + "loss": 1.3353, + "mean_token_accuracy": 0.6372344493865967, + "num_tokens": 4487178.0, + "step": 176 + }, + { + "epoch": 0.019437733362618052, + "grad_norm": 4.897689342498779, + "learning_rate": 9.659714599341383e-08, + "loss": 1.3307, + "mean_token_accuracy": 0.6360574960708618, + "num_tokens": 4510403.0, + "step": 177 + }, + { + "epoch": 0.019547551065231716, + "grad_norm": 3.931006908416748, + "learning_rate": 9.714599341383095e-08, + "loss": 1.359, + "mean_token_accuracy": 0.6269409656524658, + "num_tokens": 4539220.0, + "step": 178 + }, + { + "epoch": 0.019657368767845376, + "grad_norm": 3.8466410636901855, + "learning_rate": 9.769484083424807e-08, + "loss": 1.2958, + "mean_token_accuracy": 0.6455695033073425, + "num_tokens": 4567988.0, + "step": 179 + }, + { + "epoch": 0.01976718647045904, + "grad_norm": 4.430795192718506, + "learning_rate": 9.82436882546652e-08, + "loss": 1.3589, + "mean_token_accuracy": 0.6254836320877075, + "num_tokens": 4590434.0, + "step": 180 + }, + { + "epoch": 0.0198770041730727, + "grad_norm": 3.6762492656707764, + "learning_rate": 9.879253567508232e-08, + "loss": 1.3003, + "mean_token_accuracy": 0.646118700504303, + "num_tokens": 4622046.0, + "step": 181 + }, + { + "epoch": 0.019986821875686362, + "grad_norm": 4.137203216552734, + "learning_rate": 9.934138309549945e-08, + "loss": 1.2862, + "mean_token_accuracy": 0.6444594860076904, + "num_tokens": 4652986.0, + "step": 182 + }, + { + "epoch": 0.020096639578300022, + "grad_norm": 4.197319507598877, + "learning_rate": 9.989023051591656e-08, + "loss": 1.317, + "mean_token_accuracy": 0.633347749710083, + "num_tokens": 4682565.0, + "step": 183 + }, + { + "epoch": 0.020206457280913682, + "grad_norm": 3.8559656143188477, + "learning_rate": 1.0043907793633369e-07, + "loss": 1.3248, + "mean_token_accuracy": 0.6358917355537415, + "num_tokens": 4713362.0, + "step": 184 + }, + { + "epoch": 0.020316274983527345, + "grad_norm": 4.390886306762695, + "learning_rate": 1.0098792535675082e-07, + "loss": 1.2937, + "mean_token_accuracy": 0.6468855142593384, + "num_tokens": 4735726.0, + "step": 185 + }, + { + "epoch": 0.020426092686141005, + "grad_norm": 5.611185550689697, + "learning_rate": 1.0153677277716795e-07, + "loss": 1.2947, + "mean_token_accuracy": 0.6457133889198303, + "num_tokens": 4752876.0, + "step": 186 + }, + { + "epoch": 0.02053591038875467, + "grad_norm": 4.104154586791992, + "learning_rate": 1.0208562019758508e-07, + "loss": 1.2966, + "mean_token_accuracy": 0.642177164554596, + "num_tokens": 4777132.0, + "step": 187 + }, + { + "epoch": 0.020645728091368328, + "grad_norm": 4.536445140838623, + "learning_rate": 1.0263446761800219e-07, + "loss": 1.3137, + "mean_token_accuracy": 0.6387553215026855, + "num_tokens": 4801584.0, + "step": 188 + }, + { + "epoch": 0.02075554579398199, + "grad_norm": 3.942979335784912, + "learning_rate": 1.0318331503841932e-07, + "loss": 1.3069, + "mean_token_accuracy": 0.6413029432296753, + "num_tokens": 4827781.0, + "step": 189 + }, + { + "epoch": 0.02086536349659565, + "grad_norm": 4.781456470489502, + "learning_rate": 1.0373216245883643e-07, + "loss": 1.4687, + "mean_token_accuracy": 0.6150054931640625, + "num_tokens": 4849628.0, + "step": 190 + }, + { + "epoch": 0.02097518119920931, + "grad_norm": 3.638355016708374, + "learning_rate": 1.0428100987925356e-07, + "loss": 1.3231, + "mean_token_accuracy": 0.6321548819541931, + "num_tokens": 4879360.0, + "step": 191 + }, + { + "epoch": 0.021084998901822975, + "grad_norm": 4.133786678314209, + "learning_rate": 1.0482985729967068e-07, + "loss": 1.3311, + "mean_token_accuracy": 0.6319586038589478, + "num_tokens": 4907670.0, + "step": 192 + }, + { + "epoch": 0.021194816604436634, + "grad_norm": 3.6184022426605225, + "learning_rate": 1.0537870472008781e-07, + "loss": 1.3458, + "mean_token_accuracy": 0.6314202547073364, + "num_tokens": 4939982.0, + "step": 193 + }, + { + "epoch": 0.021304634307050298, + "grad_norm": 4.625723361968994, + "learning_rate": 1.0592755214050494e-07, + "loss": 1.4155, + "mean_token_accuracy": 0.6147107481956482, + "num_tokens": 4963392.0, + "step": 194 + }, + { + "epoch": 0.021414452009663958, + "grad_norm": 4.389723300933838, + "learning_rate": 1.0647639956092205e-07, + "loss": 1.282, + "mean_token_accuracy": 0.6450481414794922, + "num_tokens": 4985123.0, + "step": 195 + }, + { + "epoch": 0.021524269712277617, + "grad_norm": 5.008697509765625, + "learning_rate": 1.0702524698133918e-07, + "loss": 1.3093, + "mean_token_accuracy": 0.6449372172355652, + "num_tokens": 5008092.0, + "step": 196 + }, + { + "epoch": 0.02163408741489128, + "grad_norm": 4.624210834503174, + "learning_rate": 1.0757409440175631e-07, + "loss": 1.2866, + "mean_token_accuracy": 0.6433948278427124, + "num_tokens": 5035400.0, + "step": 197 + }, + { + "epoch": 0.02174390511750494, + "grad_norm": 5.546116352081299, + "learning_rate": 1.0812294182217344e-07, + "loss": 1.2917, + "mean_token_accuracy": 0.6432743072509766, + "num_tokens": 5059756.0, + "step": 198 + }, + { + "epoch": 0.021853722820118604, + "grad_norm": 6.257110118865967, + "learning_rate": 1.0867178924259056e-07, + "loss": 1.2235, + "mean_token_accuracy": 0.6518353223800659, + "num_tokens": 5076156.0, + "step": 199 + }, + { + "epoch": 0.021963540522732264, + "grad_norm": 4.82987117767334, + "learning_rate": 1.0922063666300768e-07, + "loss": 1.3394, + "mean_token_accuracy": 0.6277899742126465, + "num_tokens": 5103375.0, + "step": 200 + }, + { + "epoch": 0.022073358225345927, + "grad_norm": 5.087642669677734, + "learning_rate": 1.0976948408342481e-07, + "loss": 1.329, + "mean_token_accuracy": 0.6269881129264832, + "num_tokens": 5129861.0, + "step": 201 + }, + { + "epoch": 0.022183175927959587, + "grad_norm": 4.79668664932251, + "learning_rate": 1.1031833150384192e-07, + "loss": 1.3731, + "mean_token_accuracy": 0.6178420186042786, + "num_tokens": 5159264.0, + "step": 202 + }, + { + "epoch": 0.022292993630573247, + "grad_norm": 4.163984298706055, + "learning_rate": 1.1086717892425905e-07, + "loss": 1.1457, + "mean_token_accuracy": 0.6699997186660767, + "num_tokens": 5183540.0, + "step": 203 + }, + { + "epoch": 0.02240281133318691, + "grad_norm": 5.352416515350342, + "learning_rate": 1.1141602634467617e-07, + "loss": 1.2841, + "mean_token_accuracy": 0.6409816145896912, + "num_tokens": 5203934.0, + "step": 204 + }, + { + "epoch": 0.02251262903580057, + "grad_norm": 4.408868312835693, + "learning_rate": 1.119648737650933e-07, + "loss": 1.3417, + "mean_token_accuracy": 0.6330046653747559, + "num_tokens": 5231251.0, + "step": 205 + }, + { + "epoch": 0.022622446738414233, + "grad_norm": 4.1620259284973145, + "learning_rate": 1.1251372118551042e-07, + "loss": 1.3099, + "mean_token_accuracy": 0.6343173384666443, + "num_tokens": 5258183.0, + "step": 206 + }, + { + "epoch": 0.022732264441027893, + "grad_norm": 4.398978233337402, + "learning_rate": 1.1306256860592754e-07, + "loss": 1.3259, + "mean_token_accuracy": 0.6363443732261658, + "num_tokens": 5284139.0, + "step": 207 + }, + { + "epoch": 0.022842082143641557, + "grad_norm": 4.526630401611328, + "learning_rate": 1.1361141602634467e-07, + "loss": 1.321, + "mean_token_accuracy": 0.636906623840332, + "num_tokens": 5311524.0, + "step": 208 + }, + { + "epoch": 0.022951899846255217, + "grad_norm": 4.493083953857422, + "learning_rate": 1.141602634467618e-07, + "loss": 1.3144, + "mean_token_accuracy": 0.6374711990356445, + "num_tokens": 5339772.0, + "step": 209 + }, + { + "epoch": 0.023061717548868876, + "grad_norm": 4.203313827514648, + "learning_rate": 1.1470911086717892e-07, + "loss": 1.374, + "mean_token_accuracy": 0.6270907521247864, + "num_tokens": 5370725.0, + "step": 210 + }, + { + "epoch": 0.02317153525148254, + "grad_norm": 4.89697265625, + "learning_rate": 1.1525795828759604e-07, + "loss": 1.2377, + "mean_token_accuracy": 0.6533530354499817, + "num_tokens": 5396935.0, + "step": 211 + }, + { + "epoch": 0.0232813529540962, + "grad_norm": 4.140298366546631, + "learning_rate": 1.1580680570801317e-07, + "loss": 1.3589, + "mean_token_accuracy": 0.6225362420082092, + "num_tokens": 5428078.0, + "step": 212 + }, + { + "epoch": 0.023391170656709863, + "grad_norm": 4.716123580932617, + "learning_rate": 1.163556531284303e-07, + "loss": 1.3574, + "mean_token_accuracy": 0.638552725315094, + "num_tokens": 5456065.0, + "step": 213 + }, + { + "epoch": 0.023500988359323523, + "grad_norm": 8.733257293701172, + "learning_rate": 1.1690450054884741e-07, + "loss": 1.2436, + "mean_token_accuracy": 0.6544306874275208, + "num_tokens": 5478269.0, + "step": 214 + }, + { + "epoch": 0.023610806061937183, + "grad_norm": 5.112420082092285, + "learning_rate": 1.1745334796926454e-07, + "loss": 1.2843, + "mean_token_accuracy": 0.6445306539535522, + "num_tokens": 5508939.0, + "step": 215 + }, + { + "epoch": 0.023720623764550846, + "grad_norm": 5.3275532722473145, + "learning_rate": 1.1800219538968166e-07, + "loss": 1.2314, + "mean_token_accuracy": 0.6532975435256958, + "num_tokens": 5528782.0, + "step": 216 + }, + { + "epoch": 0.023830441467164506, + "grad_norm": 5.5103607177734375, + "learning_rate": 1.1855104281009878e-07, + "loss": 1.3502, + "mean_token_accuracy": 0.6323364973068237, + "num_tokens": 5560635.0, + "step": 217 + }, + { + "epoch": 0.02394025916977817, + "grad_norm": 5.151368141174316, + "learning_rate": 1.1909989023051591e-07, + "loss": 1.3028, + "mean_token_accuracy": 0.6438085436820984, + "num_tokens": 5589708.0, + "step": 218 + }, + { + "epoch": 0.02405007687239183, + "grad_norm": 5.0589094161987305, + "learning_rate": 1.1964873765093303e-07, + "loss": 1.2857, + "mean_token_accuracy": 0.6427679657936096, + "num_tokens": 5617684.0, + "step": 219 + }, + { + "epoch": 0.024159894575005492, + "grad_norm": 5.249506950378418, + "learning_rate": 1.2019758507135017e-07, + "loss": 1.3035, + "mean_token_accuracy": 0.6342928409576416, + "num_tokens": 5638957.0, + "step": 220 + }, + { + "epoch": 0.024269712277619152, + "grad_norm": 5.235424041748047, + "learning_rate": 1.2074643249176729e-07, + "loss": 1.2477, + "mean_token_accuracy": 0.6532148122787476, + "num_tokens": 5670336.0, + "step": 221 + }, + { + "epoch": 0.024379529980232812, + "grad_norm": 6.2552924156188965, + "learning_rate": 1.212952799121844e-07, + "loss": 1.2918, + "mean_token_accuracy": 0.6492533683776855, + "num_tokens": 5695133.0, + "step": 222 + }, + { + "epoch": 0.024489347682846475, + "grad_norm": 7.829499244689941, + "learning_rate": 1.2184412733260154e-07, + "loss": 1.2321, + "mean_token_accuracy": 0.6609385013580322, + "num_tokens": 5714384.0, + "step": 223 + }, + { + "epoch": 0.024599165385460135, + "grad_norm": 6.566165924072266, + "learning_rate": 1.2239297475301866e-07, + "loss": 1.1597, + "mean_token_accuracy": 0.6650270819664001, + "num_tokens": 5736244.0, + "step": 224 + }, + { + "epoch": 0.0247089830880738, + "grad_norm": 6.360326766967773, + "learning_rate": 1.2294182217343577e-07, + "loss": 1.2489, + "mean_token_accuracy": 0.6431970596313477, + "num_tokens": 5755357.0, + "step": 225 + }, + { + "epoch": 0.02481880079068746, + "grad_norm": 6.343930244445801, + "learning_rate": 1.2349066959385291e-07, + "loss": 1.2333, + "mean_token_accuracy": 0.6528059244155884, + "num_tokens": 5783182.0, + "step": 226 + }, + { + "epoch": 0.024928618493301122, + "grad_norm": 4.649899005889893, + "learning_rate": 1.2403951701427003e-07, + "loss": 1.3158, + "mean_token_accuracy": 0.6206269264221191, + "num_tokens": 5811712.0, + "step": 227 + }, + { + "epoch": 0.02503843619591478, + "grad_norm": 5.121761798858643, + "learning_rate": 1.2458836443468714e-07, + "loss": 1.2161, + "mean_token_accuracy": 0.6555817127227783, + "num_tokens": 5834971.0, + "step": 228 + }, + { + "epoch": 0.02514825389852844, + "grad_norm": 5.52947473526001, + "learning_rate": 1.2513721185510429e-07, + "loss": 1.1749, + "mean_token_accuracy": 0.6712073087692261, + "num_tokens": 5861472.0, + "step": 229 + }, + { + "epoch": 0.025258071601142105, + "grad_norm": 5.212065696716309, + "learning_rate": 1.256860592755214e-07, + "loss": 1.2179, + "mean_token_accuracy": 0.6619479656219482, + "num_tokens": 5890126.0, + "step": 230 + }, + { + "epoch": 0.025367889303755765, + "grad_norm": 4.599435806274414, + "learning_rate": 1.2623490669593854e-07, + "loss": 1.2583, + "mean_token_accuracy": 0.6507871150970459, + "num_tokens": 5919048.0, + "step": 231 + }, + { + "epoch": 0.025477707006369428, + "grad_norm": 5.945980548858643, + "learning_rate": 1.2678375411635563e-07, + "loss": 1.3379, + "mean_token_accuracy": 0.6418988704681396, + "num_tokens": 5940875.0, + "step": 232 + }, + { + "epoch": 0.025587524708983088, + "grad_norm": 7.738152980804443, + "learning_rate": 1.2733260153677277e-07, + "loss": 1.3119, + "mean_token_accuracy": 0.6293278932571411, + "num_tokens": 5962854.0, + "step": 233 + }, + { + "epoch": 0.025697342411596748, + "grad_norm": 6.897226810455322, + "learning_rate": 1.278814489571899e-07, + "loss": 1.2953, + "mean_token_accuracy": 0.6354627013206482, + "num_tokens": 5985851.0, + "step": 234 + }, + { + "epoch": 0.02580716011421041, + "grad_norm": 4.878690719604492, + "learning_rate": 1.2843029637760703e-07, + "loss": 1.1912, + "mean_token_accuracy": 0.6762239933013916, + "num_tokens": 6015303.0, + "step": 235 + }, + { + "epoch": 0.02591697781682407, + "grad_norm": 4.825289726257324, + "learning_rate": 1.2897914379802412e-07, + "loss": 1.2405, + "mean_token_accuracy": 0.649608314037323, + "num_tokens": 6041027.0, + "step": 236 + }, + { + "epoch": 0.026026795519437734, + "grad_norm": 5.234969615936279, + "learning_rate": 1.2952799121844126e-07, + "loss": 1.1483, + "mean_token_accuracy": 0.6766155362129211, + "num_tokens": 6066663.0, + "step": 237 + }, + { + "epoch": 0.026136613222051394, + "grad_norm": 5.509663105010986, + "learning_rate": 1.300768386388584e-07, + "loss": 1.2009, + "mean_token_accuracy": 0.664031982421875, + "num_tokens": 6092326.0, + "step": 238 + }, + { + "epoch": 0.026246430924665057, + "grad_norm": 6.480860233306885, + "learning_rate": 1.3062568605927552e-07, + "loss": 1.2758, + "mean_token_accuracy": 0.6450652480125427, + "num_tokens": 6119073.0, + "step": 239 + }, + { + "epoch": 0.026356248627278717, + "grad_norm": 5.661990642547607, + "learning_rate": 1.3117453347969266e-07, + "loss": 1.2407, + "mean_token_accuracy": 0.6493629813194275, + "num_tokens": 6141780.0, + "step": 240 + }, + { + "epoch": 0.026466066329892377, + "grad_norm": 6.969415664672852, + "learning_rate": 1.3172338090010975e-07, + "loss": 1.2232, + "mean_token_accuracy": 0.6588865518569946, + "num_tokens": 6162886.0, + "step": 241 + }, + { + "epoch": 0.02657588403250604, + "grad_norm": 5.418170928955078, + "learning_rate": 1.322722283205269e-07, + "loss": 1.3232, + "mean_token_accuracy": 0.6280111074447632, + "num_tokens": 6188639.0, + "step": 242 + }, + { + "epoch": 0.0266857017351197, + "grad_norm": 5.0611958503723145, + "learning_rate": 1.32821075740944e-07, + "loss": 1.2532, + "mean_token_accuracy": 0.6541692614555359, + "num_tokens": 6217501.0, + "step": 243 + }, + { + "epoch": 0.026795519437733364, + "grad_norm": 6.574410915374756, + "learning_rate": 1.3336992316136115e-07, + "loss": 1.3825, + "mean_token_accuracy": 0.6147276163101196, + "num_tokens": 6243835.0, + "step": 244 + }, + { + "epoch": 0.026905337140347024, + "grad_norm": 6.856051445007324, + "learning_rate": 1.3391877058177826e-07, + "loss": 1.2172, + "mean_token_accuracy": 0.6767973899841309, + "num_tokens": 6267918.0, + "step": 245 + }, + { + "epoch": 0.027015154842960687, + "grad_norm": 6.0586161613464355, + "learning_rate": 1.3446761800219538e-07, + "loss": 1.3521, + "mean_token_accuracy": 0.6288238763809204, + "num_tokens": 6292294.0, + "step": 246 + }, + { + "epoch": 0.027124972545574347, + "grad_norm": 5.2837653160095215, + "learning_rate": 1.350164654226125e-07, + "loss": 1.2297, + "mean_token_accuracy": 0.6499220132827759, + "num_tokens": 6317144.0, + "step": 247 + }, + { + "epoch": 0.027234790248188007, + "grad_norm": 7.483640670776367, + "learning_rate": 1.3556531284302963e-07, + "loss": 1.1965, + "mean_token_accuracy": 0.6604255437850952, + "num_tokens": 6337930.0, + "step": 248 + }, + { + "epoch": 0.02734460795080167, + "grad_norm": 4.95797061920166, + "learning_rate": 1.3611416026344675e-07, + "loss": 1.2053, + "mean_token_accuracy": 0.6568809747695923, + "num_tokens": 6361097.0, + "step": 249 + }, + { + "epoch": 0.02745442565341533, + "grad_norm": 5.306252956390381, + "learning_rate": 1.366630076838639e-07, + "loss": 1.3007, + "mean_token_accuracy": 0.6414871215820312, + "num_tokens": 6389324.0, + "step": 250 + }, + { + "epoch": 0.027564243356028993, + "grad_norm": 5.198586940765381, + "learning_rate": 1.37211855104281e-07, + "loss": 1.2006, + "mean_token_accuracy": 0.6569438576698303, + "num_tokens": 6413029.0, + "step": 251 + }, + { + "epoch": 0.027674061058642653, + "grad_norm": 4.8919901847839355, + "learning_rate": 1.3776070252469812e-07, + "loss": 1.2217, + "mean_token_accuracy": 0.649081826210022, + "num_tokens": 6436213.0, + "step": 252 + }, + { + "epoch": 0.027783878761256313, + "grad_norm": 5.383180618286133, + "learning_rate": 1.3830954994511526e-07, + "loss": 1.2734, + "mean_token_accuracy": 0.6393853425979614, + "num_tokens": 6457900.0, + "step": 253 + }, + { + "epoch": 0.027893696463869976, + "grad_norm": 6.62015438079834, + "learning_rate": 1.3885839736553238e-07, + "loss": 1.1186, + "mean_token_accuracy": 0.6793749332427979, + "num_tokens": 6479036.0, + "step": 254 + }, + { + "epoch": 0.028003514166483636, + "grad_norm": 4.836477756500244, + "learning_rate": 1.394072447859495e-07, + "loss": 1.2209, + "mean_token_accuracy": 0.6519668102264404, + "num_tokens": 6506436.0, + "step": 255 + }, + { + "epoch": 0.0281133318690973, + "grad_norm": 6.114823341369629, + "learning_rate": 1.399560922063666e-07, + "loss": 1.1941, + "mean_token_accuracy": 0.6636967658996582, + "num_tokens": 6529091.0, + "step": 256 + }, + { + "epoch": 0.02822314957171096, + "grad_norm": 5.549185752868652, + "learning_rate": 1.4050493962678375e-07, + "loss": 1.2997, + "mean_token_accuracy": 0.6370053887367249, + "num_tokens": 6555232.0, + "step": 257 + }, + { + "epoch": 0.028332967274324623, + "grad_norm": 4.7065911293029785, + "learning_rate": 1.4105378704720087e-07, + "loss": 1.2259, + "mean_token_accuracy": 0.660826563835144, + "num_tokens": 6580294.0, + "step": 258 + }, + { + "epoch": 0.028442784976938282, + "grad_norm": 5.977413654327393, + "learning_rate": 1.41602634467618e-07, + "loss": 1.2577, + "mean_token_accuracy": 0.6637356281280518, + "num_tokens": 6602660.0, + "step": 259 + }, + { + "epoch": 0.028552602679551942, + "grad_norm": 5.96026611328125, + "learning_rate": 1.421514818880351e-07, + "loss": 1.2704, + "mean_token_accuracy": 0.6457116603851318, + "num_tokens": 6624266.0, + "step": 260 + }, + { + "epoch": 0.028662420382165606, + "grad_norm": 4.318169116973877, + "learning_rate": 1.4270032930845224e-07, + "loss": 1.2093, + "mean_token_accuracy": 0.6591264605522156, + "num_tokens": 6654614.0, + "step": 261 + }, + { + "epoch": 0.028772238084779266, + "grad_norm": 5.302940845489502, + "learning_rate": 1.4324917672886938e-07, + "loss": 1.279, + "mean_token_accuracy": 0.6389685869216919, + "num_tokens": 6679742.0, + "step": 262 + }, + { + "epoch": 0.02888205578739293, + "grad_norm": 5.7312331199646, + "learning_rate": 1.437980241492865e-07, + "loss": 1.1876, + "mean_token_accuracy": 0.6635587215423584, + "num_tokens": 6704847.0, + "step": 263 + }, + { + "epoch": 0.02899187349000659, + "grad_norm": 5.446577072143555, + "learning_rate": 1.4434687156970364e-07, + "loss": 1.2694, + "mean_token_accuracy": 0.6425389051437378, + "num_tokens": 6730067.0, + "step": 264 + }, + { + "epoch": 0.029101691192620252, + "grad_norm": 6.36404275894165, + "learning_rate": 1.4489571899012073e-07, + "loss": 1.3259, + "mean_token_accuracy": 0.6313263177871704, + "num_tokens": 6750969.0, + "step": 265 + }, + { + "epoch": 0.029211508895233912, + "grad_norm": 5.332357883453369, + "learning_rate": 1.4544456641053787e-07, + "loss": 1.1841, + "mean_token_accuracy": 0.6688269376754761, + "num_tokens": 6777261.0, + "step": 266 + }, + { + "epoch": 0.029321326597847572, + "grad_norm": 5.694687366485596, + "learning_rate": 1.4599341383095498e-07, + "loss": 1.3256, + "mean_token_accuracy": 0.6394702196121216, + "num_tokens": 6802671.0, + "step": 267 + }, + { + "epoch": 0.029431144300461235, + "grad_norm": 4.881686687469482, + "learning_rate": 1.4654226125137212e-07, + "loss": 1.2461, + "mean_token_accuracy": 0.6538093090057373, + "num_tokens": 6829791.0, + "step": 268 + }, + { + "epoch": 0.029540962003074895, + "grad_norm": 6.541938304901123, + "learning_rate": 1.4709110867178924e-07, + "loss": 1.2287, + "mean_token_accuracy": 0.6542690396308899, + "num_tokens": 6856919.0, + "step": 269 + }, + { + "epoch": 0.02965077970568856, + "grad_norm": 4.955570220947266, + "learning_rate": 1.4763995609220636e-07, + "loss": 1.2662, + "mean_token_accuracy": 0.6475072503089905, + "num_tokens": 6887137.0, + "step": 270 + }, + { + "epoch": 0.029760597408302218, + "grad_norm": 5.674773216247559, + "learning_rate": 1.4818880351262347e-07, + "loss": 1.2893, + "mean_token_accuracy": 0.6419798731803894, + "num_tokens": 6914016.0, + "step": 271 + }, + { + "epoch": 0.029870415110915878, + "grad_norm": 5.9081010818481445, + "learning_rate": 1.487376509330406e-07, + "loss": 1.1734, + "mean_token_accuracy": 0.6727598905563354, + "num_tokens": 6938238.0, + "step": 272 + }, + { + "epoch": 0.02998023281352954, + "grad_norm": 6.148460865020752, + "learning_rate": 1.4928649835345773e-07, + "loss": 1.3134, + "mean_token_accuracy": 0.6442654132843018, + "num_tokens": 6964595.0, + "step": 273 + }, + { + "epoch": 0.0300900505161432, + "grad_norm": 5.503339767456055, + "learning_rate": 1.4983534577387484e-07, + "loss": 1.1882, + "mean_token_accuracy": 0.667168140411377, + "num_tokens": 6991332.0, + "step": 274 + }, + { + "epoch": 0.030199868218756865, + "grad_norm": 5.058751106262207, + "learning_rate": 1.5038419319429198e-07, + "loss": 1.2153, + "mean_token_accuracy": 0.6464576125144958, + "num_tokens": 7017043.0, + "step": 275 + }, + { + "epoch": 0.030309685921370524, + "grad_norm": 5.64825439453125, + "learning_rate": 1.509330406147091e-07, + "loss": 1.1083, + "mean_token_accuracy": 0.6790839433670044, + "num_tokens": 7041085.0, + "step": 276 + }, + { + "epoch": 0.030419503623984188, + "grad_norm": 6.9663519859313965, + "learning_rate": 1.5148188803512624e-07, + "loss": 1.2649, + "mean_token_accuracy": 0.6454502940177917, + "num_tokens": 7066096.0, + "step": 277 + }, + { + "epoch": 0.030529321326597848, + "grad_norm": 5.028481960296631, + "learning_rate": 1.5203073545554336e-07, + "loss": 1.2498, + "mean_token_accuracy": 0.6509221792221069, + "num_tokens": 7092328.0, + "step": 278 + }, + { + "epoch": 0.030639139029211507, + "grad_norm": 6.247194290161133, + "learning_rate": 1.5257958287596047e-07, + "loss": 1.1962, + "mean_token_accuracy": 0.6618742942810059, + "num_tokens": 7115980.0, + "step": 279 + }, + { + "epoch": 0.03074895673182517, + "grad_norm": 5.4936203956604, + "learning_rate": 1.531284302963776e-07, + "loss": 1.1873, + "mean_token_accuracy": 0.6670287251472473, + "num_tokens": 7141184.0, + "step": 280 + }, + { + "epoch": 0.03085877443443883, + "grad_norm": 4.802577972412109, + "learning_rate": 1.5367727771679473e-07, + "loss": 1.3181, + "mean_token_accuracy": 0.6302636861801147, + "num_tokens": 7172099.0, + "step": 281 + }, + { + "epoch": 0.030968592137052494, + "grad_norm": 4.828968524932861, + "learning_rate": 1.5422612513721184e-07, + "loss": 1.2635, + "mean_token_accuracy": 0.6478919982910156, + "num_tokens": 7201732.0, + "step": 282 + }, + { + "epoch": 0.031078409839666154, + "grad_norm": 5.546483039855957, + "learning_rate": 1.5477497255762899e-07, + "loss": 1.2292, + "mean_token_accuracy": 0.6530047655105591, + "num_tokens": 7229730.0, + "step": 283 + }, + { + "epoch": 0.031188227542279817, + "grad_norm": 4.373313903808594, + "learning_rate": 1.5532381997804607e-07, + "loss": 1.2589, + "mean_token_accuracy": 0.6474318504333496, + "num_tokens": 7256113.0, + "step": 284 + }, + { + "epoch": 0.03129804524489348, + "grad_norm": 5.393400192260742, + "learning_rate": 1.5587266739846322e-07, + "loss": 1.3084, + "mean_token_accuracy": 0.6469357013702393, + "num_tokens": 7279595.0, + "step": 285 + }, + { + "epoch": 0.03140786294750714, + "grad_norm": 6.226471900939941, + "learning_rate": 1.5642151481888036e-07, + "loss": 1.1342, + "mean_token_accuracy": 0.6735808253288269, + "num_tokens": 7299428.0, + "step": 286 + }, + { + "epoch": 0.0315176806501208, + "grad_norm": 5.789180278778076, + "learning_rate": 1.5697036223929747e-07, + "loss": 1.1883, + "mean_token_accuracy": 0.6645945310592651, + "num_tokens": 7319886.0, + "step": 287 + }, + { + "epoch": 0.031627498352734464, + "grad_norm": 4.744363307952881, + "learning_rate": 1.5751920965971461e-07, + "loss": 1.2447, + "mean_token_accuracy": 0.6527084112167358, + "num_tokens": 7352941.0, + "step": 288 + }, + { + "epoch": 0.03173731605534812, + "grad_norm": 6.063812255859375, + "learning_rate": 1.580680570801317e-07, + "loss": 1.1916, + "mean_token_accuracy": 0.6485726237297058, + "num_tokens": 7376356.0, + "step": 289 + }, + { + "epoch": 0.03184713375796178, + "grad_norm": 4.603757381439209, + "learning_rate": 1.5861690450054885e-07, + "loss": 1.1855, + "mean_token_accuracy": 0.6626735329627991, + "num_tokens": 7405594.0, + "step": 290 + }, + { + "epoch": 0.03195695146057544, + "grad_norm": 6.799347877502441, + "learning_rate": 1.5916575192096596e-07, + "loss": 1.2596, + "mean_token_accuracy": 0.6504641771316528, + "num_tokens": 7430154.0, + "step": 291 + }, + { + "epoch": 0.0320667691631891, + "grad_norm": 4.464807987213135, + "learning_rate": 1.597145993413831e-07, + "loss": 1.3338, + "mean_token_accuracy": 0.6271020174026489, + "num_tokens": 7462656.0, + "step": 292 + }, + { + "epoch": 0.03217658686580277, + "grad_norm": 4.476639747619629, + "learning_rate": 1.602634467618002e-07, + "loss": 1.2785, + "mean_token_accuracy": 0.6414609551429749, + "num_tokens": 7492089.0, + "step": 293 + }, + { + "epoch": 0.03228640456841643, + "grad_norm": 4.816463470458984, + "learning_rate": 1.6081229418221733e-07, + "loss": 1.2694, + "mean_token_accuracy": 0.6382818818092346, + "num_tokens": 7520667.0, + "step": 294 + }, + { + "epoch": 0.03239622227103009, + "grad_norm": 3.6076762676239014, + "learning_rate": 1.6136114160263445e-07, + "loss": 1.1682, + "mean_token_accuracy": 0.6647048592567444, + "num_tokens": 7547948.0, + "step": 295 + }, + { + "epoch": 0.03250603997364375, + "grad_norm": 4.821333885192871, + "learning_rate": 1.619099890230516e-07, + "loss": 1.1655, + "mean_token_accuracy": 0.6611570715904236, + "num_tokens": 7576491.0, + "step": 296 + }, + { + "epoch": 0.03261585767625741, + "grad_norm": 5.022906303405762, + "learning_rate": 1.6245883644346873e-07, + "loss": 1.2501, + "mean_token_accuracy": 0.6421570181846619, + "num_tokens": 7602228.0, + "step": 297 + }, + { + "epoch": 0.032725675378871076, + "grad_norm": 5.56349515914917, + "learning_rate": 1.6300768386388582e-07, + "loss": 1.1244, + "mean_token_accuracy": 0.6791588068008423, + "num_tokens": 7625036.0, + "step": 298 + }, + { + "epoch": 0.032835493081484736, + "grad_norm": 4.996785640716553, + "learning_rate": 1.6355653128430296e-07, + "loss": 1.2846, + "mean_token_accuracy": 0.6450839042663574, + "num_tokens": 7647349.0, + "step": 299 + }, + { + "epoch": 0.032945310784098396, + "grad_norm": 4.037537097930908, + "learning_rate": 1.6410537870472008e-07, + "loss": 1.1806, + "mean_token_accuracy": 0.6630178093910217, + "num_tokens": 7677517.0, + "step": 300 + }, + { + "epoch": 0.033055128486712056, + "grad_norm": 5.468443393707275, + "learning_rate": 1.6465422612513722e-07, + "loss": 1.1533, + "mean_token_accuracy": 0.6670456528663635, + "num_tokens": 7699596.0, + "step": 301 + }, + { + "epoch": 0.03316494618932572, + "grad_norm": 6.511560440063477, + "learning_rate": 1.6520307354555433e-07, + "loss": 1.1783, + "mean_token_accuracy": 0.6587928533554077, + "num_tokens": 7719512.0, + "step": 302 + }, + { + "epoch": 0.03327476389193938, + "grad_norm": 5.256762504577637, + "learning_rate": 1.6575192096597145e-07, + "loss": 1.2459, + "mean_token_accuracy": 0.6568674445152283, + "num_tokens": 7742568.0, + "step": 303 + }, + { + "epoch": 0.03338458159455304, + "grad_norm": 4.452115535736084, + "learning_rate": 1.6630076838638856e-07, + "loss": 1.2309, + "mean_token_accuracy": 0.6530786752700806, + "num_tokens": 7775570.0, + "step": 304 + }, + { + "epoch": 0.0334943992971667, + "grad_norm": 4.179479122161865, + "learning_rate": 1.668496158068057e-07, + "loss": 1.2118, + "mean_token_accuracy": 0.6527677774429321, + "num_tokens": 7801825.0, + "step": 305 + }, + { + "epoch": 0.03360421699978036, + "grad_norm": 4.7929534912109375, + "learning_rate": 1.6739846322722282e-07, + "loss": 1.1524, + "mean_token_accuracy": 0.6721697449684143, + "num_tokens": 7831562.0, + "step": 306 + }, + { + "epoch": 0.03371403470239403, + "grad_norm": 5.521035671234131, + "learning_rate": 1.6794731064763996e-07, + "loss": 1.2554, + "mean_token_accuracy": 0.6495943069458008, + "num_tokens": 7858636.0, + "step": 307 + }, + { + "epoch": 0.03382385240500769, + "grad_norm": 5.3450493812561035, + "learning_rate": 1.6849615806805705e-07, + "loss": 1.241, + "mean_token_accuracy": 0.6477991342544556, + "num_tokens": 7882277.0, + "step": 308 + }, + { + "epoch": 0.03393367010762135, + "grad_norm": 5.1329803466796875, + "learning_rate": 1.690450054884742e-07, + "loss": 1.192, + "mean_token_accuracy": 0.6627562046051025, + "num_tokens": 7905998.0, + "step": 309 + }, + { + "epoch": 0.03404348781023501, + "grad_norm": 5.367285251617432, + "learning_rate": 1.6959385290889134e-07, + "loss": 1.2407, + "mean_token_accuracy": 0.6529823541641235, + "num_tokens": 7937012.0, + "step": 310 + }, + { + "epoch": 0.03415330551284867, + "grad_norm": 5.733229637145996, + "learning_rate": 1.7014270032930845e-07, + "loss": 1.2402, + "mean_token_accuracy": 0.6516827344894409, + "num_tokens": 7969805.0, + "step": 311 + }, + { + "epoch": 0.034263123215462335, + "grad_norm": 4.741449356079102, + "learning_rate": 1.7069154774972557e-07, + "loss": 1.2563, + "mean_token_accuracy": 0.6435132622718811, + "num_tokens": 7997448.0, + "step": 312 + }, + { + "epoch": 0.034372940918075995, + "grad_norm": 5.369378089904785, + "learning_rate": 1.7124039517014268e-07, + "loss": 1.1471, + "mean_token_accuracy": 0.6797165870666504, + "num_tokens": 8021556.0, + "step": 313 + }, + { + "epoch": 0.034482758620689655, + "grad_norm": 4.462498664855957, + "learning_rate": 1.7178924259055982e-07, + "loss": 1.2435, + "mean_token_accuracy": 0.6500431299209595, + "num_tokens": 8050045.0, + "step": 314 + }, + { + "epoch": 0.034592576323303315, + "grad_norm": 5.212384223937988, + "learning_rate": 1.7233809001097694e-07, + "loss": 1.1394, + "mean_token_accuracy": 0.6774324178695679, + "num_tokens": 8074140.0, + "step": 315 + }, + { + "epoch": 0.034702394025916974, + "grad_norm": 6.954823970794678, + "learning_rate": 1.7288693743139408e-07, + "loss": 1.1204, + "mean_token_accuracy": 0.675693929195404, + "num_tokens": 8099589.0, + "step": 316 + }, + { + "epoch": 0.03481221172853064, + "grad_norm": 4.0648512840271, + "learning_rate": 1.7343578485181117e-07, + "loss": 1.1582, + "mean_token_accuracy": 0.6658995747566223, + "num_tokens": 8121906.0, + "step": 317 + }, + { + "epoch": 0.0349220294311443, + "grad_norm": 4.062045574188232, + "learning_rate": 1.739846322722283e-07, + "loss": 1.2026, + "mean_token_accuracy": 0.6628013253211975, + "num_tokens": 8151382.0, + "step": 318 + }, + { + "epoch": 0.03503184713375796, + "grad_norm": 4.515096664428711, + "learning_rate": 1.7453347969264543e-07, + "loss": 1.1132, + "mean_token_accuracy": 0.6750463843345642, + "num_tokens": 8176874.0, + "step": 319 + }, + { + "epoch": 0.03514166483637162, + "grad_norm": 5.529139518737793, + "learning_rate": 1.7508232711306257e-07, + "loss": 1.1744, + "mean_token_accuracy": 0.6693198084831238, + "num_tokens": 8197813.0, + "step": 320 + }, + { + "epoch": 0.03525148253898529, + "grad_norm": 6.273200511932373, + "learning_rate": 1.756311745334797e-07, + "loss": 1.2093, + "mean_token_accuracy": 0.6637544631958008, + "num_tokens": 8216482.0, + "step": 321 + }, + { + "epoch": 0.03536130024159895, + "grad_norm": 3.5019726753234863, + "learning_rate": 1.761800219538968e-07, + "loss": 1.2046, + "mean_token_accuracy": 0.6645109057426453, + "num_tokens": 8244375.0, + "step": 322 + }, + { + "epoch": 0.03547111794421261, + "grad_norm": 5.300239562988281, + "learning_rate": 1.7672886937431394e-07, + "loss": 1.1103, + "mean_token_accuracy": 0.6717514991760254, + "num_tokens": 8262306.0, + "step": 323 + }, + { + "epoch": 0.03558093564682627, + "grad_norm": 6.247761249542236, + "learning_rate": 1.7727771679473105e-07, + "loss": 1.2791, + "mean_token_accuracy": 0.6341515779495239, + "num_tokens": 8289269.0, + "step": 324 + }, + { + "epoch": 0.03569075334943993, + "grad_norm": 6.319210529327393, + "learning_rate": 1.778265642151482e-07, + "loss": 1.29, + "mean_token_accuracy": 0.6426863670349121, + "num_tokens": 8313882.0, + "step": 325 + }, + { + "epoch": 0.035800571052053594, + "grad_norm": 5.348138809204102, + "learning_rate": 1.7837541163556529e-07, + "loss": 1.1176, + "mean_token_accuracy": 0.675033450126648, + "num_tokens": 8335826.0, + "step": 326 + }, + { + "epoch": 0.035910388754667254, + "grad_norm": 3.953932523727417, + "learning_rate": 1.7892425905598243e-07, + "loss": 1.2656, + "mean_token_accuracy": 0.6453923583030701, + "num_tokens": 8363222.0, + "step": 327 + }, + { + "epoch": 0.036020206457280914, + "grad_norm": 5.097783088684082, + "learning_rate": 1.7947310647639954e-07, + "loss": 1.2119, + "mean_token_accuracy": 0.6665951609611511, + "num_tokens": 8392585.0, + "step": 328 + }, + { + "epoch": 0.03613002415989457, + "grad_norm": 4.050562858581543, + "learning_rate": 1.8002195389681668e-07, + "loss": 1.2689, + "mean_token_accuracy": 0.6394222974777222, + "num_tokens": 8418848.0, + "step": 329 + }, + { + "epoch": 0.03623984186250823, + "grad_norm": 4.034219264984131, + "learning_rate": 1.805708013172338e-07, + "loss": 1.3198, + "mean_token_accuracy": 0.6384641528129578, + "num_tokens": 8445670.0, + "step": 330 + }, + { + "epoch": 0.0363496595651219, + "grad_norm": 4.84985876083374, + "learning_rate": 1.8111964873765091e-07, + "loss": 1.3068, + "mean_token_accuracy": 0.6318179368972778, + "num_tokens": 8470479.0, + "step": 331 + }, + { + "epoch": 0.03645947726773556, + "grad_norm": 5.057329177856445, + "learning_rate": 1.8166849615806803e-07, + "loss": 1.227, + "mean_token_accuracy": 0.6472724676132202, + "num_tokens": 8495725.0, + "step": 332 + }, + { + "epoch": 0.03656929497034922, + "grad_norm": 4.533729076385498, + "learning_rate": 1.8221734357848517e-07, + "loss": 1.1453, + "mean_token_accuracy": 0.6733964681625366, + "num_tokens": 8519161.0, + "step": 333 + }, + { + "epoch": 0.03667911267296288, + "grad_norm": 4.103779315948486, + "learning_rate": 1.827661909989023e-07, + "loss": 1.2151, + "mean_token_accuracy": 0.658466637134552, + "num_tokens": 8550775.0, + "step": 334 + }, + { + "epoch": 0.03678893037557654, + "grad_norm": 6.017364025115967, + "learning_rate": 1.8331503841931943e-07, + "loss": 1.1279, + "mean_token_accuracy": 0.6784398555755615, + "num_tokens": 8571373.0, + "step": 335 + }, + { + "epoch": 0.036898748078190206, + "grad_norm": 4.331385135650635, + "learning_rate": 1.8386388583973654e-07, + "loss": 1.1615, + "mean_token_accuracy": 0.6637086868286133, + "num_tokens": 8591503.0, + "step": 336 + }, + { + "epoch": 0.037008565780803866, + "grad_norm": 4.2774224281311035, + "learning_rate": 1.8441273326015366e-07, + "loss": 1.1581, + "mean_token_accuracy": 0.6711850166320801, + "num_tokens": 8615031.0, + "step": 337 + }, + { + "epoch": 0.037118383483417526, + "grad_norm": 3.9136672019958496, + "learning_rate": 1.849615806805708e-07, + "loss": 1.1699, + "mean_token_accuracy": 0.6591304540634155, + "num_tokens": 8640153.0, + "step": 338 + }, + { + "epoch": 0.037228201186031186, + "grad_norm": 4.694100856781006, + "learning_rate": 1.8551042810098792e-07, + "loss": 1.1489, + "mean_token_accuracy": 0.6706222891807556, + "num_tokens": 8664669.0, + "step": 339 + }, + { + "epoch": 0.03733801888864485, + "grad_norm": 4.887158393859863, + "learning_rate": 1.8605927552140506e-07, + "loss": 1.136, + "mean_token_accuracy": 0.6789441108703613, + "num_tokens": 8684457.0, + "step": 340 + }, + { + "epoch": 0.03744783659125851, + "grad_norm": 4.15380334854126, + "learning_rate": 1.8660812294182215e-07, + "loss": 1.24, + "mean_token_accuracy": 0.6549307703971863, + "num_tokens": 8709664.0, + "step": 341 + }, + { + "epoch": 0.03755765429387217, + "grad_norm": 3.7226176261901855, + "learning_rate": 1.871569703622393e-07, + "loss": 1.2619, + "mean_token_accuracy": 0.6395746469497681, + "num_tokens": 8739516.0, + "step": 342 + }, + { + "epoch": 0.03766747199648583, + "grad_norm": 4.13930606842041, + "learning_rate": 1.877058177826564e-07, + "loss": 1.183, + "mean_token_accuracy": 0.664172887802124, + "num_tokens": 8762872.0, + "step": 343 + }, + { + "epoch": 0.03777728969909949, + "grad_norm": 4.891940116882324, + "learning_rate": 1.8825466520307354e-07, + "loss": 1.2431, + "mean_token_accuracy": 0.6474275588989258, + "num_tokens": 8787182.0, + "step": 344 + }, + { + "epoch": 0.03788710740171316, + "grad_norm": 6.439157009124756, + "learning_rate": 1.8880351262349069e-07, + "loss": 1.1153, + "mean_token_accuracy": 0.6754665374755859, + "num_tokens": 8806108.0, + "step": 345 + }, + { + "epoch": 0.03799692510432682, + "grad_norm": 4.826892375946045, + "learning_rate": 1.8935236004390778e-07, + "loss": 1.2645, + "mean_token_accuracy": 0.642389714717865, + "num_tokens": 8835240.0, + "step": 346 + }, + { + "epoch": 0.03810674280694048, + "grad_norm": 5.584826469421387, + "learning_rate": 1.8990120746432492e-07, + "loss": 1.2703, + "mean_token_accuracy": 0.6368387937545776, + "num_tokens": 8864544.0, + "step": 347 + }, + { + "epoch": 0.03821656050955414, + "grad_norm": 4.898873805999756, + "learning_rate": 1.9045005488474203e-07, + "loss": 1.2321, + "mean_token_accuracy": 0.6544634103775024, + "num_tokens": 8896595.0, + "step": 348 + }, + { + "epoch": 0.0383263782121678, + "grad_norm": 4.937606334686279, + "learning_rate": 1.9099890230515917e-07, + "loss": 1.172, + "mean_token_accuracy": 0.6600698828697205, + "num_tokens": 8919214.0, + "step": 349 + }, + { + "epoch": 0.038436195914781465, + "grad_norm": 5.440916061401367, + "learning_rate": 1.9154774972557626e-07, + "loss": 1.0933, + "mean_token_accuracy": 0.675153374671936, + "num_tokens": 8938103.0, + "step": 350 + }, + { + "epoch": 0.038546013617395125, + "grad_norm": 4.735672473907471, + "learning_rate": 1.920965971459934e-07, + "loss": 1.2143, + "mean_token_accuracy": 0.6697666049003601, + "num_tokens": 8962791.0, + "step": 351 + }, + { + "epoch": 0.038655831320008785, + "grad_norm": 4.5057783126831055, + "learning_rate": 1.9264544456641052e-07, + "loss": 1.1153, + "mean_token_accuracy": 0.6766290664672852, + "num_tokens": 8986037.0, + "step": 352 + }, + { + "epoch": 0.038765649022622445, + "grad_norm": 4.1262993812561035, + "learning_rate": 1.9319429198682766e-07, + "loss": 1.297, + "mean_token_accuracy": 0.6372253894805908, + "num_tokens": 9015429.0, + "step": 353 + }, + { + "epoch": 0.038875466725236105, + "grad_norm": 5.189697265625, + "learning_rate": 1.9374313940724478e-07, + "loss": 1.1737, + "mean_token_accuracy": 0.6634988784790039, + "num_tokens": 9037793.0, + "step": 354 + }, + { + "epoch": 0.03898528442784977, + "grad_norm": 4.189990997314453, + "learning_rate": 1.942919868276619e-07, + "loss": 1.2387, + "mean_token_accuracy": 0.6483986377716064, + "num_tokens": 9059741.0, + "step": 355 + }, + { + "epoch": 0.03909510213046343, + "grad_norm": 5.238173961639404, + "learning_rate": 1.9484083424807903e-07, + "loss": 1.1758, + "mean_token_accuracy": 0.6651731133460999, + "num_tokens": 9086452.0, + "step": 356 + }, + { + "epoch": 0.03920491983307709, + "grad_norm": 4.715301513671875, + "learning_rate": 1.9538968166849615e-07, + "loss": 1.2631, + "mean_token_accuracy": 0.6514566540718079, + "num_tokens": 9112482.0, + "step": 357 + }, + { + "epoch": 0.03931473753569075, + "grad_norm": 4.50177526473999, + "learning_rate": 1.959385290889133e-07, + "loss": 1.1651, + "mean_token_accuracy": 0.668006420135498, + "num_tokens": 9136630.0, + "step": 358 + }, + { + "epoch": 0.03942455523830442, + "grad_norm": 4.888970851898193, + "learning_rate": 1.964873765093304e-07, + "loss": 1.1958, + "mean_token_accuracy": 0.654298722743988, + "num_tokens": 9163692.0, + "step": 359 + }, + { + "epoch": 0.03953437294091808, + "grad_norm": 4.675551414489746, + "learning_rate": 1.9703622392974752e-07, + "loss": 1.1817, + "mean_token_accuracy": 0.6732252836227417, + "num_tokens": 9186102.0, + "step": 360 + }, + { + "epoch": 0.03964419064353174, + "grad_norm": 3.7936294078826904, + "learning_rate": 1.9758507135016464e-07, + "loss": 1.1887, + "mean_token_accuracy": 0.654995322227478, + "num_tokens": 9210929.0, + "step": 361 + }, + { + "epoch": 0.0397540083461454, + "grad_norm": 3.95939302444458, + "learning_rate": 1.9813391877058178e-07, + "loss": 1.2024, + "mean_token_accuracy": 0.6537796258926392, + "num_tokens": 9239172.0, + "step": 362 + }, + { + "epoch": 0.03986382604875906, + "grad_norm": 4.048371315002441, + "learning_rate": 1.986827661909989e-07, + "loss": 1.2109, + "mean_token_accuracy": 0.6546812057495117, + "num_tokens": 9269257.0, + "step": 363 + }, + { + "epoch": 0.039973643751372724, + "grad_norm": 4.7207560539245605, + "learning_rate": 1.9923161361141603e-07, + "loss": 1.222, + "mean_token_accuracy": 0.6650037169456482, + "num_tokens": 9294419.0, + "step": 364 + }, + { + "epoch": 0.040083461453986384, + "grad_norm": 3.259023427963257, + "learning_rate": 1.9978046103183312e-07, + "loss": 1.2613, + "mean_token_accuracy": 0.643643856048584, + "num_tokens": 9324179.0, + "step": 365 + }, + { + "epoch": 0.040193279156600044, + "grad_norm": 4.75868034362793, + "learning_rate": 2.0032930845225027e-07, + "loss": 1.2256, + "mean_token_accuracy": 0.6449351906776428, + "num_tokens": 9345086.0, + "step": 366 + }, + { + "epoch": 0.040303096859213704, + "grad_norm": 3.711921215057373, + "learning_rate": 2.0087815587266738e-07, + "loss": 1.1761, + "mean_token_accuracy": 0.6754946112632751, + "num_tokens": 9374249.0, + "step": 367 + }, + { + "epoch": 0.040412914561827364, + "grad_norm": 4.2553887367248535, + "learning_rate": 2.0142700329308452e-07, + "loss": 1.1706, + "mean_token_accuracy": 0.6600046753883362, + "num_tokens": 9400142.0, + "step": 368 + }, + { + "epoch": 0.04052273226444103, + "grad_norm": 4.268572807312012, + "learning_rate": 2.0197585071350164e-07, + "loss": 1.1933, + "mean_token_accuracy": 0.6814652681350708, + "num_tokens": 9424556.0, + "step": 369 + }, + { + "epoch": 0.04063254996705469, + "grad_norm": 3.623140573501587, + "learning_rate": 2.0252469813391875e-07, + "loss": 1.0661, + "mean_token_accuracy": 0.6949169635772705, + "num_tokens": 9448711.0, + "step": 370 + }, + { + "epoch": 0.04074236766966835, + "grad_norm": 4.02710485458374, + "learning_rate": 2.030735455543359e-07, + "loss": 1.1237, + "mean_token_accuracy": 0.6927603483200073, + "num_tokens": 9472501.0, + "step": 371 + }, + { + "epoch": 0.04085218537228201, + "grad_norm": 3.0236499309539795, + "learning_rate": 2.03622392974753e-07, + "loss": 1.1521, + "mean_token_accuracy": 0.6668046712875366, + "num_tokens": 9504355.0, + "step": 372 + }, + { + "epoch": 0.04096200307489567, + "grad_norm": 4.469701766967773, + "learning_rate": 2.0417124039517015e-07, + "loss": 1.1863, + "mean_token_accuracy": 0.6688437461853027, + "num_tokens": 9527799.0, + "step": 373 + }, + { + "epoch": 0.04107182077750934, + "grad_norm": 3.024138927459717, + "learning_rate": 2.0472008781558724e-07, + "loss": 1.1745, + "mean_token_accuracy": 0.6651522517204285, + "num_tokens": 9564101.0, + "step": 374 + }, + { + "epoch": 0.041181638480122996, + "grad_norm": 3.66931414604187, + "learning_rate": 2.0526893523600438e-07, + "loss": 1.2609, + "mean_token_accuracy": 0.6427276730537415, + "num_tokens": 9593570.0, + "step": 375 + }, + { + "epoch": 0.041291456182736656, + "grad_norm": 3.6672158241271973, + "learning_rate": 2.058177826564215e-07, + "loss": 1.21, + "mean_token_accuracy": 0.6596144437789917, + "num_tokens": 9623302.0, + "step": 376 + }, + { + "epoch": 0.041401273885350316, + "grad_norm": 5.773532867431641, + "learning_rate": 2.0636663007683864e-07, + "loss": 1.2092, + "mean_token_accuracy": 0.6511343717575073, + "num_tokens": 9644180.0, + "step": 377 + }, + { + "epoch": 0.04151109158796398, + "grad_norm": 4.749263286590576, + "learning_rate": 2.0691547749725575e-07, + "loss": 1.232, + "mean_token_accuracy": 0.646763801574707, + "num_tokens": 9669671.0, + "step": 378 + }, + { + "epoch": 0.04162090929057764, + "grad_norm": 3.0822417736053467, + "learning_rate": 2.0746432491767287e-07, + "loss": 1.2655, + "mean_token_accuracy": 0.6337388753890991, + "num_tokens": 9698683.0, + "step": 379 + }, + { + "epoch": 0.0417307269931913, + "grad_norm": 4.2301554679870605, + "learning_rate": 2.0801317233809e-07, + "loss": 1.2453, + "mean_token_accuracy": 0.6475851535797119, + "num_tokens": 9725289.0, + "step": 380 + }, + { + "epoch": 0.04184054469580496, + "grad_norm": 3.6676247119903564, + "learning_rate": 2.0856201975850713e-07, + "loss": 1.1953, + "mean_token_accuracy": 0.6645731329917908, + "num_tokens": 9756771.0, + "step": 381 + }, + { + "epoch": 0.04195036239841862, + "grad_norm": 3.8085086345672607, + "learning_rate": 2.0911086717892427e-07, + "loss": 1.1494, + "mean_token_accuracy": 0.6654822826385498, + "num_tokens": 9782092.0, + "step": 382 + }, + { + "epoch": 0.04206018010103229, + "grad_norm": 3.436958074569702, + "learning_rate": 2.0965971459934136e-07, + "loss": 1.2549, + "mean_token_accuracy": 0.6475471258163452, + "num_tokens": 9809832.0, + "step": 383 + }, + { + "epoch": 0.04216999780364595, + "grad_norm": 3.7103052139282227, + "learning_rate": 2.102085620197585e-07, + "loss": 1.1565, + "mean_token_accuracy": 0.6673566699028015, + "num_tokens": 9836083.0, + "step": 384 + }, + { + "epoch": 0.04227981550625961, + "grad_norm": 4.011978626251221, + "learning_rate": 2.1075740944017561e-07, + "loss": 1.1882, + "mean_token_accuracy": 0.6588300466537476, + "num_tokens": 9861745.0, + "step": 385 + }, + { + "epoch": 0.04238963320887327, + "grad_norm": 3.8953354358673096, + "learning_rate": 2.1130625686059276e-07, + "loss": 1.1927, + "mean_token_accuracy": 0.6566797494888306, + "num_tokens": 9886605.0, + "step": 386 + }, + { + "epoch": 0.04249945091148693, + "grad_norm": 3.0709218978881836, + "learning_rate": 2.1185510428100987e-07, + "loss": 1.1977, + "mean_token_accuracy": 0.6588414311408997, + "num_tokens": 9914338.0, + "step": 387 + }, + { + "epoch": 0.042609268614100596, + "grad_norm": 3.4494431018829346, + "learning_rate": 2.1240395170142699e-07, + "loss": 1.1581, + "mean_token_accuracy": 0.6689519286155701, + "num_tokens": 9941832.0, + "step": 388 + }, + { + "epoch": 0.042719086316714255, + "grad_norm": 3.8786728382110596, + "learning_rate": 2.129527991218441e-07, + "loss": 1.1383, + "mean_token_accuracy": 0.6746435761451721, + "num_tokens": 9964132.0, + "step": 389 + }, + { + "epoch": 0.042828904019327915, + "grad_norm": 4.765464782714844, + "learning_rate": 2.1350164654226124e-07, + "loss": 1.1501, + "mean_token_accuracy": 0.6630638837814331, + "num_tokens": 9983197.0, + "step": 390 + }, + { + "epoch": 0.042938721721941575, + "grad_norm": 3.839832305908203, + "learning_rate": 2.1405049396267836e-07, + "loss": 1.1482, + "mean_token_accuracy": 0.668270468711853, + "num_tokens": 10010528.0, + "step": 391 + }, + { + "epoch": 0.043048539424555235, + "grad_norm": 4.366824150085449, + "learning_rate": 2.145993413830955e-07, + "loss": 1.1229, + "mean_token_accuracy": 0.6739550828933716, + "num_tokens": 10035807.0, + "step": 392 + }, + { + "epoch": 0.0431583571271689, + "grad_norm": 3.6575570106506348, + "learning_rate": 2.1514818880351262e-07, + "loss": 1.119, + "mean_token_accuracy": 0.6797218322753906, + "num_tokens": 10058082.0, + "step": 393 + }, + { + "epoch": 0.04326817482978256, + "grad_norm": 3.4364829063415527, + "learning_rate": 2.1569703622392973e-07, + "loss": 1.2347, + "mean_token_accuracy": 0.6456183195114136, + "num_tokens": 10083095.0, + "step": 394 + }, + { + "epoch": 0.04337799253239622, + "grad_norm": 3.5117597579956055, + "learning_rate": 2.1624588364434687e-07, + "loss": 1.2123, + "mean_token_accuracy": 0.6689536571502686, + "num_tokens": 10111012.0, + "step": 395 + }, + { + "epoch": 0.04348781023500988, + "grad_norm": 3.6074440479278564, + "learning_rate": 2.16794731064764e-07, + "loss": 1.2793, + "mean_token_accuracy": 0.6496286392211914, + "num_tokens": 10137151.0, + "step": 396 + }, + { + "epoch": 0.04359762793762355, + "grad_norm": 5.184638977050781, + "learning_rate": 2.1734357848518113e-07, + "loss": 1.1394, + "mean_token_accuracy": 0.6711114645004272, + "num_tokens": 10154519.0, + "step": 397 + }, + { + "epoch": 0.04370744564023721, + "grad_norm": 3.405958414077759, + "learning_rate": 2.1789242590559822e-07, + "loss": 1.2515, + "mean_token_accuracy": 0.6436643004417419, + "num_tokens": 10182501.0, + "step": 398 + }, + { + "epoch": 0.04381726334285087, + "grad_norm": 2.88177752494812, + "learning_rate": 2.1844127332601536e-07, + "loss": 1.2184, + "mean_token_accuracy": 0.6502169966697693, + "num_tokens": 10215318.0, + "step": 399 + }, + { + "epoch": 0.04392708104546453, + "grad_norm": 3.604428768157959, + "learning_rate": 2.1899012074643247e-07, + "loss": 1.1319, + "mean_token_accuracy": 0.6724711656570435, + "num_tokens": 10244704.0, + "step": 400 + }, + { + "epoch": 0.04403689874807819, + "grad_norm": 4.019777297973633, + "learning_rate": 2.1953896816684962e-07, + "loss": 1.0758, + "mean_token_accuracy": 0.6882049441337585, + "num_tokens": 10267691.0, + "step": 401 + }, + { + "epoch": 0.044146716450691854, + "grad_norm": 2.998218536376953, + "learning_rate": 2.200878155872667e-07, + "loss": 1.197, + "mean_token_accuracy": 0.6605222225189209, + "num_tokens": 10299091.0, + "step": 402 + }, + { + "epoch": 0.044256534153305514, + "grad_norm": 3.267089605331421, + "learning_rate": 2.2063666300768385e-07, + "loss": 1.2375, + "mean_token_accuracy": 0.6430157423019409, + "num_tokens": 10324611.0, + "step": 403 + }, + { + "epoch": 0.044366351855919174, + "grad_norm": 3.3609507083892822, + "learning_rate": 2.21185510428101e-07, + "loss": 1.3252, + "mean_token_accuracy": 0.6362073421478271, + "num_tokens": 10353167.0, + "step": 404 + }, + { + "epoch": 0.044476169558532834, + "grad_norm": 2.888148546218872, + "learning_rate": 2.217343578485181e-07, + "loss": 1.2217, + "mean_token_accuracy": 0.652732253074646, + "num_tokens": 10383177.0, + "step": 405 + }, + { + "epoch": 0.044585987261146494, + "grad_norm": 4.155028343200684, + "learning_rate": 2.2228320526893525e-07, + "loss": 1.2453, + "mean_token_accuracy": 0.6458773612976074, + "num_tokens": 10409982.0, + "step": 406 + }, + { + "epoch": 0.04469580496376016, + "grad_norm": 2.937037467956543, + "learning_rate": 2.2283205268935233e-07, + "loss": 1.1723, + "mean_token_accuracy": 0.6717647314071655, + "num_tokens": 10437985.0, + "step": 407 + }, + { + "epoch": 0.04480562266637382, + "grad_norm": 3.5562639236450195, + "learning_rate": 2.2338090010976948e-07, + "loss": 1.0941, + "mean_token_accuracy": 0.6853363513946533, + "num_tokens": 10461764.0, + "step": 408 + }, + { + "epoch": 0.04491544036898748, + "grad_norm": 3.5212488174438477, + "learning_rate": 2.239297475301866e-07, + "loss": 1.1278, + "mean_token_accuracy": 0.6721394658088684, + "num_tokens": 10486935.0, + "step": 409 + }, + { + "epoch": 0.04502525807160114, + "grad_norm": 3.6588735580444336, + "learning_rate": 2.2447859495060373e-07, + "loss": 1.1909, + "mean_token_accuracy": 0.6664232015609741, + "num_tokens": 10511948.0, + "step": 410 + }, + { + "epoch": 0.0451350757742148, + "grad_norm": 3.638486623764038, + "learning_rate": 2.2502744237102085e-07, + "loss": 1.1856, + "mean_token_accuracy": 0.6668457388877869, + "num_tokens": 10540081.0, + "step": 411 + }, + { + "epoch": 0.04524489347682847, + "grad_norm": 3.781965494155884, + "learning_rate": 2.2557628979143796e-07, + "loss": 1.1603, + "mean_token_accuracy": 0.677981972694397, + "num_tokens": 10562382.0, + "step": 412 + }, + { + "epoch": 0.04535471117944213, + "grad_norm": 3.688533306121826, + "learning_rate": 2.2612513721185508e-07, + "loss": 1.1565, + "mean_token_accuracy": 0.6631014943122864, + "num_tokens": 10587354.0, + "step": 413 + }, + { + "epoch": 0.04546452888205579, + "grad_norm": 4.158097267150879, + "learning_rate": 2.2667398463227222e-07, + "loss": 1.137, + "mean_token_accuracy": 0.6761300563812256, + "num_tokens": 10606813.0, + "step": 414 + }, + { + "epoch": 0.045574346584669446, + "grad_norm": 3.2834205627441406, + "learning_rate": 2.2722283205268934e-07, + "loss": 1.1882, + "mean_token_accuracy": 0.6563148498535156, + "num_tokens": 10632971.0, + "step": 415 + }, + { + "epoch": 0.04568416428728311, + "grad_norm": 3.3908958435058594, + "learning_rate": 2.2777167947310648e-07, + "loss": 1.194, + "mean_token_accuracy": 0.6600471138954163, + "num_tokens": 10663225.0, + "step": 416 + }, + { + "epoch": 0.04579398198989677, + "grad_norm": 4.306527614593506, + "learning_rate": 2.283205268935236e-07, + "loss": 1.1834, + "mean_token_accuracy": 0.6613795757293701, + "num_tokens": 10684130.0, + "step": 417 + }, + { + "epoch": 0.04590379969251043, + "grad_norm": 3.384212017059326, + "learning_rate": 2.288693743139407e-07, + "loss": 1.1808, + "mean_token_accuracy": 0.6658638715744019, + "num_tokens": 10714965.0, + "step": 418 + }, + { + "epoch": 0.04601361739512409, + "grad_norm": 3.3324050903320312, + "learning_rate": 2.2941822173435785e-07, + "loss": 1.1562, + "mean_token_accuracy": 0.6665259599685669, + "num_tokens": 10742741.0, + "step": 419 + }, + { + "epoch": 0.04612343509773775, + "grad_norm": 3.2829275131225586, + "learning_rate": 2.2996706915477496e-07, + "loss": 1.1447, + "mean_token_accuracy": 0.6711521148681641, + "num_tokens": 10765170.0, + "step": 420 + }, + { + "epoch": 0.04623325280035142, + "grad_norm": 3.6673688888549805, + "learning_rate": 2.3051591657519208e-07, + "loss": 1.1019, + "mean_token_accuracy": 0.6761807799339294, + "num_tokens": 10789701.0, + "step": 421 + }, + { + "epoch": 0.04634307050296508, + "grad_norm": 4.485995769500732, + "learning_rate": 2.310647639956092e-07, + "loss": 1.2889, + "mean_token_accuracy": 0.6340348124504089, + "num_tokens": 10813774.0, + "step": 422 + }, + { + "epoch": 0.04645288820557874, + "grad_norm": 3.494939088821411, + "learning_rate": 2.3161361141602634e-07, + "loss": 1.1995, + "mean_token_accuracy": 0.6624654531478882, + "num_tokens": 10837484.0, + "step": 423 + }, + { + "epoch": 0.0465627059081924, + "grad_norm": 3.7744832038879395, + "learning_rate": 2.3216245883644345e-07, + "loss": 1.1147, + "mean_token_accuracy": 0.6776486039161682, + "num_tokens": 10857625.0, + "step": 424 + }, + { + "epoch": 0.04667252361080606, + "grad_norm": 3.6291744709014893, + "learning_rate": 2.327113062568606e-07, + "loss": 1.1879, + "mean_token_accuracy": 0.6584762930870056, + "num_tokens": 10879524.0, + "step": 425 + }, + { + "epoch": 0.046782341313419726, + "grad_norm": 3.314236640930176, + "learning_rate": 2.3326015367727768e-07, + "loss": 1.0899, + "mean_token_accuracy": 0.6855401396751404, + "num_tokens": 10904085.0, + "step": 426 + }, + { + "epoch": 0.046892159016033386, + "grad_norm": 3.5933823585510254, + "learning_rate": 2.3380900109769482e-07, + "loss": 1.2448, + "mean_token_accuracy": 0.6590924859046936, + "num_tokens": 10928698.0, + "step": 427 + }, + { + "epoch": 0.047001976718647046, + "grad_norm": 3.206603527069092, + "learning_rate": 2.3435784851811197e-07, + "loss": 1.1267, + "mean_token_accuracy": 0.6689953804016113, + "num_tokens": 10955323.0, + "step": 428 + }, + { + "epoch": 0.047111794421260705, + "grad_norm": 2.991243362426758, + "learning_rate": 2.3490669593852908e-07, + "loss": 1.1426, + "mean_token_accuracy": 0.6687731742858887, + "num_tokens": 10983533.0, + "step": 429 + }, + { + "epoch": 0.047221612123874365, + "grad_norm": 3.0867669582366943, + "learning_rate": 2.3545554335894622e-07, + "loss": 1.2599, + "mean_token_accuracy": 0.6400494575500488, + "num_tokens": 11015269.0, + "step": 430 + }, + { + "epoch": 0.04733142982648803, + "grad_norm": 3.4198384284973145, + "learning_rate": 2.360043907793633e-07, + "loss": 1.1438, + "mean_token_accuracy": 0.673407793045044, + "num_tokens": 11039311.0, + "step": 431 + }, + { + "epoch": 0.04744124752910169, + "grad_norm": 2.909893035888672, + "learning_rate": 2.3655323819978045e-07, + "loss": 1.1965, + "mean_token_accuracy": 0.6521835327148438, + "num_tokens": 11067134.0, + "step": 432 + }, + { + "epoch": 0.04755106523171535, + "grad_norm": 3.1667168140411377, + "learning_rate": 2.3710208562019757e-07, + "loss": 1.1372, + "mean_token_accuracy": 0.6756761074066162, + "num_tokens": 11092533.0, + "step": 433 + }, + { + "epoch": 0.04766088293432901, + "grad_norm": 2.6944737434387207, + "learning_rate": 2.376509330406147e-07, + "loss": 1.222, + "mean_token_accuracy": 0.6482876539230347, + "num_tokens": 11123846.0, + "step": 434 + }, + { + "epoch": 0.04777070063694268, + "grad_norm": 3.4058749675750732, + "learning_rate": 2.3819978046103183e-07, + "loss": 1.1694, + "mean_token_accuracy": 0.6594732999801636, + "num_tokens": 11143835.0, + "step": 435 + }, + { + "epoch": 0.04788051833955634, + "grad_norm": 3.7248458862304688, + "learning_rate": 2.3874862788144894e-07, + "loss": 1.1185, + "mean_token_accuracy": 0.6784825325012207, + "num_tokens": 11164146.0, + "step": 436 + }, + { + "epoch": 0.04799033604217, + "grad_norm": 3.1146609783172607, + "learning_rate": 2.3929747530186606e-07, + "loss": 1.0946, + "mean_token_accuracy": 0.6846172213554382, + "num_tokens": 11192751.0, + "step": 437 + }, + { + "epoch": 0.04810015374478366, + "grad_norm": 4.009331703186035, + "learning_rate": 2.3984632272228317e-07, + "loss": 1.2073, + "mean_token_accuracy": 0.6531702876091003, + "num_tokens": 11213181.0, + "step": 438 + }, + { + "epoch": 0.04820997144739732, + "grad_norm": 3.2112724781036377, + "learning_rate": 2.4039517014270034e-07, + "loss": 1.168, + "mean_token_accuracy": 0.6570603847503662, + "num_tokens": 11241950.0, + "step": 439 + }, + { + "epoch": 0.048319789150010985, + "grad_norm": 4.453896522521973, + "learning_rate": 2.4094401756311745e-07, + "loss": 1.2182, + "mean_token_accuracy": 0.6505071520805359, + "num_tokens": 11261970.0, + "step": 440 + }, + { + "epoch": 0.048429606852624645, + "grad_norm": 3.2072997093200684, + "learning_rate": 2.4149286498353457e-07, + "loss": 1.1819, + "mean_token_accuracy": 0.6565448045730591, + "num_tokens": 11291513.0, + "step": 441 + }, + { + "epoch": 0.048539424555238304, + "grad_norm": 3.271212577819824, + "learning_rate": 2.420417124039517e-07, + "loss": 1.1196, + "mean_token_accuracy": 0.6738558411598206, + "num_tokens": 11313296.0, + "step": 442 + }, + { + "epoch": 0.048649242257851964, + "grad_norm": 3.0094218254089355, + "learning_rate": 2.425905598243688e-07, + "loss": 1.2248, + "mean_token_accuracy": 0.649979293346405, + "num_tokens": 11343903.0, + "step": 443 + }, + { + "epoch": 0.048759059960465624, + "grad_norm": 2.875504732131958, + "learning_rate": 2.431394072447859e-07, + "loss": 1.1441, + "mean_token_accuracy": 0.675141453742981, + "num_tokens": 11371729.0, + "step": 444 + }, + { + "epoch": 0.04886887766307929, + "grad_norm": 3.193444013595581, + "learning_rate": 2.436882546652031e-07, + "loss": 1.1476, + "mean_token_accuracy": 0.6718856692314148, + "num_tokens": 11395166.0, + "step": 445 + }, + { + "epoch": 0.04897869536569295, + "grad_norm": 3.2464723587036133, + "learning_rate": 2.442371020856202e-07, + "loss": 1.0675, + "mean_token_accuracy": 0.688653826713562, + "num_tokens": 11418239.0, + "step": 446 + }, + { + "epoch": 0.04908851306830661, + "grad_norm": 3.637112617492676, + "learning_rate": 2.447859495060373e-07, + "loss": 1.1672, + "mean_token_accuracy": 0.661407470703125, + "num_tokens": 11437987.0, + "step": 447 + }, + { + "epoch": 0.04919833077092027, + "grad_norm": 3.6163103580474854, + "learning_rate": 2.4533479692645443e-07, + "loss": 1.1941, + "mean_token_accuracy": 0.6520886421203613, + "num_tokens": 11458480.0, + "step": 448 + }, + { + "epoch": 0.04930814847353393, + "grad_norm": 3.104384422302246, + "learning_rate": 2.4588364434687154e-07, + "loss": 1.1432, + "mean_token_accuracy": 0.6711790561676025, + "num_tokens": 11483223.0, + "step": 449 + }, + { + "epoch": 0.0494179661761476, + "grad_norm": 3.2915234565734863, + "learning_rate": 2.4643249176728866e-07, + "loss": 1.1317, + "mean_token_accuracy": 0.6786198616027832, + "num_tokens": 11509950.0, + "step": 450 + }, + { + "epoch": 0.04952778387876126, + "grad_norm": 3.6694202423095703, + "learning_rate": 2.4698133918770583e-07, + "loss": 1.1983, + "mean_token_accuracy": 0.655379056930542, + "num_tokens": 11532421.0, + "step": 451 + }, + { + "epoch": 0.04963760158137492, + "grad_norm": 3.1870038509368896, + "learning_rate": 2.4753018660812294e-07, + "loss": 1.1327, + "mean_token_accuracy": 0.6700639724731445, + "num_tokens": 11556043.0, + "step": 452 + }, + { + "epoch": 0.04974741928398858, + "grad_norm": 3.120913505554199, + "learning_rate": 2.4807903402854006e-07, + "loss": 1.189, + "mean_token_accuracy": 0.6544355750083923, + "num_tokens": 11580844.0, + "step": 453 + }, + { + "epoch": 0.049857236986602244, + "grad_norm": 3.139374256134033, + "learning_rate": 2.486278814489572e-07, + "loss": 1.1864, + "mean_token_accuracy": 0.65957111120224, + "num_tokens": 11602968.0, + "step": 454 + }, + { + "epoch": 0.0499670546892159, + "grad_norm": 3.9725425243377686, + "learning_rate": 2.491767288693743e-07, + "loss": 1.1119, + "mean_token_accuracy": 0.6680035591125488, + "num_tokens": 11624823.0, + "step": 455 + }, + { + "epoch": 0.05007687239182956, + "grad_norm": 3.299583911895752, + "learning_rate": 2.4972557628979146e-07, + "loss": 1.2495, + "mean_token_accuracy": 0.6425378322601318, + "num_tokens": 11654828.0, + "step": 456 + }, + { + "epoch": 0.05018669009444322, + "grad_norm": 2.9051015377044678, + "learning_rate": 2.5027442371020857e-07, + "loss": 1.1416, + "mean_token_accuracy": 0.6871888637542725, + "num_tokens": 11682116.0, + "step": 457 + }, + { + "epoch": 0.05029650779705688, + "grad_norm": 3.0268282890319824, + "learning_rate": 2.508232711306257e-07, + "loss": 1.1847, + "mean_token_accuracy": 0.6600742936134338, + "num_tokens": 11707263.0, + "step": 458 + }, + { + "epoch": 0.05040632549967055, + "grad_norm": 2.832902669906616, + "learning_rate": 2.513721185510428e-07, + "loss": 1.1675, + "mean_token_accuracy": 0.6695923209190369, + "num_tokens": 11734837.0, + "step": 459 + }, + { + "epoch": 0.05051614320228421, + "grad_norm": 2.6057660579681396, + "learning_rate": 2.519209659714599e-07, + "loss": 1.2149, + "mean_token_accuracy": 0.6507817506790161, + "num_tokens": 11769476.0, + "step": 460 + }, + { + "epoch": 0.05062596090489787, + "grad_norm": 2.96091365814209, + "learning_rate": 2.524698133918771e-07, + "loss": 1.1055, + "mean_token_accuracy": 0.6770296096801758, + "num_tokens": 11798574.0, + "step": 461 + }, + { + "epoch": 0.05073577860751153, + "grad_norm": 3.864748954772949, + "learning_rate": 2.5301866081229415e-07, + "loss": 1.1826, + "mean_token_accuracy": 0.659891664981842, + "num_tokens": 11825216.0, + "step": 462 + }, + { + "epoch": 0.05084559631012519, + "grad_norm": 2.9474010467529297, + "learning_rate": 2.5356750823271126e-07, + "loss": 1.146, + "mean_token_accuracy": 0.6686737537384033, + "num_tokens": 11853278.0, + "step": 463 + }, + { + "epoch": 0.050955414012738856, + "grad_norm": 2.934345006942749, + "learning_rate": 2.5411635565312843e-07, + "loss": 1.155, + "mean_token_accuracy": 0.665048360824585, + "num_tokens": 11877729.0, + "step": 464 + }, + { + "epoch": 0.051065231715352516, + "grad_norm": 2.8513998985290527, + "learning_rate": 2.5466520307354555e-07, + "loss": 1.1181, + "mean_token_accuracy": 0.6822975277900696, + "num_tokens": 11908587.0, + "step": 465 + }, + { + "epoch": 0.051175049417966176, + "grad_norm": 3.472447633743286, + "learning_rate": 2.5521405049396266e-07, + "loss": 1.1107, + "mean_token_accuracy": 0.6743884086608887, + "num_tokens": 11929499.0, + "step": 466 + }, + { + "epoch": 0.051284867120579836, + "grad_norm": 3.751211166381836, + "learning_rate": 2.557628979143798e-07, + "loss": 1.1241, + "mean_token_accuracy": 0.6748337745666504, + "num_tokens": 11952254.0, + "step": 467 + }, + { + "epoch": 0.051394684823193496, + "grad_norm": 3.1591737270355225, + "learning_rate": 2.5631174533479695e-07, + "loss": 1.0255, + "mean_token_accuracy": 0.6969254612922668, + "num_tokens": 11975920.0, + "step": 468 + }, + { + "epoch": 0.05150450252580716, + "grad_norm": 3.061526298522949, + "learning_rate": 2.5686059275521406e-07, + "loss": 1.1664, + "mean_token_accuracy": 0.6674725413322449, + "num_tokens": 12000570.0, + "step": 469 + }, + { + "epoch": 0.05161432022842082, + "grad_norm": 2.728339433670044, + "learning_rate": 2.574094401756312e-07, + "loss": 1.169, + "mean_token_accuracy": 0.6673228740692139, + "num_tokens": 12026496.0, + "step": 470 + }, + { + "epoch": 0.05172413793103448, + "grad_norm": 3.0500779151916504, + "learning_rate": 2.5795828759604824e-07, + "loss": 1.3119, + "mean_token_accuracy": 0.6300621628761292, + "num_tokens": 12056978.0, + "step": 471 + }, + { + "epoch": 0.05183395563364814, + "grad_norm": 3.0879197120666504, + "learning_rate": 2.585071350164654e-07, + "loss": 1.1456, + "mean_token_accuracy": 0.663142740726471, + "num_tokens": 12079553.0, + "step": 472 + }, + { + "epoch": 0.05194377333626181, + "grad_norm": 2.8943235874176025, + "learning_rate": 2.590559824368825e-07, + "loss": 1.0993, + "mean_token_accuracy": 0.679355800151825, + "num_tokens": 12105013.0, + "step": 473 + }, + { + "epoch": 0.05205359103887547, + "grad_norm": 3.05190372467041, + "learning_rate": 2.5960482985729964e-07, + "loss": 1.1033, + "mean_token_accuracy": 0.6804770827293396, + "num_tokens": 12127266.0, + "step": 474 + }, + { + "epoch": 0.05216340874148913, + "grad_norm": 2.698665142059326, + "learning_rate": 2.601536772777168e-07, + "loss": 1.1511, + "mean_token_accuracy": 0.6741541624069214, + "num_tokens": 12155889.0, + "step": 475 + }, + { + "epoch": 0.05227322644410279, + "grad_norm": 2.8912854194641113, + "learning_rate": 2.607025246981339e-07, + "loss": 1.1627, + "mean_token_accuracy": 0.6666158437728882, + "num_tokens": 12182764.0, + "step": 476 + }, + { + "epoch": 0.05238304414671645, + "grad_norm": 3.0239369869232178, + "learning_rate": 2.6125137211855104e-07, + "loss": 1.2243, + "mean_token_accuracy": 0.6605309844017029, + "num_tokens": 12208149.0, + "step": 477 + }, + { + "epoch": 0.052492861849330115, + "grad_norm": 3.1189894676208496, + "learning_rate": 2.6180021953896815e-07, + "loss": 1.0997, + "mean_token_accuracy": 0.6829297542572021, + "num_tokens": 12229726.0, + "step": 478 + }, + { + "epoch": 0.052602679551943775, + "grad_norm": 2.860358476638794, + "learning_rate": 2.623490669593853e-07, + "loss": 1.2339, + "mean_token_accuracy": 0.6468026638031006, + "num_tokens": 12254289.0, + "step": 479 + }, + { + "epoch": 0.052712497254557435, + "grad_norm": 2.8653922080993652, + "learning_rate": 2.6289791437980244e-07, + "loss": 1.0655, + "mean_token_accuracy": 0.6880159378051758, + "num_tokens": 12278764.0, + "step": 480 + }, + { + "epoch": 0.052822314957171095, + "grad_norm": 2.8469693660736084, + "learning_rate": 2.634467618002195e-07, + "loss": 1.1193, + "mean_token_accuracy": 0.6770780682563782, + "num_tokens": 12305620.0, + "step": 481 + }, + { + "epoch": 0.052932132659784754, + "grad_norm": 2.812034845352173, + "learning_rate": 2.639956092206366e-07, + "loss": 1.032, + "mean_token_accuracy": 0.6950036883354187, + "num_tokens": 12331802.0, + "step": 482 + }, + { + "epoch": 0.05304195036239842, + "grad_norm": 2.507997751235962, + "learning_rate": 2.645444566410538e-07, + "loss": 1.071, + "mean_token_accuracy": 0.6811461448669434, + "num_tokens": 12362433.0, + "step": 483 + }, + { + "epoch": 0.05315176806501208, + "grad_norm": 2.570697784423828, + "learning_rate": 2.650933040614709e-07, + "loss": 1.2798, + "mean_token_accuracy": 0.641563892364502, + "num_tokens": 12391939.0, + "step": 484 + }, + { + "epoch": 0.05326158576762574, + "grad_norm": 3.5249526500701904, + "learning_rate": 2.65642151481888e-07, + "loss": 1.1694, + "mean_token_accuracy": 0.6612480878829956, + "num_tokens": 12412212.0, + "step": 485 + }, + { + "epoch": 0.0533714034702394, + "grad_norm": 3.243630886077881, + "learning_rate": 2.6619099890230513e-07, + "loss": 1.2356, + "mean_token_accuracy": 0.6507245302200317, + "num_tokens": 12437053.0, + "step": 486 + }, + { + "epoch": 0.05348122117285306, + "grad_norm": 3.2384700775146484, + "learning_rate": 2.667398463227223e-07, + "loss": 1.0706, + "mean_token_accuracy": 0.6823394298553467, + "num_tokens": 12459635.0, + "step": 487 + }, + { + "epoch": 0.05359103887546673, + "grad_norm": 2.659662961959839, + "learning_rate": 2.672886937431394e-07, + "loss": 1.1203, + "mean_token_accuracy": 0.6781542301177979, + "num_tokens": 12488734.0, + "step": 488 + }, + { + "epoch": 0.05370085657808039, + "grad_norm": 2.928159713745117, + "learning_rate": 2.678375411635565e-07, + "loss": 1.0901, + "mean_token_accuracy": 0.6775121688842773, + "num_tokens": 12512692.0, + "step": 489 + }, + { + "epoch": 0.05381067428069405, + "grad_norm": 2.998805046081543, + "learning_rate": 2.6838638858397364e-07, + "loss": 1.1008, + "mean_token_accuracy": 0.6754859685897827, + "num_tokens": 12534988.0, + "step": 490 + }, + { + "epoch": 0.05392049198330771, + "grad_norm": 3.0806362628936768, + "learning_rate": 2.6893523600439076e-07, + "loss": 1.1448, + "mean_token_accuracy": 0.6699921488761902, + "num_tokens": 12559613.0, + "step": 491 + }, + { + "epoch": 0.054030309685921374, + "grad_norm": 3.318110942840576, + "learning_rate": 2.6948408342480787e-07, + "loss": 1.115, + "mean_token_accuracy": 0.6727299690246582, + "num_tokens": 12582802.0, + "step": 492 + }, + { + "epoch": 0.054140127388535034, + "grad_norm": 2.8725171089172363, + "learning_rate": 2.70032930845225e-07, + "loss": 1.0848, + "mean_token_accuracy": 0.6808451414108276, + "num_tokens": 12609562.0, + "step": 493 + }, + { + "epoch": 0.054249945091148694, + "grad_norm": 3.0242481231689453, + "learning_rate": 2.7058177826564215e-07, + "loss": 1.1864, + "mean_token_accuracy": 0.6643209457397461, + "num_tokens": 12636530.0, + "step": 494 + }, + { + "epoch": 0.05435976279376235, + "grad_norm": 3.14785099029541, + "learning_rate": 2.7113062568605927e-07, + "loss": 1.1146, + "mean_token_accuracy": 0.6773680448532104, + "num_tokens": 12659377.0, + "step": 495 + }, + { + "epoch": 0.05446958049637601, + "grad_norm": 2.642909049987793, + "learning_rate": 2.716794731064764e-07, + "loss": 1.1572, + "mean_token_accuracy": 0.6672093868255615, + "num_tokens": 12686690.0, + "step": 496 + }, + { + "epoch": 0.05457939819898968, + "grad_norm": 3.1507723331451416, + "learning_rate": 2.722283205268935e-07, + "loss": 1.0759, + "mean_token_accuracy": 0.6871058940887451, + "num_tokens": 12707192.0, + "step": 497 + }, + { + "epoch": 0.05468921590160334, + "grad_norm": 2.763690948486328, + "learning_rate": 2.7277716794731067e-07, + "loss": 1.2387, + "mean_token_accuracy": 0.6447410583496094, + "num_tokens": 12735934.0, + "step": 498 + }, + { + "epoch": 0.054799033604217, + "grad_norm": 2.8265914916992188, + "learning_rate": 2.733260153677278e-07, + "loss": 1.1817, + "mean_token_accuracy": 0.6579384207725525, + "num_tokens": 12764925.0, + "step": 499 + }, + { + "epoch": 0.05490885130683066, + "grad_norm": 2.9403467178344727, + "learning_rate": 2.7387486278814485e-07, + "loss": 1.0984, + "mean_token_accuracy": 0.6762644052505493, + "num_tokens": 12788148.0, + "step": 500 + }, + { + "epoch": 0.05501866900944432, + "grad_norm": 2.702709197998047, + "learning_rate": 2.74423710208562e-07, + "loss": 1.1159, + "mean_token_accuracy": 0.6753844022750854, + "num_tokens": 12812727.0, + "step": 501 + }, + { + "epoch": 0.055128486712057986, + "grad_norm": 2.4502596855163574, + "learning_rate": 2.7497255762897913e-07, + "loss": 1.2186, + "mean_token_accuracy": 0.651238203048706, + "num_tokens": 12843363.0, + "step": 502 + }, + { + "epoch": 0.055238304414671646, + "grad_norm": 3.1884233951568604, + "learning_rate": 2.7552140504939624e-07, + "loss": 1.1712, + "mean_token_accuracy": 0.6508257985115051, + "num_tokens": 12863909.0, + "step": 503 + }, + { + "epoch": 0.055348122117285306, + "grad_norm": 2.5589919090270996, + "learning_rate": 2.7607025246981336e-07, + "loss": 1.111, + "mean_token_accuracy": 0.6777164936065674, + "num_tokens": 12891970.0, + "step": 504 + }, + { + "epoch": 0.055457939819898966, + "grad_norm": 3.373656749725342, + "learning_rate": 2.7661909989023053e-07, + "loss": 1.0278, + "mean_token_accuracy": 0.6972694396972656, + "num_tokens": 12913180.0, + "step": 505 + }, + { + "epoch": 0.055567757522512626, + "grad_norm": 2.858383893966675, + "learning_rate": 2.7716794731064764e-07, + "loss": 1.1832, + "mean_token_accuracy": 0.6640771627426147, + "num_tokens": 12940680.0, + "step": 506 + }, + { + "epoch": 0.05567757522512629, + "grad_norm": 3.1474525928497314, + "learning_rate": 2.7771679473106476e-07, + "loss": 1.088, + "mean_token_accuracy": 0.683661937713623, + "num_tokens": 12963813.0, + "step": 507 + }, + { + "epoch": 0.05578739292773995, + "grad_norm": 3.0919301509857178, + "learning_rate": 2.7826564215148187e-07, + "loss": 1.1522, + "mean_token_accuracy": 0.6679531931877136, + "num_tokens": 12985195.0, + "step": 508 + }, + { + "epoch": 0.05589721063035361, + "grad_norm": 2.839087724685669, + "learning_rate": 2.78814489571899e-07, + "loss": 1.0202, + "mean_token_accuracy": 0.6912680864334106, + "num_tokens": 13006656.0, + "step": 509 + }, + { + "epoch": 0.05600702833296727, + "grad_norm": 2.9322948455810547, + "learning_rate": 2.793633369923161e-07, + "loss": 1.0467, + "mean_token_accuracy": 0.6970127820968628, + "num_tokens": 13026122.0, + "step": 510 + }, + { + "epoch": 0.05611684603558094, + "grad_norm": 2.718881130218506, + "learning_rate": 2.799121844127332e-07, + "loss": 1.1717, + "mean_token_accuracy": 0.663446307182312, + "num_tokens": 13051570.0, + "step": 511 + }, + { + "epoch": 0.0562266637381946, + "grad_norm": 3.0106136798858643, + "learning_rate": 2.804610318331504e-07, + "loss": 1.1251, + "mean_token_accuracy": 0.6755059957504272, + "num_tokens": 13073766.0, + "step": 512 + }, + { + "epoch": 0.05633648144080826, + "grad_norm": 3.2423365116119385, + "learning_rate": 2.810098792535675e-07, + "loss": 1.0498, + "mean_token_accuracy": 0.6919777393341064, + "num_tokens": 13094973.0, + "step": 513 + }, + { + "epoch": 0.05644629914342192, + "grad_norm": 3.585421562194824, + "learning_rate": 2.815587266739846e-07, + "loss": 1.0995, + "mean_token_accuracy": 0.6770241856575012, + "num_tokens": 13112632.0, + "step": 514 + }, + { + "epoch": 0.05655611684603558, + "grad_norm": 3.1150009632110596, + "learning_rate": 2.8210757409440173e-07, + "loss": 1.1145, + "mean_token_accuracy": 0.6766128540039062, + "num_tokens": 13134381.0, + "step": 515 + }, + { + "epoch": 0.056665934548649245, + "grad_norm": 2.5707106590270996, + "learning_rate": 2.826564215148189e-07, + "loss": 1.1933, + "mean_token_accuracy": 0.6511658430099487, + "num_tokens": 13162396.0, + "step": 516 + }, + { + "epoch": 0.056775752251262905, + "grad_norm": 2.834235191345215, + "learning_rate": 2.83205268935236e-07, + "loss": 1.0717, + "mean_token_accuracy": 0.6963566541671753, + "num_tokens": 13185170.0, + "step": 517 + }, + { + "epoch": 0.056885569953876565, + "grad_norm": 2.4463768005371094, + "learning_rate": 2.8375411635565313e-07, + "loss": 1.1197, + "mean_token_accuracy": 0.675258219242096, + "num_tokens": 13216566.0, + "step": 518 + }, + { + "epoch": 0.056995387656490225, + "grad_norm": 2.6684420108795166, + "learning_rate": 2.843029637760702e-07, + "loss": 1.191, + "mean_token_accuracy": 0.6578055620193481, + "num_tokens": 13244028.0, + "step": 519 + }, + { + "epoch": 0.057105205359103885, + "grad_norm": 2.8252527713775635, + "learning_rate": 2.8485181119648736e-07, + "loss": 1.0203, + "mean_token_accuracy": 0.6978711485862732, + "num_tokens": 13270417.0, + "step": 520 + }, + { + "epoch": 0.05721502306171755, + "grad_norm": 2.718156337738037, + "learning_rate": 2.854006586169045e-07, + "loss": 1.1211, + "mean_token_accuracy": 0.6760726571083069, + "num_tokens": 13296284.0, + "step": 521 + }, + { + "epoch": 0.05732484076433121, + "grad_norm": 2.4290332794189453, + "learning_rate": 2.859495060373216e-07, + "loss": 1.2109, + "mean_token_accuracy": 0.65793377161026, + "num_tokens": 13325777.0, + "step": 522 + }, + { + "epoch": 0.05743465846694487, + "grad_norm": 2.6973493099212646, + "learning_rate": 2.8649835345773876e-07, + "loss": 1.2045, + "mean_token_accuracy": 0.6629723310470581, + "num_tokens": 13350256.0, + "step": 523 + }, + { + "epoch": 0.05754447616955853, + "grad_norm": 2.877103567123413, + "learning_rate": 2.870472008781559e-07, + "loss": 1.1586, + "mean_token_accuracy": 0.6597898006439209, + "num_tokens": 13371937.0, + "step": 524 + }, + { + "epoch": 0.05765429387217219, + "grad_norm": 2.752389907836914, + "learning_rate": 2.87596048298573e-07, + "loss": 1.1805, + "mean_token_accuracy": 0.6546033620834351, + "num_tokens": 13397020.0, + "step": 525 + }, + { + "epoch": 0.05776411157478586, + "grad_norm": 2.9767158031463623, + "learning_rate": 2.881448957189901e-07, + "loss": 1.1116, + "mean_token_accuracy": 0.6633789539337158, + "num_tokens": 13418705.0, + "step": 526 + }, + { + "epoch": 0.05787392927739952, + "grad_norm": 2.821265459060669, + "learning_rate": 2.886937431394073e-07, + "loss": 1.1337, + "mean_token_accuracy": 0.6668974757194519, + "num_tokens": 13441901.0, + "step": 527 + }, + { + "epoch": 0.05798374698001318, + "grad_norm": 2.711885452270508, + "learning_rate": 2.8924259055982434e-07, + "loss": 1.1568, + "mean_token_accuracy": 0.6776034832000732, + "num_tokens": 13466397.0, + "step": 528 + }, + { + "epoch": 0.05809356468262684, + "grad_norm": 3.0137741565704346, + "learning_rate": 2.8979143798024145e-07, + "loss": 1.128, + "mean_token_accuracy": 0.6691799163818359, + "num_tokens": 13486653.0, + "step": 529 + }, + { + "epoch": 0.058203382385240504, + "grad_norm": 2.742936372756958, + "learning_rate": 2.9034028540065857e-07, + "loss": 1.1138, + "mean_token_accuracy": 0.6709808111190796, + "num_tokens": 13511184.0, + "step": 530 + }, + { + "epoch": 0.058313200087854164, + "grad_norm": 2.88207745552063, + "learning_rate": 2.9088913282107574e-07, + "loss": 1.0897, + "mean_token_accuracy": 0.6821142435073853, + "num_tokens": 13535787.0, + "step": 531 + }, + { + "epoch": 0.058423017790467824, + "grad_norm": 2.685570240020752, + "learning_rate": 2.9143798024149285e-07, + "loss": 1.1554, + "mean_token_accuracy": 0.6713097095489502, + "num_tokens": 13561325.0, + "step": 532 + }, + { + "epoch": 0.058532835493081484, + "grad_norm": 2.63828706741333, + "learning_rate": 2.9198682766190997e-07, + "loss": 1.0534, + "mean_token_accuracy": 0.6860761642456055, + "num_tokens": 13587477.0, + "step": 533 + }, + { + "epoch": 0.058642653195695144, + "grad_norm": 3.048537254333496, + "learning_rate": 2.9253567508232713e-07, + "loss": 1.0034, + "mean_token_accuracy": 0.7041735649108887, + "num_tokens": 13607104.0, + "step": 534 + }, + { + "epoch": 0.05875247089830881, + "grad_norm": 3.3394548892974854, + "learning_rate": 2.9308452250274425e-07, + "loss": 1.1225, + "mean_token_accuracy": 0.6696520447731018, + "num_tokens": 13625485.0, + "step": 535 + }, + { + "epoch": 0.05886228860092247, + "grad_norm": 2.847606897354126, + "learning_rate": 2.9363336992316136e-07, + "loss": 1.1114, + "mean_token_accuracy": 0.678138256072998, + "num_tokens": 13649512.0, + "step": 536 + }, + { + "epoch": 0.05897210630353613, + "grad_norm": 2.9257071018218994, + "learning_rate": 2.941822173435785e-07, + "loss": 1.0624, + "mean_token_accuracy": 0.6848617196083069, + "num_tokens": 13670913.0, + "step": 537 + }, + { + "epoch": 0.05908192400614979, + "grad_norm": 2.345378875732422, + "learning_rate": 2.947310647639956e-07, + "loss": 1.1106, + "mean_token_accuracy": 0.6810805797576904, + "num_tokens": 13702919.0, + "step": 538 + }, + { + "epoch": 0.05919174170876345, + "grad_norm": 2.3207361698150635, + "learning_rate": 2.952799121844127e-07, + "loss": 1.1769, + "mean_token_accuracy": 0.6550666689872742, + "num_tokens": 13733919.0, + "step": 539 + }, + { + "epoch": 0.05930155941137712, + "grad_norm": 2.54240083694458, + "learning_rate": 2.958287596048298e-07, + "loss": 1.2054, + "mean_token_accuracy": 0.6534004211425781, + "num_tokens": 13762959.0, + "step": 540 + }, + { + "epoch": 0.059411377113990776, + "grad_norm": 2.859239101409912, + "learning_rate": 2.9637760702524694e-07, + "loss": 1.1834, + "mean_token_accuracy": 0.6613554358482361, + "num_tokens": 13786659.0, + "step": 541 + }, + { + "epoch": 0.059521194816604436, + "grad_norm": 2.497947931289673, + "learning_rate": 2.969264544456641e-07, + "loss": 1.0962, + "mean_token_accuracy": 0.6834759712219238, + "num_tokens": 13813781.0, + "step": 542 + }, + { + "epoch": 0.059631012519218096, + "grad_norm": 2.6186108589172363, + "learning_rate": 2.974753018660812e-07, + "loss": 1.1279, + "mean_token_accuracy": 0.6822501420974731, + "num_tokens": 13839309.0, + "step": 543 + }, + { + "epoch": 0.059740830221831756, + "grad_norm": 2.4768266677856445, + "learning_rate": 2.9802414928649834e-07, + "loss": 1.1782, + "mean_token_accuracy": 0.6607061624526978, + "num_tokens": 13869038.0, + "step": 544 + }, + { + "epoch": 0.05985064792444542, + "grad_norm": 2.6816959381103516, + "learning_rate": 2.9857299670691546e-07, + "loss": 1.1895, + "mean_token_accuracy": 0.6561024188995361, + "num_tokens": 13894689.0, + "step": 545 + }, + { + "epoch": 0.05996046562705908, + "grad_norm": 2.5929877758026123, + "learning_rate": 2.991218441273326e-07, + "loss": 1.1014, + "mean_token_accuracy": 0.6758500933647156, + "num_tokens": 13923958.0, + "step": 546 + }, + { + "epoch": 0.06007028332967274, + "grad_norm": 2.5225517749786377, + "learning_rate": 2.996706915477497e-07, + "loss": 1.1169, + "mean_token_accuracy": 0.6754350662231445, + "num_tokens": 13952991.0, + "step": 547 + }, + { + "epoch": 0.0601801010322864, + "grad_norm": 2.2489969730377197, + "learning_rate": 3.002195389681668e-07, + "loss": 1.2453, + "mean_token_accuracy": 0.638812780380249, + "num_tokens": 13990058.0, + "step": 548 + }, + { + "epoch": 0.06028991873490007, + "grad_norm": 2.4870193004608154, + "learning_rate": 3.0076838638858397e-07, + "loss": 1.1056, + "mean_token_accuracy": 0.675102174282074, + "num_tokens": 14018410.0, + "step": 549 + }, + { + "epoch": 0.06039973643751373, + "grad_norm": 2.6048877239227295, + "learning_rate": 3.013172338090011e-07, + "loss": 1.1479, + "mean_token_accuracy": 0.6659572720527649, + "num_tokens": 14046469.0, + "step": 550 + }, + { + "epoch": 0.06050955414012739, + "grad_norm": 2.4725704193115234, + "learning_rate": 3.018660812294182e-07, + "loss": 1.0784, + "mean_token_accuracy": 0.6814204454421997, + "num_tokens": 14075094.0, + "step": 551 + }, + { + "epoch": 0.06061937184274105, + "grad_norm": 2.624206066131592, + "learning_rate": 3.024149286498353e-07, + "loss": 1.1811, + "mean_token_accuracy": 0.6581487655639648, + "num_tokens": 14100496.0, + "step": 552 + }, + { + "epoch": 0.06072918954535471, + "grad_norm": 2.4121036529541016, + "learning_rate": 3.029637760702525e-07, + "loss": 1.1031, + "mean_token_accuracy": 0.6819007396697998, + "num_tokens": 14131600.0, + "step": 553 + }, + { + "epoch": 0.060839007247968376, + "grad_norm": 2.3669536113739014, + "learning_rate": 3.035126234906696e-07, + "loss": 1.2297, + "mean_token_accuracy": 0.6505866050720215, + "num_tokens": 14164442.0, + "step": 554 + }, + { + "epoch": 0.060948824950582035, + "grad_norm": 2.6850829124450684, + "learning_rate": 3.040614709110867e-07, + "loss": 1.1605, + "mean_token_accuracy": 0.6647282838821411, + "num_tokens": 14190058.0, + "step": 555 + }, + { + "epoch": 0.061058642653195695, + "grad_norm": 3.069627285003662, + "learning_rate": 3.046103183315038e-07, + "loss": 1.1545, + "mean_token_accuracy": 0.671165943145752, + "num_tokens": 14210043.0, + "step": 556 + }, + { + "epoch": 0.061168460355809355, + "grad_norm": 2.746685266494751, + "learning_rate": 3.0515916575192094e-07, + "loss": 1.1197, + "mean_token_accuracy": 0.6777287721633911, + "num_tokens": 14233000.0, + "step": 557 + }, + { + "epoch": 0.061278278058423015, + "grad_norm": 2.51068377494812, + "learning_rate": 3.0570801317233806e-07, + "loss": 1.1425, + "mean_token_accuracy": 0.671364963054657, + "num_tokens": 14261210.0, + "step": 558 + }, + { + "epoch": 0.06138809576103668, + "grad_norm": 2.3953969478607178, + "learning_rate": 3.062568605927552e-07, + "loss": 1.0603, + "mean_token_accuracy": 0.688676118850708, + "num_tokens": 14288776.0, + "step": 559 + }, + { + "epoch": 0.06149791346365034, + "grad_norm": 2.883497953414917, + "learning_rate": 3.0680570801317234e-07, + "loss": 1.1246, + "mean_token_accuracy": 0.6711962819099426, + "num_tokens": 14313201.0, + "step": 560 + }, + { + "epoch": 0.061607731166264, + "grad_norm": 2.627168893814087, + "learning_rate": 3.0735455543358946e-07, + "loss": 1.2007, + "mean_token_accuracy": 0.6667066812515259, + "num_tokens": 14338096.0, + "step": 561 + }, + { + "epoch": 0.06171754886887766, + "grad_norm": 2.120388984680176, + "learning_rate": 3.0790340285400657e-07, + "loss": 1.0956, + "mean_token_accuracy": 0.6816165447235107, + "num_tokens": 14374924.0, + "step": 562 + }, + { + "epoch": 0.06182736657149132, + "grad_norm": 2.870120048522949, + "learning_rate": 3.084522502744237e-07, + "loss": 1.092, + "mean_token_accuracy": 0.6792371273040771, + "num_tokens": 14398155.0, + "step": 563 + }, + { + "epoch": 0.06193718427410499, + "grad_norm": 2.2712764739990234, + "learning_rate": 3.0900109769484086e-07, + "loss": 1.2434, + "mean_token_accuracy": 0.641977071762085, + "num_tokens": 14433356.0, + "step": 564 + }, + { + "epoch": 0.06204700197671865, + "grad_norm": 2.7212700843811035, + "learning_rate": 3.0954994511525797e-07, + "loss": 1.0717, + "mean_token_accuracy": 0.6786229014396667, + "num_tokens": 14458821.0, + "step": 565 + }, + { + "epoch": 0.06215681967933231, + "grad_norm": 2.9363479614257812, + "learning_rate": 3.1009879253567503e-07, + "loss": 1.1527, + "mean_token_accuracy": 0.6700756549835205, + "num_tokens": 14479854.0, + "step": 566 + }, + { + "epoch": 0.06226663738194597, + "grad_norm": 2.2048227787017822, + "learning_rate": 3.1064763995609215e-07, + "loss": 1.1621, + "mean_token_accuracy": 0.6684665083885193, + "num_tokens": 14515721.0, + "step": 567 + }, + { + "epoch": 0.062376455084559634, + "grad_norm": 2.2660446166992188, + "learning_rate": 3.111964873765093e-07, + "loss": 1.1542, + "mean_token_accuracy": 0.6647942662239075, + "num_tokens": 14546052.0, + "step": 568 + }, + { + "epoch": 0.062486272787173294, + "grad_norm": 2.695948362350464, + "learning_rate": 3.1174533479692643e-07, + "loss": 1.0997, + "mean_token_accuracy": 0.6814300417900085, + "num_tokens": 14568887.0, + "step": 569 + }, + { + "epoch": 0.06259609048978695, + "grad_norm": 2.211932420730591, + "learning_rate": 3.1229418221734355e-07, + "loss": 1.2187, + "mean_token_accuracy": 0.6524953842163086, + "num_tokens": 14603289.0, + "step": 570 + }, + { + "epoch": 0.06270590819240061, + "grad_norm": 3.1615514755249023, + "learning_rate": 3.128430296377607e-07, + "loss": 1.0315, + "mean_token_accuracy": 0.6969609260559082, + "num_tokens": 14623198.0, + "step": 571 + }, + { + "epoch": 0.06281572589501427, + "grad_norm": 2.7230160236358643, + "learning_rate": 3.1339187705817783e-07, + "loss": 1.1547, + "mean_token_accuracy": 0.6540288329124451, + "num_tokens": 14648915.0, + "step": 572 + }, + { + "epoch": 0.06292554359762793, + "grad_norm": 2.909539222717285, + "learning_rate": 3.1394072447859495e-07, + "loss": 1.0738, + "mean_token_accuracy": 0.6816576719284058, + "num_tokens": 14670929.0, + "step": 573 + }, + { + "epoch": 0.0630353613002416, + "grad_norm": 2.6637351512908936, + "learning_rate": 3.1448957189901206e-07, + "loss": 1.1335, + "mean_token_accuracy": 0.6727254986763, + "num_tokens": 14698420.0, + "step": 574 + }, + { + "epoch": 0.06314517900285525, + "grad_norm": 2.4837005138397217, + "learning_rate": 3.1503841931942923e-07, + "loss": 1.1342, + "mean_token_accuracy": 0.6695652008056641, + "num_tokens": 14727390.0, + "step": 575 + }, + { + "epoch": 0.06325499670546893, + "grad_norm": 2.931217908859253, + "learning_rate": 3.155872667398463e-07, + "loss": 1.095, + "mean_token_accuracy": 0.6780778765678406, + "num_tokens": 14748544.0, + "step": 576 + }, + { + "epoch": 0.06336481440808259, + "grad_norm": 2.0890023708343506, + "learning_rate": 3.161361141602634e-07, + "loss": 1.2998, + "mean_token_accuracy": 0.6274779438972473, + "num_tokens": 14785061.0, + "step": 577 + }, + { + "epoch": 0.06347463211069625, + "grad_norm": 2.461941719055176, + "learning_rate": 3.166849615806805e-07, + "loss": 1.1258, + "mean_token_accuracy": 0.6718968749046326, + "num_tokens": 14813645.0, + "step": 578 + }, + { + "epoch": 0.0635844498133099, + "grad_norm": 2.6198012828826904, + "learning_rate": 3.172338090010977e-07, + "loss": 1.1828, + "mean_token_accuracy": 0.6620968580245972, + "num_tokens": 14841218.0, + "step": 579 + }, + { + "epoch": 0.06369426751592357, + "grad_norm": 2.7523937225341797, + "learning_rate": 3.177826564215148e-07, + "loss": 1.0945, + "mean_token_accuracy": 0.6813254356384277, + "num_tokens": 14865695.0, + "step": 580 + }, + { + "epoch": 0.06380408521853723, + "grad_norm": 2.828402280807495, + "learning_rate": 3.183315038419319e-07, + "loss": 1.0963, + "mean_token_accuracy": 0.6789636611938477, + "num_tokens": 14888051.0, + "step": 581 + }, + { + "epoch": 0.06391390292115089, + "grad_norm": 2.5534565448760986, + "learning_rate": 3.188803512623491e-07, + "loss": 1.2266, + "mean_token_accuracy": 0.6427608728408813, + "num_tokens": 14914698.0, + "step": 582 + }, + { + "epoch": 0.06402372062376455, + "grad_norm": 2.7864015102386475, + "learning_rate": 3.194291986827662e-07, + "loss": 1.1297, + "mean_token_accuracy": 0.671078622341156, + "num_tokens": 14937160.0, + "step": 583 + }, + { + "epoch": 0.0641335383263782, + "grad_norm": 3.01108717918396, + "learning_rate": 3.199780461031833e-07, + "loss": 1.1663, + "mean_token_accuracy": 0.6798018217086792, + "num_tokens": 14958287.0, + "step": 584 + }, + { + "epoch": 0.06424335602899188, + "grad_norm": 2.5995287895202637, + "learning_rate": 3.205268935236004e-07, + "loss": 1.214, + "mean_token_accuracy": 0.6443029642105103, + "num_tokens": 14985194.0, + "step": 585 + }, + { + "epoch": 0.06435317373160554, + "grad_norm": 2.9659206867218018, + "learning_rate": 3.2107574094401755e-07, + "loss": 1.1657, + "mean_token_accuracy": 0.6665798425674438, + "num_tokens": 15005115.0, + "step": 586 + }, + { + "epoch": 0.0644629914342192, + "grad_norm": 2.4473628997802734, + "learning_rate": 3.2162458836443467e-07, + "loss": 1.0302, + "mean_token_accuracy": 0.7023302316665649, + "num_tokens": 15031971.0, + "step": 587 + }, + { + "epoch": 0.06457280913683286, + "grad_norm": 2.4602677822113037, + "learning_rate": 3.221734357848518e-07, + "loss": 1.1906, + "mean_token_accuracy": 0.6643208265304565, + "num_tokens": 15059917.0, + "step": 588 + }, + { + "epoch": 0.06468262683944652, + "grad_norm": 2.617886543273926, + "learning_rate": 3.227222832052689e-07, + "loss": 1.1263, + "mean_token_accuracy": 0.6750026345252991, + "num_tokens": 15086114.0, + "step": 589 + }, + { + "epoch": 0.06479244454206018, + "grad_norm": 2.2159318923950195, + "learning_rate": 3.2327113062568606e-07, + "loss": 1.0864, + "mean_token_accuracy": 0.692084789276123, + "num_tokens": 15116725.0, + "step": 590 + }, + { + "epoch": 0.06490226224467384, + "grad_norm": 2.864705801010132, + "learning_rate": 3.238199780461032e-07, + "loss": 1.0895, + "mean_token_accuracy": 0.6865394115447998, + "num_tokens": 15137555.0, + "step": 591 + }, + { + "epoch": 0.0650120799472875, + "grad_norm": 2.4366776943206787, + "learning_rate": 3.243688254665203e-07, + "loss": 1.1067, + "mean_token_accuracy": 0.6775074005126953, + "num_tokens": 15164902.0, + "step": 592 + }, + { + "epoch": 0.06512189764990116, + "grad_norm": 3.023458242416382, + "learning_rate": 3.2491767288693746e-07, + "loss": 0.9215, + "mean_token_accuracy": 0.7213287353515625, + "num_tokens": 15182444.0, + "step": 593 + }, + { + "epoch": 0.06523171535251482, + "grad_norm": 2.501758098602295, + "learning_rate": 3.254665203073546e-07, + "loss": 1.0009, + "mean_token_accuracy": 0.7041459083557129, + "num_tokens": 15207565.0, + "step": 594 + }, + { + "epoch": 0.06534153305512849, + "grad_norm": 2.6970229148864746, + "learning_rate": 3.2601536772777164e-07, + "loss": 1.1117, + "mean_token_accuracy": 0.679779052734375, + "num_tokens": 15230800.0, + "step": 595 + }, + { + "epoch": 0.06545135075774215, + "grad_norm": 2.3112401962280273, + "learning_rate": 3.2656421514818876e-07, + "loss": 1.0698, + "mean_token_accuracy": 0.6913890838623047, + "num_tokens": 15262809.0, + "step": 596 + }, + { + "epoch": 0.06556116846035581, + "grad_norm": 2.405503749847412, + "learning_rate": 3.271130625686059e-07, + "loss": 1.1379, + "mean_token_accuracy": 0.6728039383888245, + "num_tokens": 15289939.0, + "step": 597 + }, + { + "epoch": 0.06567098616296947, + "grad_norm": 2.3376941680908203, + "learning_rate": 3.2766190998902304e-07, + "loss": 1.1114, + "mean_token_accuracy": 0.6780323386192322, + "num_tokens": 15320702.0, + "step": 598 + }, + { + "epoch": 0.06578080386558313, + "grad_norm": 2.755075216293335, + "learning_rate": 3.2821075740944015e-07, + "loss": 1.1375, + "mean_token_accuracy": 0.6749676465988159, + "num_tokens": 15342846.0, + "step": 599 + }, + { + "epoch": 0.06589062156819679, + "grad_norm": 2.705010414123535, + "learning_rate": 3.2875960482985727e-07, + "loss": 1.095, + "mean_token_accuracy": 0.6806840300559998, + "num_tokens": 15364802.0, + "step": 600 + }, + { + "epoch": 0.06600043927081045, + "grad_norm": 2.470677375793457, + "learning_rate": 3.2930845225027444e-07, + "loss": 1.2092, + "mean_token_accuracy": 0.6552208662033081, + "num_tokens": 15393356.0, + "step": 601 + }, + { + "epoch": 0.06611025697342411, + "grad_norm": 2.471034049987793, + "learning_rate": 3.2985729967069155e-07, + "loss": 1.0355, + "mean_token_accuracy": 0.6977280974388123, + "num_tokens": 15418267.0, + "step": 602 + }, + { + "epoch": 0.06622007467603777, + "grad_norm": 2.5970916748046875, + "learning_rate": 3.3040614709110867e-07, + "loss": 1.1373, + "mean_token_accuracy": 0.6712170243263245, + "num_tokens": 15442023.0, + "step": 603 + }, + { + "epoch": 0.06632989237865144, + "grad_norm": 2.208739757537842, + "learning_rate": 3.3095499451152573e-07, + "loss": 1.1161, + "mean_token_accuracy": 0.6750038266181946, + "num_tokens": 15474883.0, + "step": 604 + }, + { + "epoch": 0.0664397100812651, + "grad_norm": 2.886747360229492, + "learning_rate": 3.315038419319429e-07, + "loss": 1.0827, + "mean_token_accuracy": 0.676034688949585, + "num_tokens": 15493861.0, + "step": 605 + }, + { + "epoch": 0.06654952778387876, + "grad_norm": 2.5146892070770264, + "learning_rate": 3.3205268935236e-07, + "loss": 1.0483, + "mean_token_accuracy": 0.6978753209114075, + "num_tokens": 15517930.0, + "step": 606 + }, + { + "epoch": 0.06665934548649242, + "grad_norm": 2.222362995147705, + "learning_rate": 3.3260153677277713e-07, + "loss": 1.1713, + "mean_token_accuracy": 0.6547473669052124, + "num_tokens": 15548047.0, + "step": 607 + }, + { + "epoch": 0.06676916318910608, + "grad_norm": 2.762101411819458, + "learning_rate": 3.331503841931943e-07, + "loss": 1.1839, + "mean_token_accuracy": 0.6602906584739685, + "num_tokens": 15569970.0, + "step": 608 + }, + { + "epoch": 0.06687898089171974, + "grad_norm": 2.526479959487915, + "learning_rate": 3.336992316136114e-07, + "loss": 1.0543, + "mean_token_accuracy": 0.6867443323135376, + "num_tokens": 15594296.0, + "step": 609 + }, + { + "epoch": 0.0669887985943334, + "grad_norm": 2.2375354766845703, + "learning_rate": 3.3424807903402853e-07, + "loss": 1.1334, + "mean_token_accuracy": 0.6651285290718079, + "num_tokens": 15625494.0, + "step": 610 + }, + { + "epoch": 0.06709861629694706, + "grad_norm": 2.5597779750823975, + "learning_rate": 3.3479692645444564e-07, + "loss": 1.0616, + "mean_token_accuracy": 0.6953325271606445, + "num_tokens": 15651483.0, + "step": 611 + }, + { + "epoch": 0.06720843399956072, + "grad_norm": 2.644479513168335, + "learning_rate": 3.353457738748628e-07, + "loss": 1.0468, + "mean_token_accuracy": 0.6915411949157715, + "num_tokens": 15674285.0, + "step": 612 + }, + { + "epoch": 0.06731825170217438, + "grad_norm": 2.6208863258361816, + "learning_rate": 3.3589462129527993e-07, + "loss": 1.1041, + "mean_token_accuracy": 0.6751848459243774, + "num_tokens": 15699162.0, + "step": 613 + }, + { + "epoch": 0.06742806940478806, + "grad_norm": 2.284453868865967, + "learning_rate": 3.36443468715697e-07, + "loss": 1.0853, + "mean_token_accuracy": 0.6934300065040588, + "num_tokens": 15729401.0, + "step": 614 + }, + { + "epoch": 0.06753788710740172, + "grad_norm": 2.974421739578247, + "learning_rate": 3.369923161361141e-07, + "loss": 1.1285, + "mean_token_accuracy": 0.6803876757621765, + "num_tokens": 15749076.0, + "step": 615 + }, + { + "epoch": 0.06764770481001538, + "grad_norm": 2.48643159866333, + "learning_rate": 3.3754116355653127e-07, + "loss": 1.0678, + "mean_token_accuracy": 0.6915161609649658, + "num_tokens": 15772988.0, + "step": 616 + }, + { + "epoch": 0.06775752251262904, + "grad_norm": 2.745469808578491, + "learning_rate": 3.380900109769484e-07, + "loss": 1.1761, + "mean_token_accuracy": 0.6573842763900757, + "num_tokens": 15794756.0, + "step": 617 + }, + { + "epoch": 0.0678673402152427, + "grad_norm": 2.5475921630859375, + "learning_rate": 3.386388583973655e-07, + "loss": 1.0938, + "mean_token_accuracy": 0.673376739025116, + "num_tokens": 15818585.0, + "step": 618 + }, + { + "epoch": 0.06797715791785636, + "grad_norm": 2.199611186981201, + "learning_rate": 3.3918770581778267e-07, + "loss": 1.1221, + "mean_token_accuracy": 0.6664491891860962, + "num_tokens": 15851209.0, + "step": 619 + }, + { + "epoch": 0.06808697562047002, + "grad_norm": 2.568486213684082, + "learning_rate": 3.397365532381998e-07, + "loss": 1.0943, + "mean_token_accuracy": 0.680225133895874, + "num_tokens": 15877976.0, + "step": 620 + }, + { + "epoch": 0.06819679332308368, + "grad_norm": 2.428103446960449, + "learning_rate": 3.402854006586169e-07, + "loss": 1.1794, + "mean_token_accuracy": 0.6516431570053101, + "num_tokens": 15906571.0, + "step": 621 + }, + { + "epoch": 0.06830661102569734, + "grad_norm": 2.6162359714508057, + "learning_rate": 3.40834248079034e-07, + "loss": 1.15, + "mean_token_accuracy": 0.6645765900611877, + "num_tokens": 15930672.0, + "step": 622 + }, + { + "epoch": 0.06841642872831101, + "grad_norm": 2.704446792602539, + "learning_rate": 3.4138309549945113e-07, + "loss": 1.0787, + "mean_token_accuracy": 0.6833314299583435, + "num_tokens": 15953583.0, + "step": 623 + }, + { + "epoch": 0.06852624643092467, + "grad_norm": 2.856313467025757, + "learning_rate": 3.4193194291986825e-07, + "loss": 1.1208, + "mean_token_accuracy": 0.6816681623458862, + "num_tokens": 15975522.0, + "step": 624 + }, + { + "epoch": 0.06863606413353833, + "grad_norm": 2.6461238861083984, + "learning_rate": 3.4248079034028536e-07, + "loss": 1.0252, + "mean_token_accuracy": 0.6900660991668701, + "num_tokens": 15997534.0, + "step": 625 + }, + { + "epoch": 0.06874588183615199, + "grad_norm": 2.5567924976348877, + "learning_rate": 3.430296377607025e-07, + "loss": 1.1127, + "mean_token_accuracy": 0.6779779195785522, + "num_tokens": 16022149.0, + "step": 626 + }, + { + "epoch": 0.06885569953876565, + "grad_norm": 2.5272741317749023, + "learning_rate": 3.4357848518111965e-07, + "loss": 1.0654, + "mean_token_accuracy": 0.688036322593689, + "num_tokens": 16046964.0, + "step": 627 + }, + { + "epoch": 0.06896551724137931, + "grad_norm": 2.7262895107269287, + "learning_rate": 3.4412733260153676e-07, + "loss": 1.1096, + "mean_token_accuracy": 0.6804586052894592, + "num_tokens": 16068675.0, + "step": 628 + }, + { + "epoch": 0.06907533494399297, + "grad_norm": 2.16037917137146, + "learning_rate": 3.446761800219539e-07, + "loss": 1.1011, + "mean_token_accuracy": 0.6883939504623413, + "num_tokens": 16099882.0, + "step": 629 + }, + { + "epoch": 0.06918515264660663, + "grad_norm": 2.888213872909546, + "learning_rate": 3.4522502744237104e-07, + "loss": 1.056, + "mean_token_accuracy": 0.6930402517318726, + "num_tokens": 16120146.0, + "step": 630 + }, + { + "epoch": 0.06929497034922029, + "grad_norm": 2.0785515308380127, + "learning_rate": 3.4577387486278816e-07, + "loss": 1.1176, + "mean_token_accuracy": 0.676007866859436, + "num_tokens": 16154686.0, + "step": 631 + }, + { + "epoch": 0.06940478805183395, + "grad_norm": 2.336667060852051, + "learning_rate": 3.463227222832053e-07, + "loss": 1.096, + "mean_token_accuracy": 0.6761786341667175, + "num_tokens": 16182763.0, + "step": 632 + }, + { + "epoch": 0.06951460575444762, + "grad_norm": 2.398885488510132, + "learning_rate": 3.4687156970362234e-07, + "loss": 1.0385, + "mean_token_accuracy": 0.6928608417510986, + "num_tokens": 16208306.0, + "step": 633 + }, + { + "epoch": 0.06962442345706128, + "grad_norm": 2.9185080528259277, + "learning_rate": 3.474204171240395e-07, + "loss": 1.2433, + "mean_token_accuracy": 0.6605685353279114, + "num_tokens": 16229086.0, + "step": 634 + }, + { + "epoch": 0.06973424115967494, + "grad_norm": 2.668635368347168, + "learning_rate": 3.479692645444566e-07, + "loss": 1.0541, + "mean_token_accuracy": 0.6876848936080933, + "num_tokens": 16253210.0, + "step": 635 + }, + { + "epoch": 0.0698440588622886, + "grad_norm": 2.4912009239196777, + "learning_rate": 3.4851811196487374e-07, + "loss": 1.1848, + "mean_token_accuracy": 0.6599127054214478, + "num_tokens": 16278614.0, + "step": 636 + }, + { + "epoch": 0.06995387656490226, + "grad_norm": 2.827753782272339, + "learning_rate": 3.4906695938529085e-07, + "loss": 1.1785, + "mean_token_accuracy": 0.6580864191055298, + "num_tokens": 16302362.0, + "step": 637 + }, + { + "epoch": 0.07006369426751592, + "grad_norm": 2.2551159858703613, + "learning_rate": 3.49615806805708e-07, + "loss": 1.0587, + "mean_token_accuracy": 0.6911108493804932, + "num_tokens": 16329983.0, + "step": 638 + }, + { + "epoch": 0.07017351197012958, + "grad_norm": 2.8654043674468994, + "learning_rate": 3.5016465422612513e-07, + "loss": 1.0496, + "mean_token_accuracy": 0.6925676465034485, + "num_tokens": 16350894.0, + "step": 639 + }, + { + "epoch": 0.07028332967274324, + "grad_norm": 3.0840229988098145, + "learning_rate": 3.5071350164654225e-07, + "loss": 1.0076, + "mean_token_accuracy": 0.7016417980194092, + "num_tokens": 16370957.0, + "step": 640 + }, + { + "epoch": 0.0703931473753569, + "grad_norm": 2.2424709796905518, + "learning_rate": 3.512623490669594e-07, + "loss": 1.1873, + "mean_token_accuracy": 0.6559809446334839, + "num_tokens": 16400382.0, + "step": 641 + }, + { + "epoch": 0.07050296507797058, + "grad_norm": 2.7112655639648438, + "learning_rate": 3.518111964873765e-07, + "loss": 1.0569, + "mean_token_accuracy": 0.6837412118911743, + "num_tokens": 16421173.0, + "step": 642 + }, + { + "epoch": 0.07061278278058424, + "grad_norm": 2.96142840385437, + "learning_rate": 3.523600439077936e-07, + "loss": 1.0883, + "mean_token_accuracy": 0.6823481321334839, + "num_tokens": 16440549.0, + "step": 643 + }, + { + "epoch": 0.0707226004831979, + "grad_norm": 2.4184184074401855, + "learning_rate": 3.529088913282107e-07, + "loss": 1.1288, + "mean_token_accuracy": 0.6703730225563049, + "num_tokens": 16466137.0, + "step": 644 + }, + { + "epoch": 0.07083241818581155, + "grad_norm": 2.507002830505371, + "learning_rate": 3.534577387486279e-07, + "loss": 1.1195, + "mean_token_accuracy": 0.6751563549041748, + "num_tokens": 16489998.0, + "step": 645 + }, + { + "epoch": 0.07094223588842521, + "grad_norm": 2.266913652420044, + "learning_rate": 3.54006586169045e-07, + "loss": 1.179, + "mean_token_accuracy": 0.6564456224441528, + "num_tokens": 16519662.0, + "step": 646 + }, + { + "epoch": 0.07105205359103887, + "grad_norm": 2.3756446838378906, + "learning_rate": 3.545554335894621e-07, + "loss": 1.204, + "mean_token_accuracy": 0.6593765020370483, + "num_tokens": 16549333.0, + "step": 647 + }, + { + "epoch": 0.07116187129365253, + "grad_norm": 2.7434279918670654, + "learning_rate": 3.551042810098792e-07, + "loss": 1.1729, + "mean_token_accuracy": 0.6692070960998535, + "num_tokens": 16571566.0, + "step": 648 + }, + { + "epoch": 0.0712716889962662, + "grad_norm": 2.4965522289276123, + "learning_rate": 3.556531284302964e-07, + "loss": 1.0191, + "mean_token_accuracy": 0.7012209892272949, + "num_tokens": 16595296.0, + "step": 649 + }, + { + "epoch": 0.07138150669887985, + "grad_norm": 2.508690357208252, + "learning_rate": 3.562019758507135e-07, + "loss": 1.0748, + "mean_token_accuracy": 0.6848982572555542, + "num_tokens": 16620496.0, + "step": 650 + }, + { + "epoch": 0.07149132440149351, + "grad_norm": 2.866421699523926, + "learning_rate": 3.5675082327113057e-07, + "loss": 1.1495, + "mean_token_accuracy": 0.6704912185668945, + "num_tokens": 16639951.0, + "step": 651 + }, + { + "epoch": 0.07160114210410719, + "grad_norm": 2.3979287147521973, + "learning_rate": 3.5729967069154774e-07, + "loss": 1.0592, + "mean_token_accuracy": 0.6863941550254822, + "num_tokens": 16666993.0, + "step": 652 + }, + { + "epoch": 0.07171095980672085, + "grad_norm": 2.311506509780884, + "learning_rate": 3.5784851811196485e-07, + "loss": 1.0949, + "mean_token_accuracy": 0.6791126728057861, + "num_tokens": 16695468.0, + "step": 653 + }, + { + "epoch": 0.07182077750933451, + "grad_norm": 2.6637673377990723, + "learning_rate": 3.5839736553238197e-07, + "loss": 1.119, + "mean_token_accuracy": 0.6749553084373474, + "num_tokens": 16719214.0, + "step": 654 + }, + { + "epoch": 0.07193059521194817, + "grad_norm": 2.8224215507507324, + "learning_rate": 3.589462129527991e-07, + "loss": 1.0271, + "mean_token_accuracy": 0.6928434371948242, + "num_tokens": 16740957.0, + "step": 655 + }, + { + "epoch": 0.07204041291456183, + "grad_norm": 2.5292630195617676, + "learning_rate": 3.5949506037321625e-07, + "loss": 1.0892, + "mean_token_accuracy": 0.6820632219314575, + "num_tokens": 16766646.0, + "step": 656 + }, + { + "epoch": 0.07215023061717549, + "grad_norm": 2.6541619300842285, + "learning_rate": 3.6004390779363337e-07, + "loss": 1.1865, + "mean_token_accuracy": 0.6551270484924316, + "num_tokens": 16790286.0, + "step": 657 + }, + { + "epoch": 0.07226004831978915, + "grad_norm": 2.2646615505218506, + "learning_rate": 3.605927552140505e-07, + "loss": 1.1688, + "mean_token_accuracy": 0.6640782356262207, + "num_tokens": 16820146.0, + "step": 658 + }, + { + "epoch": 0.0723698660224028, + "grad_norm": 2.735342264175415, + "learning_rate": 3.611416026344676e-07, + "loss": 1.1309, + "mean_token_accuracy": 0.668781042098999, + "num_tokens": 16842170.0, + "step": 659 + }, + { + "epoch": 0.07247968372501647, + "grad_norm": 2.4466969966888428, + "learning_rate": 3.6169045005488477e-07, + "loss": 1.101, + "mean_token_accuracy": 0.6831360459327698, + "num_tokens": 16867071.0, + "step": 660 + }, + { + "epoch": 0.07258950142763014, + "grad_norm": 2.3566300868988037, + "learning_rate": 3.6223929747530183e-07, + "loss": 1.0968, + "mean_token_accuracy": 0.6861728429794312, + "num_tokens": 16897159.0, + "step": 661 + }, + { + "epoch": 0.0726993191302438, + "grad_norm": 2.3779854774475098, + "learning_rate": 3.6278814489571894e-07, + "loss": 1.1187, + "mean_token_accuracy": 0.6738275289535522, + "num_tokens": 16924315.0, + "step": 662 + }, + { + "epoch": 0.07280913683285746, + "grad_norm": 2.3956289291381836, + "learning_rate": 3.6333699231613606e-07, + "loss": 1.1823, + "mean_token_accuracy": 0.6539854407310486, + "num_tokens": 16952973.0, + "step": 663 + }, + { + "epoch": 0.07291895453547112, + "grad_norm": 2.419362783432007, + "learning_rate": 3.6388583973655323e-07, + "loss": 1.1553, + "mean_token_accuracy": 0.6631338000297546, + "num_tokens": 16980377.0, + "step": 664 + }, + { + "epoch": 0.07302877223808478, + "grad_norm": 2.494213104248047, + "learning_rate": 3.6443468715697034e-07, + "loss": 1.095, + "mean_token_accuracy": 0.6798760890960693, + "num_tokens": 17007175.0, + "step": 665 + }, + { + "epoch": 0.07313858994069844, + "grad_norm": 2.308943271636963, + "learning_rate": 3.6498353457738746e-07, + "loss": 1.0916, + "mean_token_accuracy": 0.6787681579589844, + "num_tokens": 17033925.0, + "step": 666 + }, + { + "epoch": 0.0732484076433121, + "grad_norm": 2.506861925125122, + "learning_rate": 3.655323819978046e-07, + "loss": 1.0657, + "mean_token_accuracy": 0.6843278408050537, + "num_tokens": 17057857.0, + "step": 667 + }, + { + "epoch": 0.07335822534592576, + "grad_norm": 2.648054838180542, + "learning_rate": 3.6608122941822174e-07, + "loss": 1.1765, + "mean_token_accuracy": 0.6685245633125305, + "num_tokens": 17081588.0, + "step": 668 + }, + { + "epoch": 0.07346804304853942, + "grad_norm": 2.7125630378723145, + "learning_rate": 3.6663007683863886e-07, + "loss": 1.0011, + "mean_token_accuracy": 0.7029314041137695, + "num_tokens": 17101390.0, + "step": 669 + }, + { + "epoch": 0.07357786075115308, + "grad_norm": 2.4115002155303955, + "learning_rate": 3.671789242590559e-07, + "loss": 1.1158, + "mean_token_accuracy": 0.6772300601005554, + "num_tokens": 17126884.0, + "step": 670 + }, + { + "epoch": 0.07368767845376675, + "grad_norm": 2.251235246658325, + "learning_rate": 3.677277716794731e-07, + "loss": 1.1067, + "mean_token_accuracy": 0.6822038292884827, + "num_tokens": 17155385.0, + "step": 671 + }, + { + "epoch": 0.07379749615638041, + "grad_norm": 2.458993434906006, + "learning_rate": 3.682766190998902e-07, + "loss": 1.1306, + "mean_token_accuracy": 0.6734921932220459, + "num_tokens": 17181508.0, + "step": 672 + }, + { + "epoch": 0.07390731385899407, + "grad_norm": 2.5588293075561523, + "learning_rate": 3.688254665203073e-07, + "loss": 1.0079, + "mean_token_accuracy": 0.7031605243682861, + "num_tokens": 17205246.0, + "step": 673 + }, + { + "epoch": 0.07401713156160773, + "grad_norm": 2.343190908432007, + "learning_rate": 3.6937431394072443e-07, + "loss": 1.153, + "mean_token_accuracy": 0.6605772376060486, + "num_tokens": 17232180.0, + "step": 674 + }, + { + "epoch": 0.07412694926422139, + "grad_norm": 2.2822985649108887, + "learning_rate": 3.699231613611416e-07, + "loss": 1.1368, + "mean_token_accuracy": 0.6683691143989563, + "num_tokens": 17263733.0, + "step": 675 + }, + { + "epoch": 0.07423676696683505, + "grad_norm": 2.7353737354278564, + "learning_rate": 3.704720087815587e-07, + "loss": 1.0095, + "mean_token_accuracy": 0.703224778175354, + "num_tokens": 17284437.0, + "step": 676 + }, + { + "epoch": 0.07434658466944871, + "grad_norm": 2.804532051086426, + "learning_rate": 3.7102085620197583e-07, + "loss": 1.1467, + "mean_token_accuracy": 0.6593762636184692, + "num_tokens": 17305757.0, + "step": 677 + }, + { + "epoch": 0.07445640237206237, + "grad_norm": 3.203603506088257, + "learning_rate": 3.71569703622393e-07, + "loss": 1.0948, + "mean_token_accuracy": 0.6764190196990967, + "num_tokens": 17325911.0, + "step": 678 + }, + { + "epoch": 0.07456622007467603, + "grad_norm": 2.419711112976074, + "learning_rate": 3.721185510428101e-07, + "loss": 1.0773, + "mean_token_accuracy": 0.6768589019775391, + "num_tokens": 17351111.0, + "step": 679 + }, + { + "epoch": 0.0746760377772897, + "grad_norm": 2.6770262718200684, + "learning_rate": 3.726673984632272e-07, + "loss": 1.0454, + "mean_token_accuracy": 0.7022207975387573, + "num_tokens": 17373600.0, + "step": 680 + }, + { + "epoch": 0.07478585547990337, + "grad_norm": 2.487008810043335, + "learning_rate": 3.732162458836443e-07, + "loss": 1.1216, + "mean_token_accuracy": 0.6697163581848145, + "num_tokens": 17398543.0, + "step": 681 + }, + { + "epoch": 0.07489567318251703, + "grad_norm": 2.399381637573242, + "learning_rate": 3.7376509330406146e-07, + "loss": 1.1899, + "mean_token_accuracy": 0.6538713574409485, + "num_tokens": 17426285.0, + "step": 682 + }, + { + "epoch": 0.07500549088513069, + "grad_norm": 2.546989917755127, + "learning_rate": 3.743139407244786e-07, + "loss": 1.0776, + "mean_token_accuracy": 0.6819491386413574, + "num_tokens": 17451029.0, + "step": 683 + }, + { + "epoch": 0.07511530858774434, + "grad_norm": 2.446197509765625, + "learning_rate": 3.748627881448957e-07, + "loss": 1.1056, + "mean_token_accuracy": 0.6706414818763733, + "num_tokens": 17477668.0, + "step": 684 + }, + { + "epoch": 0.075225126290358, + "grad_norm": 2.551422119140625, + "learning_rate": 3.754116355653128e-07, + "loss": 1.0633, + "mean_token_accuracy": 0.683895468711853, + "num_tokens": 17501606.0, + "step": 685 + }, + { + "epoch": 0.07533494399297166, + "grad_norm": 2.444143056869507, + "learning_rate": 3.7596048298573e-07, + "loss": 1.2328, + "mean_token_accuracy": 0.6466124653816223, + "num_tokens": 17528715.0, + "step": 686 + }, + { + "epoch": 0.07544476169558532, + "grad_norm": 2.5962324142456055, + "learning_rate": 3.765093304061471e-07, + "loss": 1.1406, + "mean_token_accuracy": 0.6730235815048218, + "num_tokens": 17552576.0, + "step": 687 + }, + { + "epoch": 0.07555457939819898, + "grad_norm": 2.309121608734131, + "learning_rate": 3.770581778265642e-07, + "loss": 1.0308, + "mean_token_accuracy": 0.6977751851081848, + "num_tokens": 17578880.0, + "step": 688 + }, + { + "epoch": 0.07566439710081264, + "grad_norm": 2.5204854011535645, + "learning_rate": 3.7760702524698137e-07, + "loss": 1.0335, + "mean_token_accuracy": 0.6922653913497925, + "num_tokens": 17603046.0, + "step": 689 + }, + { + "epoch": 0.07577421480342632, + "grad_norm": 2.221385955810547, + "learning_rate": 3.7815587266739844e-07, + "loss": 1.1031, + "mean_token_accuracy": 0.6832256317138672, + "num_tokens": 17631749.0, + "step": 690 + }, + { + "epoch": 0.07588403250603998, + "grad_norm": 2.534353733062744, + "learning_rate": 3.7870472008781555e-07, + "loss": 1.1267, + "mean_token_accuracy": 0.6705412864685059, + "num_tokens": 17656406.0, + "step": 691 + }, + { + "epoch": 0.07599385020865364, + "grad_norm": 2.5562610626220703, + "learning_rate": 3.7925356750823267e-07, + "loss": 1.0881, + "mean_token_accuracy": 0.6769446134567261, + "num_tokens": 17678193.0, + "step": 692 + }, + { + "epoch": 0.0761036679112673, + "grad_norm": 2.075159788131714, + "learning_rate": 3.7980241492864983e-07, + "loss": 1.1963, + "mean_token_accuracy": 0.6532615423202515, + "num_tokens": 17714703.0, + "step": 693 + }, + { + "epoch": 0.07621348561388096, + "grad_norm": 2.3652822971343994, + "learning_rate": 3.8035126234906695e-07, + "loss": 1.0826, + "mean_token_accuracy": 0.6849661469459534, + "num_tokens": 17744672.0, + "step": 694 + }, + { + "epoch": 0.07632330331649462, + "grad_norm": 3.128051280975342, + "learning_rate": 3.8090010976948406e-07, + "loss": 0.9839, + "mean_token_accuracy": 0.6992751955986023, + "num_tokens": 17762339.0, + "step": 695 + }, + { + "epoch": 0.07643312101910828, + "grad_norm": 2.368191957473755, + "learning_rate": 3.814489571899012e-07, + "loss": 1.0998, + "mean_token_accuracy": 0.67503821849823, + "num_tokens": 17789075.0, + "step": 696 + }, + { + "epoch": 0.07654293872172194, + "grad_norm": 2.55304217338562, + "learning_rate": 3.8199780461031835e-07, + "loss": 1.1185, + "mean_token_accuracy": 0.675503134727478, + "num_tokens": 17812332.0, + "step": 697 + }, + { + "epoch": 0.0766527564243356, + "grad_norm": 2.538299560546875, + "learning_rate": 3.8254665203073546e-07, + "loss": 1.0999, + "mean_token_accuracy": 0.6750611066818237, + "num_tokens": 17836214.0, + "step": 698 + }, + { + "epoch": 0.07676257412694927, + "grad_norm": 2.288607120513916, + "learning_rate": 3.830954994511525e-07, + "loss": 1.17, + "mean_token_accuracy": 0.6627006530761719, + "num_tokens": 17865525.0, + "step": 699 + }, + { + "epoch": 0.07687239182956293, + "grad_norm": 2.8103585243225098, + "learning_rate": 3.836443468715697e-07, + "loss": 1.1136, + "mean_token_accuracy": 0.6762309670448303, + "num_tokens": 17886614.0, + "step": 700 + }, + { + "epoch": 0.07698220953217659, + "grad_norm": 2.7870397567749023, + "learning_rate": 3.841931942919868e-07, + "loss": 0.9825, + "mean_token_accuracy": 0.7072858810424805, + "num_tokens": 17905655.0, + "step": 701 + }, + { + "epoch": 0.07709202723479025, + "grad_norm": 3.0452544689178467, + "learning_rate": 3.847420417124039e-07, + "loss": 1.0027, + "mean_token_accuracy": 0.6981643438339233, + "num_tokens": 17922596.0, + "step": 702 + }, + { + "epoch": 0.07720184493740391, + "grad_norm": 2.5733444690704346, + "learning_rate": 3.8529088913282104e-07, + "loss": 0.9681, + "mean_token_accuracy": 0.7095773220062256, + "num_tokens": 17944003.0, + "step": 703 + }, + { + "epoch": 0.07731166264001757, + "grad_norm": 2.977137804031372, + "learning_rate": 3.858397365532382e-07, + "loss": 0.9544, + "mean_token_accuracy": 0.70464026927948, + "num_tokens": 17961528.0, + "step": 704 + }, + { + "epoch": 0.07742148034263123, + "grad_norm": 2.4152157306671143, + "learning_rate": 3.863885839736553e-07, + "loss": 1.1566, + "mean_token_accuracy": 0.658139169216156, + "num_tokens": 17988918.0, + "step": 705 + }, + { + "epoch": 0.07753129804524489, + "grad_norm": 2.507800340652466, + "learning_rate": 3.8693743139407244e-07, + "loss": 1.0744, + "mean_token_accuracy": 0.6873059272766113, + "num_tokens": 18013477.0, + "step": 706 + }, + { + "epoch": 0.07764111574785855, + "grad_norm": 2.2955105304718018, + "learning_rate": 3.8748627881448955e-07, + "loss": 1.1206, + "mean_token_accuracy": 0.6751445531845093, + "num_tokens": 18043792.0, + "step": 707 + }, + { + "epoch": 0.07775093345047221, + "grad_norm": 2.514993667602539, + "learning_rate": 3.880351262349067e-07, + "loss": 1.0248, + "mean_token_accuracy": 0.6955201625823975, + "num_tokens": 18067371.0, + "step": 708 + }, + { + "epoch": 0.07786075115308588, + "grad_norm": 2.57330584526062, + "learning_rate": 3.885839736553238e-07, + "loss": 1.0212, + "mean_token_accuracy": 0.6930540800094604, + "num_tokens": 18089131.0, + "step": 709 + }, + { + "epoch": 0.07797056885569954, + "grad_norm": 2.5712811946868896, + "learning_rate": 3.891328210757409e-07, + "loss": 1.0527, + "mean_token_accuracy": 0.6852920055389404, + "num_tokens": 18113432.0, + "step": 710 + }, + { + "epoch": 0.0780803865583132, + "grad_norm": 2.3074073791503906, + "learning_rate": 3.8968166849615807e-07, + "loss": 1.1199, + "mean_token_accuracy": 0.6752992868423462, + "num_tokens": 18142460.0, + "step": 711 + }, + { + "epoch": 0.07819020426092686, + "grad_norm": 2.187676191329956, + "learning_rate": 3.902305159165752e-07, + "loss": 1.1347, + "mean_token_accuracy": 0.6761799454689026, + "num_tokens": 18173726.0, + "step": 712 + }, + { + "epoch": 0.07830002196354052, + "grad_norm": 2.613863468170166, + "learning_rate": 3.907793633369923e-07, + "loss": 1.0666, + "mean_token_accuracy": 0.6891650557518005, + "num_tokens": 18196172.0, + "step": 713 + }, + { + "epoch": 0.07840983966615418, + "grad_norm": 2.5093235969543457, + "learning_rate": 3.913282107574094e-07, + "loss": 1.0597, + "mean_token_accuracy": 0.6826335191726685, + "num_tokens": 18217197.0, + "step": 714 + }, + { + "epoch": 0.07851965736876784, + "grad_norm": 2.5121209621429443, + "learning_rate": 3.918770581778266e-07, + "loss": 1.0982, + "mean_token_accuracy": 0.6773220300674438, + "num_tokens": 18241498.0, + "step": 715 + }, + { + "epoch": 0.0786294750713815, + "grad_norm": 2.40837025642395, + "learning_rate": 3.924259055982437e-07, + "loss": 1.0395, + "mean_token_accuracy": 0.7179538011550903, + "num_tokens": 18268220.0, + "step": 716 + }, + { + "epoch": 0.07873929277399516, + "grad_norm": 2.597005844116211, + "learning_rate": 3.929747530186608e-07, + "loss": 1.0444, + "mean_token_accuracy": 0.6894104480743408, + "num_tokens": 18290926.0, + "step": 717 + }, + { + "epoch": 0.07884911047660884, + "grad_norm": 2.4049503803253174, + "learning_rate": 3.935236004390779e-07, + "loss": 1.0512, + "mean_token_accuracy": 0.6881304979324341, + "num_tokens": 18318712.0, + "step": 718 + }, + { + "epoch": 0.0789589281792225, + "grad_norm": 2.207051992416382, + "learning_rate": 3.9407244785949504e-07, + "loss": 1.1716, + "mean_token_accuracy": 0.6798619031906128, + "num_tokens": 18349495.0, + "step": 719 + }, + { + "epoch": 0.07906874588183616, + "grad_norm": 2.3311877250671387, + "learning_rate": 3.9462129527991216e-07, + "loss": 1.0497, + "mean_token_accuracy": 0.6848529577255249, + "num_tokens": 18377596.0, + "step": 720 + }, + { + "epoch": 0.07917856358444982, + "grad_norm": 2.3271498680114746, + "learning_rate": 3.9517014270032927e-07, + "loss": 1.0718, + "mean_token_accuracy": 0.6825895309448242, + "num_tokens": 18405080.0, + "step": 721 + }, + { + "epoch": 0.07928838128706348, + "grad_norm": 2.6620259284973145, + "learning_rate": 3.957189901207464e-07, + "loss": 1.1273, + "mean_token_accuracy": 0.6675500869750977, + "num_tokens": 18427435.0, + "step": 722 + }, + { + "epoch": 0.07939819898967714, + "grad_norm": 2.0717344284057617, + "learning_rate": 3.9626783754116356e-07, + "loss": 1.0887, + "mean_token_accuracy": 0.677370011806488, + "num_tokens": 18458423.0, + "step": 723 + }, + { + "epoch": 0.0795080166922908, + "grad_norm": 2.391000270843506, + "learning_rate": 3.9681668496158067e-07, + "loss": 1.0642, + "mean_token_accuracy": 0.682226300239563, + "num_tokens": 18483547.0, + "step": 724 + }, + { + "epoch": 0.07961783439490445, + "grad_norm": 2.2791731357574463, + "learning_rate": 3.973655323819978e-07, + "loss": 1.0564, + "mean_token_accuracy": 0.7007995843887329, + "num_tokens": 18512386.0, + "step": 725 + }, + { + "epoch": 0.07972765209751811, + "grad_norm": 2.4180479049682617, + "learning_rate": 3.9791437980241495e-07, + "loss": 1.0063, + "mean_token_accuracy": 0.6967347860336304, + "num_tokens": 18536490.0, + "step": 726 + }, + { + "epoch": 0.07983746980013177, + "grad_norm": 2.1878302097320557, + "learning_rate": 3.9846322722283207e-07, + "loss": 1.0519, + "mean_token_accuracy": 0.6900230646133423, + "num_tokens": 18569008.0, + "step": 727 + }, + { + "epoch": 0.07994728750274545, + "grad_norm": 2.343325138092041, + "learning_rate": 3.9901207464324913e-07, + "loss": 1.0296, + "mean_token_accuracy": 0.6993945837020874, + "num_tokens": 18596596.0, + "step": 728 + }, + { + "epoch": 0.08005710520535911, + "grad_norm": 2.6582224369049072, + "learning_rate": 3.9956092206366625e-07, + "loss": 1.119, + "mean_token_accuracy": 0.6718021631240845, + "num_tokens": 18622172.0, + "step": 729 + }, + { + "epoch": 0.08016692290797277, + "grad_norm": 3.2057764530181885, + "learning_rate": 4.001097694840834e-07, + "loss": 1.1575, + "mean_token_accuracy": 0.6846916675567627, + "num_tokens": 18640264.0, + "step": 730 + }, + { + "epoch": 0.08027674061058643, + "grad_norm": 3.3804850578308105, + "learning_rate": 4.0065861690450053e-07, + "loss": 1.0821, + "mean_token_accuracy": 0.6825416088104248, + "num_tokens": 18666928.0, + "step": 731 + }, + { + "epoch": 0.08038655831320009, + "grad_norm": 2.3053441047668457, + "learning_rate": 4.0120746432491765e-07, + "loss": 1.2165, + "mean_token_accuracy": 0.6484824419021606, + "num_tokens": 18694976.0, + "step": 732 + }, + { + "epoch": 0.08049637601581375, + "grad_norm": 2.3906409740448, + "learning_rate": 4.0175631174533476e-07, + "loss": 1.0556, + "mean_token_accuracy": 0.6812582015991211, + "num_tokens": 18724309.0, + "step": 733 + }, + { + "epoch": 0.08060619371842741, + "grad_norm": 2.288206100463867, + "learning_rate": 4.0230515916575193e-07, + "loss": 1.1416, + "mean_token_accuracy": 0.6723031997680664, + "num_tokens": 18750734.0, + "step": 734 + }, + { + "epoch": 0.08071601142104107, + "grad_norm": 2.1361613273620605, + "learning_rate": 4.0285400658616904e-07, + "loss": 1.1027, + "mean_token_accuracy": 0.6903146505355835, + "num_tokens": 18783441.0, + "step": 735 + }, + { + "epoch": 0.08082582912365473, + "grad_norm": 2.709289312362671, + "learning_rate": 4.0340285400658616e-07, + "loss": 1.0826, + "mean_token_accuracy": 0.6796688437461853, + "num_tokens": 18804729.0, + "step": 736 + }, + { + "epoch": 0.0809356468262684, + "grad_norm": 2.6287710666656494, + "learning_rate": 4.039517014270033e-07, + "loss": 0.9745, + "mean_token_accuracy": 0.7103304862976074, + "num_tokens": 18826765.0, + "step": 737 + }, + { + "epoch": 0.08104546452888206, + "grad_norm": 2.7334325313568115, + "learning_rate": 4.045005488474204e-07, + "loss": 1.1306, + "mean_token_accuracy": 0.6728687286376953, + "num_tokens": 18847238.0, + "step": 738 + }, + { + "epoch": 0.08115528223149572, + "grad_norm": 2.4154419898986816, + "learning_rate": 4.050493962678375e-07, + "loss": 1.0766, + "mean_token_accuracy": 0.6780332326889038, + "num_tokens": 18872434.0, + "step": 739 + }, + { + "epoch": 0.08126509993410938, + "grad_norm": 2.361297369003296, + "learning_rate": 4.055982436882546e-07, + "loss": 0.9703, + "mean_token_accuracy": 0.7133012413978577, + "num_tokens": 18896603.0, + "step": 740 + }, + { + "epoch": 0.08137491763672304, + "grad_norm": 2.221271276473999, + "learning_rate": 4.061470911086718e-07, + "loss": 1.0112, + "mean_token_accuracy": 0.6990737915039062, + "num_tokens": 18923751.0, + "step": 741 + }, + { + "epoch": 0.0814847353393367, + "grad_norm": 2.448230743408203, + "learning_rate": 4.066959385290889e-07, + "loss": 1.1191, + "mean_token_accuracy": 0.6725143194198608, + "num_tokens": 18948398.0, + "step": 742 + }, + { + "epoch": 0.08159455304195036, + "grad_norm": 2.238598346710205, + "learning_rate": 4.07244785949506e-07, + "loss": 1.1237, + "mean_token_accuracy": 0.6750310659408569, + "num_tokens": 18977399.0, + "step": 743 + }, + { + "epoch": 0.08170437074456402, + "grad_norm": 2.2375638484954834, + "learning_rate": 4.0779363336992313e-07, + "loss": 1.0754, + "mean_token_accuracy": 0.6816981434822083, + "num_tokens": 19004906.0, + "step": 744 + }, + { + "epoch": 0.08181418844717768, + "grad_norm": 2.3054301738739014, + "learning_rate": 4.083424807903403e-07, + "loss": 1.1118, + "mean_token_accuracy": 0.6741310954093933, + "num_tokens": 19033431.0, + "step": 745 + }, + { + "epoch": 0.08192400614979134, + "grad_norm": 2.2800633907318115, + "learning_rate": 4.0889132821075737e-07, + "loss": 0.9822, + "mean_token_accuracy": 0.7097031474113464, + "num_tokens": 19063074.0, + "step": 746 + }, + { + "epoch": 0.08203382385240501, + "grad_norm": 2.1202778816223145, + "learning_rate": 4.094401756311745e-07, + "loss": 1.0122, + "mean_token_accuracy": 0.6976656913757324, + "num_tokens": 19093483.0, + "step": 747 + }, + { + "epoch": 0.08214364155501867, + "grad_norm": 2.247711181640625, + "learning_rate": 4.0998902305159165e-07, + "loss": 1.0751, + "mean_token_accuracy": 0.6824800968170166, + "num_tokens": 19121922.0, + "step": 748 + }, + { + "epoch": 0.08225345925763233, + "grad_norm": 2.3491082191467285, + "learning_rate": 4.1053787047200876e-07, + "loss": 1.2003, + "mean_token_accuracy": 0.6513763666152954, + "num_tokens": 19150503.0, + "step": 749 + }, + { + "epoch": 0.08236327696024599, + "grad_norm": 2.7021851539611816, + "learning_rate": 4.110867178924259e-07, + "loss": 1.1168, + "mean_token_accuracy": 0.6752758026123047, + "num_tokens": 19171772.0, + "step": 750 + }, + { + "epoch": 0.08247309466285965, + "grad_norm": 2.139751672744751, + "learning_rate": 4.11635565312843e-07, + "loss": 1.0529, + "mean_token_accuracy": 0.6894055008888245, + "num_tokens": 19203123.0, + "step": 751 + }, + { + "epoch": 0.08258291236547331, + "grad_norm": 2.111232042312622, + "learning_rate": 4.1218441273326016e-07, + "loss": 1.1525, + "mean_token_accuracy": 0.6668035984039307, + "num_tokens": 19233877.0, + "step": 752 + }, + { + "epoch": 0.08269273006808697, + "grad_norm": 2.4042797088623047, + "learning_rate": 4.127332601536773e-07, + "loss": 0.987, + "mean_token_accuracy": 0.7077190279960632, + "num_tokens": 19258018.0, + "step": 753 + }, + { + "epoch": 0.08280254777070063, + "grad_norm": 2.399552345275879, + "learning_rate": 4.132821075740944e-07, + "loss": 1.1262, + "mean_token_accuracy": 0.6785398721694946, + "num_tokens": 19285429.0, + "step": 754 + }, + { + "epoch": 0.08291236547331429, + "grad_norm": 2.5871357917785645, + "learning_rate": 4.138309549945115e-07, + "loss": 1.085, + "mean_token_accuracy": 0.6833242177963257, + "num_tokens": 19309331.0, + "step": 755 + }, + { + "epoch": 0.08302218317592797, + "grad_norm": 2.2102549076080322, + "learning_rate": 4.143798024149286e-07, + "loss": 1.1248, + "mean_token_accuracy": 0.6754212379455566, + "num_tokens": 19337778.0, + "step": 756 + }, + { + "epoch": 0.08313200087854163, + "grad_norm": 2.6479296684265137, + "learning_rate": 4.1492864983534574e-07, + "loss": 1.108, + "mean_token_accuracy": 0.6723260879516602, + "num_tokens": 19358859.0, + "step": 757 + }, + { + "epoch": 0.08324181858115529, + "grad_norm": 2.6242315769195557, + "learning_rate": 4.1547749725576285e-07, + "loss": 1.0281, + "mean_token_accuracy": 0.6959081888198853, + "num_tokens": 19379435.0, + "step": 758 + }, + { + "epoch": 0.08335163628376895, + "grad_norm": 2.24951171875, + "learning_rate": 4.1602634467618e-07, + "loss": 1.0304, + "mean_token_accuracy": 0.687073826789856, + "num_tokens": 19405704.0, + "step": 759 + }, + { + "epoch": 0.0834614539863826, + "grad_norm": 2.4709584712982178, + "learning_rate": 4.1657519209659714e-07, + "loss": 0.9777, + "mean_token_accuracy": 0.7068338394165039, + "num_tokens": 19428567.0, + "step": 760 + }, + { + "epoch": 0.08357127168899627, + "grad_norm": 2.5073401927948, + "learning_rate": 4.1712403951701425e-07, + "loss": 1.008, + "mean_token_accuracy": 0.7014617323875427, + "num_tokens": 19452185.0, + "step": 761 + }, + { + "epoch": 0.08368108939160993, + "grad_norm": 2.197082757949829, + "learning_rate": 4.1767288693743137e-07, + "loss": 1.0714, + "mean_token_accuracy": 0.6896169185638428, + "num_tokens": 19483419.0, + "step": 762 + }, + { + "epoch": 0.08379090709422359, + "grad_norm": 2.5001797676086426, + "learning_rate": 4.1822173435784854e-07, + "loss": 1.0909, + "mean_token_accuracy": 0.6867016553878784, + "num_tokens": 19509299.0, + "step": 763 + }, + { + "epoch": 0.08390072479683724, + "grad_norm": 1.909211277961731, + "learning_rate": 4.1877058177826565e-07, + "loss": 1.0824, + "mean_token_accuracy": 0.677528977394104, + "num_tokens": 19544908.0, + "step": 764 + }, + { + "epoch": 0.0840105424994509, + "grad_norm": 2.07645845413208, + "learning_rate": 4.193194291986827e-07, + "loss": 1.1063, + "mean_token_accuracy": 0.677207887172699, + "num_tokens": 19577731.0, + "step": 765 + }, + { + "epoch": 0.08412036020206458, + "grad_norm": 2.489840030670166, + "learning_rate": 4.1986827661909983e-07, + "loss": 1.0587, + "mean_token_accuracy": 0.6821370124816895, + "num_tokens": 19602492.0, + "step": 766 + }, + { + "epoch": 0.08423017790467824, + "grad_norm": 2.26814341545105, + "learning_rate": 4.20417124039517e-07, + "loss": 1.1319, + "mean_token_accuracy": 0.6677074432373047, + "num_tokens": 19629977.0, + "step": 767 + }, + { + "epoch": 0.0843399956072919, + "grad_norm": 2.2160565853118896, + "learning_rate": 4.209659714599341e-07, + "loss": 1.129, + "mean_token_accuracy": 0.6679675579071045, + "num_tokens": 19659646.0, + "step": 768 + }, + { + "epoch": 0.08444981330990556, + "grad_norm": 2.247687816619873, + "learning_rate": 4.2151481888035123e-07, + "loss": 1.0922, + "mean_token_accuracy": 0.6796954870223999, + "num_tokens": 19687981.0, + "step": 769 + }, + { + "epoch": 0.08455963101251922, + "grad_norm": 2.4960262775421143, + "learning_rate": 4.2206366630076834e-07, + "loss": 1.2083, + "mean_token_accuracy": 0.653243899345398, + "num_tokens": 19714096.0, + "step": 770 + }, + { + "epoch": 0.08466944871513288, + "grad_norm": 2.2494137287139893, + "learning_rate": 4.226125137211855e-07, + "loss": 1.1639, + "mean_token_accuracy": 0.6704214215278625, + "num_tokens": 19743047.0, + "step": 771 + }, + { + "epoch": 0.08477926641774654, + "grad_norm": 3.0281124114990234, + "learning_rate": 4.231613611416026e-07, + "loss": 1.0195, + "mean_token_accuracy": 0.6940641403198242, + "num_tokens": 19761254.0, + "step": 772 + }, + { + "epoch": 0.0848890841203602, + "grad_norm": 2.1179699897766113, + "learning_rate": 4.2371020856201974e-07, + "loss": 1.0648, + "mean_token_accuracy": 0.6860027313232422, + "num_tokens": 19794107.0, + "step": 773 + }, + { + "epoch": 0.08499890182297386, + "grad_norm": 2.592663288116455, + "learning_rate": 4.242590559824369e-07, + "loss": 1.1974, + "mean_token_accuracy": 0.6561360955238342, + "num_tokens": 19820918.0, + "step": 774 + }, + { + "epoch": 0.08510871952558753, + "grad_norm": 2.2201201915740967, + "learning_rate": 4.2480790340285397e-07, + "loss": 1.1972, + "mean_token_accuracy": 0.6587015390396118, + "num_tokens": 19848887.0, + "step": 775 + }, + { + "epoch": 0.08521853722820119, + "grad_norm": 2.58333158493042, + "learning_rate": 4.253567508232711e-07, + "loss": 1.1446, + "mean_token_accuracy": 0.6687231659889221, + "num_tokens": 19871903.0, + "step": 776 + }, + { + "epoch": 0.08532835493081485, + "grad_norm": 2.1596791744232178, + "learning_rate": 4.259055982436882e-07, + "loss": 1.1715, + "mean_token_accuracy": 0.6634835004806519, + "num_tokens": 19901897.0, + "step": 777 + }, + { + "epoch": 0.08543817263342851, + "grad_norm": 2.3576018810272217, + "learning_rate": 4.2645444566410537e-07, + "loss": 1.1091, + "mean_token_accuracy": 0.6813899278640747, + "num_tokens": 19927129.0, + "step": 778 + }, + { + "epoch": 0.08554799033604217, + "grad_norm": 2.190836191177368, + "learning_rate": 4.270032930845225e-07, + "loss": 1.118, + "mean_token_accuracy": 0.6744362115859985, + "num_tokens": 19956770.0, + "step": 779 + }, + { + "epoch": 0.08565780803865583, + "grad_norm": 2.263556718826294, + "learning_rate": 4.275521405049396e-07, + "loss": 1.1366, + "mean_token_accuracy": 0.6692482829093933, + "num_tokens": 19985158.0, + "step": 780 + }, + { + "epoch": 0.08576762574126949, + "grad_norm": 2.537336587905884, + "learning_rate": 4.281009879253567e-07, + "loss": 1.014, + "mean_token_accuracy": 0.6981872916221619, + "num_tokens": 20008897.0, + "step": 781 + }, + { + "epoch": 0.08587744344388315, + "grad_norm": 2.331444025039673, + "learning_rate": 4.286498353457739e-07, + "loss": 1.1086, + "mean_token_accuracy": 0.6737514138221741, + "num_tokens": 20034797.0, + "step": 782 + }, + { + "epoch": 0.08598726114649681, + "grad_norm": 2.354487419128418, + "learning_rate": 4.29198682766191e-07, + "loss": 1.0622, + "mean_token_accuracy": 0.6850799322128296, + "num_tokens": 20061622.0, + "step": 783 + }, + { + "epoch": 0.08609707884911047, + "grad_norm": 2.207035779953003, + "learning_rate": 4.2974753018660806e-07, + "loss": 1.1454, + "mean_token_accuracy": 0.6745589971542358, + "num_tokens": 20089729.0, + "step": 784 + }, + { + "epoch": 0.08620689655172414, + "grad_norm": 2.763854742050171, + "learning_rate": 4.3029637760702523e-07, + "loss": 1.1619, + "mean_token_accuracy": 0.6655670404434204, + "num_tokens": 20112254.0, + "step": 785 + }, + { + "epoch": 0.0863167142543378, + "grad_norm": 2.663074254989624, + "learning_rate": 4.3084522502744235e-07, + "loss": 1.0424, + "mean_token_accuracy": 0.6898553371429443, + "num_tokens": 20134081.0, + "step": 786 + }, + { + "epoch": 0.08642653195695146, + "grad_norm": 2.558579683303833, + "learning_rate": 4.3139407244785946e-07, + "loss": 1.0851, + "mean_token_accuracy": 0.6895039081573486, + "num_tokens": 20156955.0, + "step": 787 + }, + { + "epoch": 0.08653634965956512, + "grad_norm": 2.6161699295043945, + "learning_rate": 4.319429198682766e-07, + "loss": 1.1484, + "mean_token_accuracy": 0.6598751544952393, + "num_tokens": 20181125.0, + "step": 788 + }, + { + "epoch": 0.08664616736217878, + "grad_norm": 2.5832221508026123, + "learning_rate": 4.3249176728869374e-07, + "loss": 1.125, + "mean_token_accuracy": 0.6716529130935669, + "num_tokens": 20206103.0, + "step": 789 + }, + { + "epoch": 0.08675598506479244, + "grad_norm": 2.357356071472168, + "learning_rate": 4.3304061470911086e-07, + "loss": 1.054, + "mean_token_accuracy": 0.6888317465782166, + "num_tokens": 20232157.0, + "step": 790 + }, + { + "epoch": 0.0868658027674061, + "grad_norm": 2.4672253131866455, + "learning_rate": 4.33589462129528e-07, + "loss": 1.1085, + "mean_token_accuracy": 0.6710876822471619, + "num_tokens": 20256859.0, + "step": 791 + }, + { + "epoch": 0.08697562047001976, + "grad_norm": 2.410008192062378, + "learning_rate": 4.341383095499451e-07, + "loss": 1.0832, + "mean_token_accuracy": 0.6810652613639832, + "num_tokens": 20281937.0, + "step": 792 + }, + { + "epoch": 0.08708543817263342, + "grad_norm": 2.569831132888794, + "learning_rate": 4.3468715697036226e-07, + "loss": 1.0527, + "mean_token_accuracy": 0.6919360160827637, + "num_tokens": 20304899.0, + "step": 793 + }, + { + "epoch": 0.0871952558752471, + "grad_norm": 2.1304523944854736, + "learning_rate": 4.352360043907793e-07, + "loss": 1.0375, + "mean_token_accuracy": 0.6899349689483643, + "num_tokens": 20334398.0, + "step": 794 + }, + { + "epoch": 0.08730507357786076, + "grad_norm": 2.084228277206421, + "learning_rate": 4.3578485181119644e-07, + "loss": 1.105, + "mean_token_accuracy": 0.6667703986167908, + "num_tokens": 20366927.0, + "step": 795 + }, + { + "epoch": 0.08741489128047442, + "grad_norm": 2.456718921661377, + "learning_rate": 4.363336992316136e-07, + "loss": 1.0621, + "mean_token_accuracy": 0.685893177986145, + "num_tokens": 20390120.0, + "step": 796 + }, + { + "epoch": 0.08752470898308808, + "grad_norm": 2.1228959560394287, + "learning_rate": 4.368825466520307e-07, + "loss": 1.1223, + "mean_token_accuracy": 0.6724272966384888, + "num_tokens": 20421725.0, + "step": 797 + }, + { + "epoch": 0.08763452668570174, + "grad_norm": 2.7209277153015137, + "learning_rate": 4.3743139407244783e-07, + "loss": 1.0956, + "mean_token_accuracy": 0.6754980087280273, + "num_tokens": 20443511.0, + "step": 798 + }, + { + "epoch": 0.0877443443883154, + "grad_norm": 2.498218536376953, + "learning_rate": 4.3798024149286495e-07, + "loss": 1.1417, + "mean_token_accuracy": 0.6664590835571289, + "num_tokens": 20468175.0, + "step": 799 + }, + { + "epoch": 0.08785416209092906, + "grad_norm": 2.6861000061035156, + "learning_rate": 4.385290889132821e-07, + "loss": 1.0643, + "mean_token_accuracy": 0.6865451335906982, + "num_tokens": 20490602.0, + "step": 800 + }, + { + "epoch": 0.08796397979354272, + "grad_norm": 2.067079782485962, + "learning_rate": 4.3907793633369923e-07, + "loss": 1.0103, + "mean_token_accuracy": 0.6955727934837341, + "num_tokens": 20521336.0, + "step": 801 + }, + { + "epoch": 0.08807379749615638, + "grad_norm": 2.424811363220215, + "learning_rate": 4.3962678375411635e-07, + "loss": 1.0261, + "mean_token_accuracy": 0.6926549673080444, + "num_tokens": 20546530.0, + "step": 802 + }, + { + "epoch": 0.08818361519877004, + "grad_norm": 2.6124112606048584, + "learning_rate": 4.401756311745334e-07, + "loss": 1.0937, + "mean_token_accuracy": 0.6708410978317261, + "num_tokens": 20571465.0, + "step": 803 + }, + { + "epoch": 0.08829343290138371, + "grad_norm": 2.4224414825439453, + "learning_rate": 4.407244785949506e-07, + "loss": 1.0184, + "mean_token_accuracy": 0.6905995607376099, + "num_tokens": 20596761.0, + "step": 804 + }, + { + "epoch": 0.08840325060399737, + "grad_norm": 2.7289295196533203, + "learning_rate": 4.412733260153677e-07, + "loss": 1.138, + "mean_token_accuracy": 0.6685812473297119, + "num_tokens": 20620289.0, + "step": 805 + }, + { + "epoch": 0.08851306830661103, + "grad_norm": 2.644758701324463, + "learning_rate": 4.418221734357848e-07, + "loss": 1.133, + "mean_token_accuracy": 0.6645418405532837, + "num_tokens": 20642644.0, + "step": 806 + }, + { + "epoch": 0.08862288600922469, + "grad_norm": 2.42580509185791, + "learning_rate": 4.42371020856202e-07, + "loss": 1.0386, + "mean_token_accuracy": 0.6797176003456116, + "num_tokens": 20667106.0, + "step": 807 + }, + { + "epoch": 0.08873270371183835, + "grad_norm": 2.3083336353302, + "learning_rate": 4.429198682766191e-07, + "loss": 1.085, + "mean_token_accuracy": 0.6728224754333496, + "num_tokens": 20694408.0, + "step": 808 + }, + { + "epoch": 0.08884252141445201, + "grad_norm": 2.461409091949463, + "learning_rate": 4.434687156970362e-07, + "loss": 0.9546, + "mean_token_accuracy": 0.7160836458206177, + "num_tokens": 20720291.0, + "step": 809 + }, + { + "epoch": 0.08895233911706567, + "grad_norm": 2.3394410610198975, + "learning_rate": 4.440175631174533e-07, + "loss": 1.131, + "mean_token_accuracy": 0.6692497730255127, + "num_tokens": 20747054.0, + "step": 810 + }, + { + "epoch": 0.08906215681967933, + "grad_norm": 2.462390422821045, + "learning_rate": 4.445664105378705e-07, + "loss": 1.1155, + "mean_token_accuracy": 0.6681057810783386, + "num_tokens": 20773240.0, + "step": 811 + }, + { + "epoch": 0.08917197452229299, + "grad_norm": 2.389068126678467, + "learning_rate": 4.451152579582876e-07, + "loss": 1.022, + "mean_token_accuracy": 0.7041659951210022, + "num_tokens": 20799862.0, + "step": 812 + }, + { + "epoch": 0.08928179222490666, + "grad_norm": 2.4365651607513428, + "learning_rate": 4.4566410537870467e-07, + "loss": 1.016, + "mean_token_accuracy": 0.6982905864715576, + "num_tokens": 20822312.0, + "step": 813 + }, + { + "epoch": 0.08939160992752032, + "grad_norm": 2.605113983154297, + "learning_rate": 4.462129527991218e-07, + "loss": 1.1452, + "mean_token_accuracy": 0.6733986139297485, + "num_tokens": 20843983.0, + "step": 814 + }, + { + "epoch": 0.08950142763013398, + "grad_norm": 2.4210293292999268, + "learning_rate": 4.4676180021953895e-07, + "loss": 1.1463, + "mean_token_accuracy": 0.6639403104782104, + "num_tokens": 20870212.0, + "step": 815 + }, + { + "epoch": 0.08961124533274764, + "grad_norm": 2.5806374549865723, + "learning_rate": 4.4731064763995607e-07, + "loss": 0.9605, + "mean_token_accuracy": 0.7079640626907349, + "num_tokens": 20891629.0, + "step": 816 + }, + { + "epoch": 0.0897210630353613, + "grad_norm": 2.6757500171661377, + "learning_rate": 4.478594950603732e-07, + "loss": 1.0208, + "mean_token_accuracy": 0.695677638053894, + "num_tokens": 20911161.0, + "step": 817 + }, + { + "epoch": 0.08983088073797496, + "grad_norm": 2.4611499309539795, + "learning_rate": 4.4840834248079035e-07, + "loss": 1.0512, + "mean_token_accuracy": 0.6934946179389954, + "num_tokens": 20935411.0, + "step": 818 + }, + { + "epoch": 0.08994069844058862, + "grad_norm": 2.6403560638427734, + "learning_rate": 4.4895718990120747e-07, + "loss": 1.0892, + "mean_token_accuracy": 0.6808372139930725, + "num_tokens": 20956845.0, + "step": 819 + }, + { + "epoch": 0.09005051614320228, + "grad_norm": 2.149343490600586, + "learning_rate": 4.495060373216246e-07, + "loss": 1.108, + "mean_token_accuracy": 0.6741326451301575, + "num_tokens": 20986300.0, + "step": 820 + }, + { + "epoch": 0.09016033384581594, + "grad_norm": 2.4599149227142334, + "learning_rate": 4.500548847420417e-07, + "loss": 1.1508, + "mean_token_accuracy": 0.6672807931900024, + "num_tokens": 21011953.0, + "step": 821 + }, + { + "epoch": 0.0902701515484296, + "grad_norm": 2.635227918624878, + "learning_rate": 4.5060373216245886e-07, + "loss": 1.0116, + "mean_token_accuracy": 0.6974774599075317, + "num_tokens": 21033790.0, + "step": 822 + }, + { + "epoch": 0.09037996925104327, + "grad_norm": 2.2329940795898438, + "learning_rate": 4.5115257958287593e-07, + "loss": 1.1278, + "mean_token_accuracy": 0.6722120046615601, + "num_tokens": 21062854.0, + "step": 823 + }, + { + "epoch": 0.09048978695365693, + "grad_norm": 2.459817409515381, + "learning_rate": 4.5170142700329304e-07, + "loss": 1.0807, + "mean_token_accuracy": 0.6806789636611938, + "num_tokens": 21088082.0, + "step": 824 + }, + { + "epoch": 0.0905996046562706, + "grad_norm": 2.346290349960327, + "learning_rate": 4.5225027442371016e-07, + "loss": 1.162, + "mean_token_accuracy": 0.6648001670837402, + "num_tokens": 21115205.0, + "step": 825 + }, + { + "epoch": 0.09070942235888425, + "grad_norm": 2.280510425567627, + "learning_rate": 4.527991218441273e-07, + "loss": 1.0942, + "mean_token_accuracy": 0.6768864393234253, + "num_tokens": 21142992.0, + "step": 826 + }, + { + "epoch": 0.09081924006149791, + "grad_norm": 2.2977092266082764, + "learning_rate": 4.5334796926454444e-07, + "loss": 1.1071, + "mean_token_accuracy": 0.6737522482872009, + "num_tokens": 21168979.0, + "step": 827 + }, + { + "epoch": 0.09092905776411157, + "grad_norm": 2.2231109142303467, + "learning_rate": 4.5389681668496156e-07, + "loss": 1.1906, + "mean_token_accuracy": 0.6620775461196899, + "num_tokens": 21198666.0, + "step": 828 + }, + { + "epoch": 0.09103887546672523, + "grad_norm": 2.5978121757507324, + "learning_rate": 4.5444566410537867e-07, + "loss": 1.082, + "mean_token_accuracy": 0.6783775687217712, + "num_tokens": 21221293.0, + "step": 829 + }, + { + "epoch": 0.09114869316933889, + "grad_norm": 2.4462645053863525, + "learning_rate": 4.5499451152579584e-07, + "loss": 1.0864, + "mean_token_accuracy": 0.6864690184593201, + "num_tokens": 21243956.0, + "step": 830 + }, + { + "epoch": 0.09125851087195255, + "grad_norm": 2.3434078693389893, + "learning_rate": 4.5554335894621295e-07, + "loss": 1.0737, + "mean_token_accuracy": 0.6770188212394714, + "num_tokens": 21270024.0, + "step": 831 + }, + { + "epoch": 0.09136832857456623, + "grad_norm": 2.172039747238159, + "learning_rate": 4.5609220636663e-07, + "loss": 1.1376, + "mean_token_accuracy": 0.6676913499832153, + "num_tokens": 21300419.0, + "step": 832 + }, + { + "epoch": 0.09147814627717989, + "grad_norm": 2.725252389907837, + "learning_rate": 4.566410537870472e-07, + "loss": 1.0547, + "mean_token_accuracy": 0.6843485832214355, + "num_tokens": 21321015.0, + "step": 833 + }, + { + "epoch": 0.09158796397979355, + "grad_norm": 2.4952197074890137, + "learning_rate": 4.571899012074643e-07, + "loss": 1.0208, + "mean_token_accuracy": 0.6940531730651855, + "num_tokens": 21342005.0, + "step": 834 + }, + { + "epoch": 0.0916977816824072, + "grad_norm": 2.3016533851623535, + "learning_rate": 4.577387486278814e-07, + "loss": 1.0495, + "mean_token_accuracy": 0.6878721714019775, + "num_tokens": 21368697.0, + "step": 835 + }, + { + "epoch": 0.09180759938502087, + "grad_norm": 2.3269619941711426, + "learning_rate": 4.5828759604829853e-07, + "loss": 1.0112, + "mean_token_accuracy": 0.7018327713012695, + "num_tokens": 21392864.0, + "step": 836 + }, + { + "epoch": 0.09191741708763453, + "grad_norm": 2.6208906173706055, + "learning_rate": 4.588364434687157e-07, + "loss": 1.0406, + "mean_token_accuracy": 0.6910344362258911, + "num_tokens": 21413445.0, + "step": 837 + }, + { + "epoch": 0.09202723479024819, + "grad_norm": 2.6270792484283447, + "learning_rate": 4.593852908891328e-07, + "loss": 1.1253, + "mean_token_accuracy": 0.666752278804779, + "num_tokens": 21434710.0, + "step": 838 + }, + { + "epoch": 0.09213705249286185, + "grad_norm": 2.4900481700897217, + "learning_rate": 4.5993413830954993e-07, + "loss": 1.0327, + "mean_token_accuracy": 0.6988094449043274, + "num_tokens": 21458275.0, + "step": 839 + }, + { + "epoch": 0.0922468701954755, + "grad_norm": 2.1824381351470947, + "learning_rate": 4.6048298572996704e-07, + "loss": 1.1182, + "mean_token_accuracy": 0.6756604909896851, + "num_tokens": 21488728.0, + "step": 840 + }, + { + "epoch": 0.09235668789808917, + "grad_norm": 3.027283191680908, + "learning_rate": 4.6103183315038416e-07, + "loss": 0.9815, + "mean_token_accuracy": 0.7128303050994873, + "num_tokens": 21507594.0, + "step": 841 + }, + { + "epoch": 0.09246650560070284, + "grad_norm": 2.328662157058716, + "learning_rate": 4.615806805708013e-07, + "loss": 1.1028, + "mean_token_accuracy": 0.6820960640907288, + "num_tokens": 21533969.0, + "step": 842 + }, + { + "epoch": 0.0925763233033165, + "grad_norm": 2.0639660358428955, + "learning_rate": 4.621295279912184e-07, + "loss": 1.1751, + "mean_token_accuracy": 0.6607956290245056, + "num_tokens": 21565616.0, + "step": 843 + }, + { + "epoch": 0.09268614100593016, + "grad_norm": 2.447929859161377, + "learning_rate": 4.6267837541163556e-07, + "loss": 1.0355, + "mean_token_accuracy": 0.6938949823379517, + "num_tokens": 21591639.0, + "step": 844 + }, + { + "epoch": 0.09279595870854382, + "grad_norm": 2.590477228164673, + "learning_rate": 4.632272228320527e-07, + "loss": 1.0259, + "mean_token_accuracy": 0.696992039680481, + "num_tokens": 21613395.0, + "step": 845 + }, + { + "epoch": 0.09290577641115748, + "grad_norm": 2.2060766220092773, + "learning_rate": 4.637760702524698e-07, + "loss": 1.0367, + "mean_token_accuracy": 0.6983504295349121, + "num_tokens": 21642097.0, + "step": 846 + }, + { + "epoch": 0.09301559411377114, + "grad_norm": 2.6357851028442383, + "learning_rate": 4.643249176728869e-07, + "loss": 1.091, + "mean_token_accuracy": 0.6807730197906494, + "num_tokens": 21663399.0, + "step": 847 + }, + { + "epoch": 0.0931254118163848, + "grad_norm": 2.5116965770721436, + "learning_rate": 4.6487376509330407e-07, + "loss": 1.1532, + "mean_token_accuracy": 0.659818708896637, + "num_tokens": 21688368.0, + "step": 848 + }, + { + "epoch": 0.09323522951899846, + "grad_norm": 2.5227959156036377, + "learning_rate": 4.654226125137212e-07, + "loss": 1.0513, + "mean_token_accuracy": 0.6885336637496948, + "num_tokens": 21709596.0, + "step": 849 + }, + { + "epoch": 0.09334504722161212, + "grad_norm": 2.8405003547668457, + "learning_rate": 4.659714599341383e-07, + "loss": 1.0177, + "mean_token_accuracy": 0.6877191066741943, + "num_tokens": 21727516.0, + "step": 850 + }, + { + "epoch": 0.09345486492422579, + "grad_norm": 2.298578977584839, + "learning_rate": 4.6652030735455537e-07, + "loss": 1.0729, + "mean_token_accuracy": 0.683324933052063, + "num_tokens": 21754854.0, + "step": 851 + }, + { + "epoch": 0.09356468262683945, + "grad_norm": 2.210618734359741, + "learning_rate": 4.6706915477497253e-07, + "loss": 1.0208, + "mean_token_accuracy": 0.6945620775222778, + "num_tokens": 21785854.0, + "step": 852 + }, + { + "epoch": 0.09367450032945311, + "grad_norm": 2.3518383502960205, + "learning_rate": 4.6761800219538965e-07, + "loss": 1.096, + "mean_token_accuracy": 0.6736420392990112, + "num_tokens": 21811798.0, + "step": 853 + }, + { + "epoch": 0.09378431803206677, + "grad_norm": 2.4908719062805176, + "learning_rate": 4.6816684961580676e-07, + "loss": 1.0777, + "mean_token_accuracy": 0.6842461824417114, + "num_tokens": 21834337.0, + "step": 854 + }, + { + "epoch": 0.09389413573468043, + "grad_norm": 2.429827928543091, + "learning_rate": 4.6871569703622393e-07, + "loss": 1.0265, + "mean_token_accuracy": 0.6936714053153992, + "num_tokens": 21857151.0, + "step": 855 + }, + { + "epoch": 0.09400395343729409, + "grad_norm": 2.3289289474487305, + "learning_rate": 4.6926454445664105e-07, + "loss": 1.0822, + "mean_token_accuracy": 0.6797447204589844, + "num_tokens": 21882252.0, + "step": 856 + }, + { + "epoch": 0.09411377113990775, + "grad_norm": 2.932119846343994, + "learning_rate": 4.6981339187705816e-07, + "loss": 1.0533, + "mean_token_accuracy": 0.689914345741272, + "num_tokens": 21908486.0, + "step": 857 + }, + { + "epoch": 0.09422358884252141, + "grad_norm": 2.286153554916382, + "learning_rate": 4.703622392974753e-07, + "loss": 1.0178, + "mean_token_accuracy": 0.6956425905227661, + "num_tokens": 21934540.0, + "step": 858 + }, + { + "epoch": 0.09433340654513507, + "grad_norm": 2.342250108718872, + "learning_rate": 4.7091108671789245e-07, + "loss": 1.0494, + "mean_token_accuracy": 0.687605619430542, + "num_tokens": 21958784.0, + "step": 859 + }, + { + "epoch": 0.09444322424774873, + "grad_norm": 2.2076611518859863, + "learning_rate": 4.714599341383095e-07, + "loss": 1.0999, + "mean_token_accuracy": 0.6755416393280029, + "num_tokens": 21986236.0, + "step": 860 + }, + { + "epoch": 0.0945530419503624, + "grad_norm": 2.404210090637207, + "learning_rate": 4.720087815587266e-07, + "loss": 1.0913, + "mean_token_accuracy": 0.6744183897972107, + "num_tokens": 22009592.0, + "step": 861 + }, + { + "epoch": 0.09466285965297606, + "grad_norm": 2.2476062774658203, + "learning_rate": 4.7255762897914374e-07, + "loss": 1.0595, + "mean_token_accuracy": 0.6898123025894165, + "num_tokens": 22036553.0, + "step": 862 + }, + { + "epoch": 0.09477267735558972, + "grad_norm": 2.4941306114196777, + "learning_rate": 4.731064763995609e-07, + "loss": 1.0932, + "mean_token_accuracy": 0.6822105646133423, + "num_tokens": 22060304.0, + "step": 863 + }, + { + "epoch": 0.09488249505820338, + "grad_norm": 2.182189702987671, + "learning_rate": 4.73655323819978e-07, + "loss": 1.0261, + "mean_token_accuracy": 0.6951183080673218, + "num_tokens": 22088479.0, + "step": 864 + }, + { + "epoch": 0.09499231276081704, + "grad_norm": 1.9010167121887207, + "learning_rate": 4.7420417124039514e-07, + "loss": 1.1219, + "mean_token_accuracy": 0.6658769249916077, + "num_tokens": 22125683.0, + "step": 865 + }, + { + "epoch": 0.0951021304634307, + "grad_norm": 2.54883074760437, + "learning_rate": 4.747530186608123e-07, + "loss": 0.9898, + "mean_token_accuracy": 0.7018548250198364, + "num_tokens": 22147353.0, + "step": 866 + }, + { + "epoch": 0.09521194816604436, + "grad_norm": 2.3961093425750732, + "learning_rate": 4.753018660812294e-07, + "loss": 0.987, + "mean_token_accuracy": 0.6992750763893127, + "num_tokens": 22169454.0, + "step": 867 + }, + { + "epoch": 0.09532176586865802, + "grad_norm": 2.479604959487915, + "learning_rate": 4.7585071350164654e-07, + "loss": 0.9827, + "mean_token_accuracy": 0.7015714645385742, + "num_tokens": 22194157.0, + "step": 868 + }, + { + "epoch": 0.09543158357127168, + "grad_norm": 2.1308581829071045, + "learning_rate": 4.7639956092206365e-07, + "loss": 1.0174, + "mean_token_accuracy": 0.7042462825775146, + "num_tokens": 22223106.0, + "step": 869 + }, + { + "epoch": 0.09554140127388536, + "grad_norm": 2.5958163738250732, + "learning_rate": 4.769484083424808e-07, + "loss": 1.1158, + "mean_token_accuracy": 0.6747497320175171, + "num_tokens": 22245895.0, + "step": 870 + }, + { + "epoch": 0.09565121897649902, + "grad_norm": 2.8946070671081543, + "learning_rate": 4.774972557628979e-07, + "loss": 1.0692, + "mean_token_accuracy": 0.6779119968414307, + "num_tokens": 22263475.0, + "step": 871 + }, + { + "epoch": 0.09576103667911268, + "grad_norm": 2.2359530925750732, + "learning_rate": 4.78046103183315e-07, + "loss": 1.1516, + "mean_token_accuracy": 0.6676587462425232, + "num_tokens": 22291659.0, + "step": 872 + }, + { + "epoch": 0.09587085438172634, + "grad_norm": 2.7843503952026367, + "learning_rate": 4.785949506037321e-07, + "loss": 1.0461, + "mean_token_accuracy": 0.6897814273834229, + "num_tokens": 22311679.0, + "step": 873 + }, + { + "epoch": 0.09598067208434, + "grad_norm": 2.3472492694854736, + "learning_rate": 4.791437980241493e-07, + "loss": 1.0658, + "mean_token_accuracy": 0.6790145039558411, + "num_tokens": 22336631.0, + "step": 874 + }, + { + "epoch": 0.09609048978695366, + "grad_norm": 2.69036865234375, + "learning_rate": 4.796926454445663e-07, + "loss": 1.0349, + "mean_token_accuracy": 0.6889045834541321, + "num_tokens": 22357546.0, + "step": 875 + }, + { + "epoch": 0.09620030748956732, + "grad_norm": 2.5174710750579834, + "learning_rate": 4.802414928649835e-07, + "loss": 1.0562, + "mean_token_accuracy": 0.6911237239837646, + "num_tokens": 22379776.0, + "step": 876 + }, + { + "epoch": 0.09631012519218098, + "grad_norm": 2.2351086139678955, + "learning_rate": 4.807903402854007e-07, + "loss": 1.0188, + "mean_token_accuracy": 0.6922352313995361, + "num_tokens": 22406623.0, + "step": 877 + }, + { + "epoch": 0.09641994289479464, + "grad_norm": 2.469292402267456, + "learning_rate": 4.813391877058177e-07, + "loss": 1.0157, + "mean_token_accuracy": 0.7005493640899658, + "num_tokens": 22429796.0, + "step": 878 + }, + { + "epoch": 0.0965297605974083, + "grad_norm": 2.4201242923736572, + "learning_rate": 4.818880351262349e-07, + "loss": 1.0553, + "mean_token_accuracy": 0.6808485984802246, + "num_tokens": 22454796.0, + "step": 879 + }, + { + "epoch": 0.09663957830002197, + "grad_norm": 2.490648031234741, + "learning_rate": 4.82436882546652e-07, + "loss": 1.0853, + "mean_token_accuracy": 0.6810115575790405, + "num_tokens": 22477862.0, + "step": 880 + }, + { + "epoch": 0.09674939600263563, + "grad_norm": 2.6163742542266846, + "learning_rate": 4.829857299670691e-07, + "loss": 1.0771, + "mean_token_accuracy": 0.6793366074562073, + "num_tokens": 22498036.0, + "step": 881 + }, + { + "epoch": 0.09685921370524929, + "grad_norm": 2.266585350036621, + "learning_rate": 4.835345773874863e-07, + "loss": 1.0376, + "mean_token_accuracy": 0.6921387910842896, + "num_tokens": 22526474.0, + "step": 882 + }, + { + "epoch": 0.09696903140786295, + "grad_norm": 2.2132296562194824, + "learning_rate": 4.840834248079034e-07, + "loss": 1.045, + "mean_token_accuracy": 0.6911003589630127, + "num_tokens": 22556013.0, + "step": 883 + }, + { + "epoch": 0.09707884911047661, + "grad_norm": 2.550616502761841, + "learning_rate": 4.846322722283204e-07, + "loss": 1.0126, + "mean_token_accuracy": 0.7020692825317383, + "num_tokens": 22578406.0, + "step": 884 + }, + { + "epoch": 0.09718866681309027, + "grad_norm": 2.3484034538269043, + "learning_rate": 4.851811196487376e-07, + "loss": 1.0017, + "mean_token_accuracy": 0.6988842487335205, + "num_tokens": 22604193.0, + "step": 885 + }, + { + "epoch": 0.09729848451570393, + "grad_norm": 2.0274746417999268, + "learning_rate": 4.857299670691548e-07, + "loss": 1.0333, + "mean_token_accuracy": 0.6957651972770691, + "num_tokens": 22636587.0, + "step": 886 + }, + { + "epoch": 0.09740830221831759, + "grad_norm": 2.8121190071105957, + "learning_rate": 4.862788144895718e-07, + "loss": 1.0931, + "mean_token_accuracy": 0.6801059246063232, + "num_tokens": 22657150.0, + "step": 887 + }, + { + "epoch": 0.09751811992093125, + "grad_norm": 2.269268035888672, + "learning_rate": 4.86827661909989e-07, + "loss": 1.0308, + "mean_token_accuracy": 0.6920309066772461, + "num_tokens": 22685175.0, + "step": 888 + }, + { + "epoch": 0.09762793762354492, + "grad_norm": 2.3776094913482666, + "learning_rate": 4.873765093304062e-07, + "loss": 0.9651, + "mean_token_accuracy": 0.7160487174987793, + "num_tokens": 22711330.0, + "step": 889 + }, + { + "epoch": 0.09773775532615858, + "grad_norm": 2.494279623031616, + "learning_rate": 4.879253567508232e-07, + "loss": 0.976, + "mean_token_accuracy": 0.7009448409080505, + "num_tokens": 22734845.0, + "step": 890 + }, + { + "epoch": 0.09784757302877224, + "grad_norm": 2.3319265842437744, + "learning_rate": 4.884742041712404e-07, + "loss": 1.0869, + "mean_token_accuracy": 0.6882351636886597, + "num_tokens": 22761008.0, + "step": 891 + }, + { + "epoch": 0.0979573907313859, + "grad_norm": 2.225543737411499, + "learning_rate": 4.890230515916576e-07, + "loss": 1.0925, + "mean_token_accuracy": 0.6781523823738098, + "num_tokens": 22787838.0, + "step": 892 + }, + { + "epoch": 0.09806720843399956, + "grad_norm": 2.214270830154419, + "learning_rate": 4.895718990120746e-07, + "loss": 1.2256, + "mean_token_accuracy": 0.6463714838027954, + "num_tokens": 22818692.0, + "step": 893 + }, + { + "epoch": 0.09817702613661322, + "grad_norm": 2.3523640632629395, + "learning_rate": 4.901207464324917e-07, + "loss": 1.0226, + "mean_token_accuracy": 0.6972104907035828, + "num_tokens": 22843223.0, + "step": 894 + }, + { + "epoch": 0.09828684383922688, + "grad_norm": 2.8114328384399414, + "learning_rate": 4.906695938529089e-07, + "loss": 1.0852, + "mean_token_accuracy": 0.6779536008834839, + "num_tokens": 22863838.0, + "step": 895 + }, + { + "epoch": 0.09839666154184054, + "grad_norm": 2.460254669189453, + "learning_rate": 4.91218441273326e-07, + "loss": 0.9266, + "mean_token_accuracy": 0.7231545448303223, + "num_tokens": 22886852.0, + "step": 896 + }, + { + "epoch": 0.0985064792444542, + "grad_norm": 2.2744088172912598, + "learning_rate": 4.917672886937431e-07, + "loss": 1.0568, + "mean_token_accuracy": 0.6847584247589111, + "num_tokens": 22912983.0, + "step": 897 + }, + { + "epoch": 0.09861629694706786, + "grad_norm": 2.3542654514312744, + "learning_rate": 4.923161361141603e-07, + "loss": 1.0994, + "mean_token_accuracy": 0.6707340478897095, + "num_tokens": 22936864.0, + "step": 898 + }, + { + "epoch": 0.09872611464968153, + "grad_norm": 2.3903539180755615, + "learning_rate": 4.928649835345773e-07, + "loss": 1.1665, + "mean_token_accuracy": 0.6831038594245911, + "num_tokens": 22962355.0, + "step": 899 + }, + { + "epoch": 0.0988359323522952, + "grad_norm": 2.370640754699707, + "learning_rate": 4.934138309549945e-07, + "loss": 1.1345, + "mean_token_accuracy": 0.6658662557601929, + "num_tokens": 22990826.0, + "step": 900 + }, + { + "epoch": 0.09894575005490885, + "grad_norm": 2.6415109634399414, + "learning_rate": 4.939626783754117e-07, + "loss": 1.1063, + "mean_token_accuracy": 0.6732673645019531, + "num_tokens": 23013188.0, + "step": 901 + }, + { + "epoch": 0.09905556775752251, + "grad_norm": 2.1205391883850098, + "learning_rate": 4.945115257958287e-07, + "loss": 1.108, + "mean_token_accuracy": 0.6797193884849548, + "num_tokens": 23045318.0, + "step": 902 + }, + { + "epoch": 0.09916538546013617, + "grad_norm": 2.3345797061920166, + "learning_rate": 4.950603732162459e-07, + "loss": 0.9558, + "mean_token_accuracy": 0.7137778997421265, + "num_tokens": 23070482.0, + "step": 903 + }, + { + "epoch": 0.09927520316274983, + "grad_norm": 2.4889473915100098, + "learning_rate": 4.95609220636663e-07, + "loss": 1.058, + "mean_token_accuracy": 0.693440854549408, + "num_tokens": 23093539.0, + "step": 904 + }, + { + "epoch": 0.0993850208653635, + "grad_norm": 2.1564958095550537, + "learning_rate": 4.961580680570801e-07, + "loss": 1.1219, + "mean_token_accuracy": 0.6738736629486084, + "num_tokens": 23123781.0, + "step": 905 + }, + { + "epoch": 0.09949483856797715, + "grad_norm": 2.342709541320801, + "learning_rate": 4.967069154774972e-07, + "loss": 1.0946, + "mean_token_accuracy": 0.6767503023147583, + "num_tokens": 23151164.0, + "step": 906 + }, + { + "epoch": 0.09960465627059081, + "grad_norm": 2.30281400680542, + "learning_rate": 4.972557628979143e-07, + "loss": 1.1059, + "mean_token_accuracy": 0.6775310039520264, + "num_tokens": 23177124.0, + "step": 907 + }, + { + "epoch": 0.09971447397320449, + "grad_norm": 2.2246899604797363, + "learning_rate": 4.978046103183315e-07, + "loss": 1.1057, + "mean_token_accuracy": 0.675663948059082, + "num_tokens": 23204762.0, + "step": 908 + }, + { + "epoch": 0.09982429167581815, + "grad_norm": 2.171180009841919, + "learning_rate": 4.983534577387486e-07, + "loss": 1.0793, + "mean_token_accuracy": 0.6852276921272278, + "num_tokens": 23233756.0, + "step": 909 + }, + { + "epoch": 0.0999341093784318, + "grad_norm": 2.066121816635132, + "learning_rate": 4.989023051591657e-07, + "loss": 1.1128, + "mean_token_accuracy": 0.6808335781097412, + "num_tokens": 23263665.0, + "step": 910 + }, + { + "epoch": 0.10004392708104547, + "grad_norm": 2.471226453781128, + "learning_rate": 4.994511525795829e-07, + "loss": 1.0649, + "mean_token_accuracy": 0.6804245710372925, + "num_tokens": 23287349.0, + "step": 911 + }, + { + "epoch": 0.10015374478365913, + "grad_norm": 2.0553138256073, + "learning_rate": 5e-07, + "loss": 1.1738, + "mean_token_accuracy": 0.6608452200889587, + "num_tokens": 23320157.0, + "step": 912 + }, + { + "epoch": 0.10026356248627279, + "grad_norm": 2.708853244781494, + "learning_rate": 5.005488474204171e-07, + "loss": 1.1494, + "mean_token_accuracy": 0.6652776002883911, + "num_tokens": 23341136.0, + "step": 913 + }, + { + "epoch": 0.10037338018888645, + "grad_norm": 2.4886085987091064, + "learning_rate": 5.010976948408342e-07, + "loss": 1.0115, + "mean_token_accuracy": 0.6964856386184692, + "num_tokens": 23364136.0, + "step": 914 + }, + { + "epoch": 0.1004831978915001, + "grad_norm": 2.0723862648010254, + "learning_rate": 5.016465422612514e-07, + "loss": 1.1143, + "mean_token_accuracy": 0.6743341088294983, + "num_tokens": 23395798.0, + "step": 915 + }, + { + "epoch": 0.10059301559411377, + "grad_norm": 2.671309232711792, + "learning_rate": 5.021953896816685e-07, + "loss": 1.0477, + "mean_token_accuracy": 0.6873657703399658, + "num_tokens": 23419795.0, + "step": 916 + }, + { + "epoch": 0.10070283329672743, + "grad_norm": 2.4470129013061523, + "learning_rate": 5.027442371020856e-07, + "loss": 1.085, + "mean_token_accuracy": 0.6825889348983765, + "num_tokens": 23444721.0, + "step": 917 + }, + { + "epoch": 0.1008126509993411, + "grad_norm": 2.3949625492095947, + "learning_rate": 5.032930845225028e-07, + "loss": 0.9904, + "mean_token_accuracy": 0.7001777291297913, + "num_tokens": 23471001.0, + "step": 918 + }, + { + "epoch": 0.10092246870195476, + "grad_norm": 2.151501417160034, + "learning_rate": 5.038419319429198e-07, + "loss": 1.0676, + "mean_token_accuracy": 0.6885464191436768, + "num_tokens": 23499426.0, + "step": 919 + }, + { + "epoch": 0.10103228640456842, + "grad_norm": 2.180629253387451, + "learning_rate": 5.04390779363337e-07, + "loss": 1.0228, + "mean_token_accuracy": 0.6928173899650574, + "num_tokens": 23527024.0, + "step": 920 + }, + { + "epoch": 0.10114210410718208, + "grad_norm": 2.416217803955078, + "learning_rate": 5.049396267837542e-07, + "loss": 1.008, + "mean_token_accuracy": 0.70012366771698, + "num_tokens": 23551376.0, + "step": 921 + }, + { + "epoch": 0.10125192180979574, + "grad_norm": 2.3359012603759766, + "learning_rate": 5.054884742041711e-07, + "loss": 1.0766, + "mean_token_accuracy": 0.6771988868713379, + "num_tokens": 23575744.0, + "step": 922 + }, + { + "epoch": 0.1013617395124094, + "grad_norm": 2.3699517250061035, + "learning_rate": 5.060373216245883e-07, + "loss": 1.0543, + "mean_token_accuracy": 0.6865376234054565, + "num_tokens": 23600425.0, + "step": 923 + }, + { + "epoch": 0.10147155721502306, + "grad_norm": 2.05851674079895, + "learning_rate": 5.065861690450055e-07, + "loss": 1.1441, + "mean_token_accuracy": 0.6647499799728394, + "num_tokens": 23631452.0, + "step": 924 + }, + { + "epoch": 0.10158137491763672, + "grad_norm": 2.3162007331848145, + "learning_rate": 5.071350164654225e-07, + "loss": 1.1466, + "mean_token_accuracy": 0.662135660648346, + "num_tokens": 23660164.0, + "step": 925 + }, + { + "epoch": 0.10169119262025038, + "grad_norm": 2.9730496406555176, + "learning_rate": 5.076838638858397e-07, + "loss": 1.0048, + "mean_token_accuracy": 0.7049232125282288, + "num_tokens": 23678007.0, + "step": 926 + }, + { + "epoch": 0.10180101032286405, + "grad_norm": 2.5300769805908203, + "learning_rate": 5.082327113062569e-07, + "loss": 1.0221, + "mean_token_accuracy": 0.6922160387039185, + "num_tokens": 23699329.0, + "step": 927 + }, + { + "epoch": 0.10191082802547771, + "grad_norm": 2.15704083442688, + "learning_rate": 5.087815587266739e-07, + "loss": 1.0393, + "mean_token_accuracy": 0.6887442469596863, + "num_tokens": 23728979.0, + "step": 928 + }, + { + "epoch": 0.10202064572809137, + "grad_norm": 2.522015333175659, + "learning_rate": 5.093304061470911e-07, + "loss": 1.0788, + "mean_token_accuracy": 0.6764909029006958, + "num_tokens": 23753266.0, + "step": 929 + }, + { + "epoch": 0.10213046343070503, + "grad_norm": 2.5262372493743896, + "learning_rate": 5.098792535675082e-07, + "loss": 1.0647, + "mean_token_accuracy": 0.690932035446167, + "num_tokens": 23777157.0, + "step": 930 + }, + { + "epoch": 0.10224028113331869, + "grad_norm": 2.294611692428589, + "learning_rate": 5.104281009879253e-07, + "loss": 1.116, + "mean_token_accuracy": 0.6713793277740479, + "num_tokens": 23807321.0, + "step": 931 + }, + { + "epoch": 0.10235009883593235, + "grad_norm": 2.5643978118896484, + "learning_rate": 5.109769484083425e-07, + "loss": 1.1312, + "mean_token_accuracy": 0.6676867008209229, + "num_tokens": 23831767.0, + "step": 932 + }, + { + "epoch": 0.10245991653854601, + "grad_norm": 2.4293200969696045, + "learning_rate": 5.115257958287596e-07, + "loss": 1.0496, + "mean_token_accuracy": 0.6874143481254578, + "num_tokens": 23856154.0, + "step": 933 + }, + { + "epoch": 0.10256973424115967, + "grad_norm": 2.4193966388702393, + "learning_rate": 5.120746432491767e-07, + "loss": 1.0766, + "mean_token_accuracy": 0.6869078874588013, + "num_tokens": 23880326.0, + "step": 934 + }, + { + "epoch": 0.10267955194377333, + "grad_norm": 2.3084194660186768, + "learning_rate": 5.126234906695939e-07, + "loss": 1.1288, + "mean_token_accuracy": 0.6736059188842773, + "num_tokens": 23907007.0, + "step": 935 + }, + { + "epoch": 0.10278936964638699, + "grad_norm": 2.4556519985198975, + "learning_rate": 5.13172338090011e-07, + "loss": 1.0822, + "mean_token_accuracy": 0.6790939569473267, + "num_tokens": 23932485.0, + "step": 936 + }, + { + "epoch": 0.10289918734900066, + "grad_norm": 2.44912052154541, + "learning_rate": 5.137211855104281e-07, + "loss": 1.0885, + "mean_token_accuracy": 0.6791197061538696, + "num_tokens": 23959778.0, + "step": 937 + }, + { + "epoch": 0.10300900505161432, + "grad_norm": 2.6294143199920654, + "learning_rate": 5.142700329308453e-07, + "loss": 1.0157, + "mean_token_accuracy": 0.6955337524414062, + "num_tokens": 23980976.0, + "step": 938 + }, + { + "epoch": 0.10311882275422798, + "grad_norm": 2.8802602291107178, + "learning_rate": 5.148188803512624e-07, + "loss": 0.9295, + "mean_token_accuracy": 0.7179599404335022, + "num_tokens": 23999382.0, + "step": 939 + }, + { + "epoch": 0.10322864045684164, + "grad_norm": 2.2976901531219482, + "learning_rate": 5.153677277716795e-07, + "loss": 1.0354, + "mean_token_accuracy": 0.6935293674468994, + "num_tokens": 24024903.0, + "step": 940 + }, + { + "epoch": 0.1033384581594553, + "grad_norm": 2.703899383544922, + "learning_rate": 5.159165751920965e-07, + "loss": 0.9862, + "mean_token_accuracy": 0.7024394869804382, + "num_tokens": 24046757.0, + "step": 941 + }, + { + "epoch": 0.10344827586206896, + "grad_norm": 2.1606898307800293, + "learning_rate": 5.164654226125136e-07, + "loss": 1.1557, + "mean_token_accuracy": 0.6585103273391724, + "num_tokens": 24077973.0, + "step": 942 + }, + { + "epoch": 0.10355809356468262, + "grad_norm": 2.258960008621216, + "learning_rate": 5.170142700329308e-07, + "loss": 1.1201, + "mean_token_accuracy": 0.6695854663848877, + "num_tokens": 24105899.0, + "step": 943 + }, + { + "epoch": 0.10366791126729628, + "grad_norm": 2.4711434841156006, + "learning_rate": 5.175631174533479e-07, + "loss": 1.0823, + "mean_token_accuracy": 0.6757874488830566, + "num_tokens": 24130261.0, + "step": 944 + }, + { + "epoch": 0.10377772896990994, + "grad_norm": 2.156235456466675, + "learning_rate": 5.18111964873765e-07, + "loss": 1.0251, + "mean_token_accuracy": 0.6950032711029053, + "num_tokens": 24159436.0, + "step": 945 + }, + { + "epoch": 0.10388754667252362, + "grad_norm": 2.221583843231201, + "learning_rate": 5.186608122941822e-07, + "loss": 0.9705, + "mean_token_accuracy": 0.7032124996185303, + "num_tokens": 24186477.0, + "step": 946 + }, + { + "epoch": 0.10399736437513728, + "grad_norm": 2.3940560817718506, + "learning_rate": 5.192096597145993e-07, + "loss": 1.0788, + "mean_token_accuracy": 0.6809991598129272, + "num_tokens": 24210702.0, + "step": 947 + }, + { + "epoch": 0.10410718207775094, + "grad_norm": 2.6425936222076416, + "learning_rate": 5.197585071350164e-07, + "loss": 0.9973, + "mean_token_accuracy": 0.7015781402587891, + "num_tokens": 24232238.0, + "step": 948 + }, + { + "epoch": 0.1042169997803646, + "grad_norm": 2.400914192199707, + "learning_rate": 5.203073545554336e-07, + "loss": 1.1265, + "mean_token_accuracy": 0.659905731678009, + "num_tokens": 24256264.0, + "step": 949 + }, + { + "epoch": 0.10432681748297826, + "grad_norm": 2.167188882827759, + "learning_rate": 5.208562019758507e-07, + "loss": 1.097, + "mean_token_accuracy": 0.6728580594062805, + "num_tokens": 24285773.0, + "step": 950 + }, + { + "epoch": 0.10443663518559192, + "grad_norm": 2.309260368347168, + "learning_rate": 5.214050493962678e-07, + "loss": 1.1132, + "mean_token_accuracy": 0.6752752661705017, + "num_tokens": 24312153.0, + "step": 951 + }, + { + "epoch": 0.10454645288820558, + "grad_norm": 2.789113998413086, + "learning_rate": 5.219538968166849e-07, + "loss": 1.0712, + "mean_token_accuracy": 0.6859441995620728, + "num_tokens": 24332263.0, + "step": 952 + }, + { + "epoch": 0.10465627059081924, + "grad_norm": 2.206923723220825, + "learning_rate": 5.225027442371021e-07, + "loss": 1.0707, + "mean_token_accuracy": 0.6927958130836487, + "num_tokens": 24358898.0, + "step": 953 + }, + { + "epoch": 0.1047660882934329, + "grad_norm": 2.4271910190582275, + "learning_rate": 5.230515916575192e-07, + "loss": 0.8983, + "mean_token_accuracy": 0.7157496213912964, + "num_tokens": 24380321.0, + "step": 954 + }, + { + "epoch": 0.10487590599604656, + "grad_norm": 2.7210795879364014, + "learning_rate": 5.236004390779363e-07, + "loss": 1.0008, + "mean_token_accuracy": 0.6981990337371826, + "num_tokens": 24400364.0, + "step": 955 + }, + { + "epoch": 0.10498572369866023, + "grad_norm": 2.337341547012329, + "learning_rate": 5.241492864983535e-07, + "loss": 1.0676, + "mean_token_accuracy": 0.6886004209518433, + "num_tokens": 24425457.0, + "step": 956 + }, + { + "epoch": 0.10509554140127389, + "grad_norm": 2.3623976707458496, + "learning_rate": 5.246981339187706e-07, + "loss": 1.0863, + "mean_token_accuracy": 0.6760930418968201, + "num_tokens": 24452864.0, + "step": 957 + }, + { + "epoch": 0.10520535910388755, + "grad_norm": 2.2092158794403076, + "learning_rate": 5.252469813391877e-07, + "loss": 1.0348, + "mean_token_accuracy": 0.6902472972869873, + "num_tokens": 24480147.0, + "step": 958 + }, + { + "epoch": 0.10531517680650121, + "grad_norm": 2.4425127506256104, + "learning_rate": 5.257958287596049e-07, + "loss": 1.0532, + "mean_token_accuracy": 0.6875948309898376, + "num_tokens": 24503896.0, + "step": 959 + }, + { + "epoch": 0.10542499450911487, + "grad_norm": 2.206336259841919, + "learning_rate": 5.263446761800219e-07, + "loss": 1.1185, + "mean_token_accuracy": 0.6697440147399902, + "num_tokens": 24530613.0, + "step": 960 + }, + { + "epoch": 0.10553481221172853, + "grad_norm": 2.1301612854003906, + "learning_rate": 5.26893523600439e-07, + "loss": 1.0028, + "mean_token_accuracy": 0.6995295286178589, + "num_tokens": 24560844.0, + "step": 961 + }, + { + "epoch": 0.10564462991434219, + "grad_norm": 2.304842233657837, + "learning_rate": 5.274423710208562e-07, + "loss": 1.0097, + "mean_token_accuracy": 0.7042255401611328, + "num_tokens": 24587670.0, + "step": 962 + }, + { + "epoch": 0.10575444761695585, + "grad_norm": 2.4004969596862793, + "learning_rate": 5.279912184412732e-07, + "loss": 1.0826, + "mean_token_accuracy": 0.6860820055007935, + "num_tokens": 24612205.0, + "step": 963 + }, + { + "epoch": 0.10586426531956951, + "grad_norm": 2.1731011867523193, + "learning_rate": 5.285400658616904e-07, + "loss": 1.0888, + "mean_token_accuracy": 0.6766583919525146, + "num_tokens": 24641900.0, + "step": 964 + }, + { + "epoch": 0.10597408302218318, + "grad_norm": 2.3617265224456787, + "learning_rate": 5.290889132821076e-07, + "loss": 1.0635, + "mean_token_accuracy": 0.6937641501426697, + "num_tokens": 24665918.0, + "step": 965 + }, + { + "epoch": 0.10608390072479684, + "grad_norm": 2.242440700531006, + "learning_rate": 5.296377607025246e-07, + "loss": 1.0734, + "mean_token_accuracy": 0.6950129270553589, + "num_tokens": 24692072.0, + "step": 966 + }, + { + "epoch": 0.1061937184274105, + "grad_norm": 2.0207924842834473, + "learning_rate": 5.301866081229418e-07, + "loss": 1.0637, + "mean_token_accuracy": 0.6793250441551208, + "num_tokens": 24722554.0, + "step": 967 + }, + { + "epoch": 0.10630353613002416, + "grad_norm": 2.351451873779297, + "learning_rate": 5.30735455543359e-07, + "loss": 1.0401, + "mean_token_accuracy": 0.6920198798179626, + "num_tokens": 24748650.0, + "step": 968 + }, + { + "epoch": 0.10641335383263782, + "grad_norm": 2.587538957595825, + "learning_rate": 5.31284302963776e-07, + "loss": 1.0268, + "mean_token_accuracy": 0.694108247756958, + "num_tokens": 24769563.0, + "step": 969 + }, + { + "epoch": 0.10652317153525148, + "grad_norm": 2.400408983230591, + "learning_rate": 5.318331503841932e-07, + "loss": 0.9607, + "mean_token_accuracy": 0.7093315124511719, + "num_tokens": 24793912.0, + "step": 970 + }, + { + "epoch": 0.10663298923786514, + "grad_norm": 2.198111057281494, + "learning_rate": 5.323819978046103e-07, + "loss": 1.0336, + "mean_token_accuracy": 0.6915076971054077, + "num_tokens": 24823333.0, + "step": 971 + }, + { + "epoch": 0.1067428069404788, + "grad_norm": 2.6178784370422363, + "learning_rate": 5.329308452250274e-07, + "loss": 1.056, + "mean_token_accuracy": 0.6903915405273438, + "num_tokens": 24844541.0, + "step": 972 + }, + { + "epoch": 0.10685262464309246, + "grad_norm": 2.09761905670166, + "learning_rate": 5.334796926454446e-07, + "loss": 1.0101, + "mean_token_accuracy": 0.7014297246932983, + "num_tokens": 24874040.0, + "step": 973 + }, + { + "epoch": 0.10696244234570612, + "grad_norm": 2.141167402267456, + "learning_rate": 5.340285400658617e-07, + "loss": 1.1035, + "mean_token_accuracy": 0.6828663349151611, + "num_tokens": 24902248.0, + "step": 974 + }, + { + "epoch": 0.1070722600483198, + "grad_norm": 2.5747663974761963, + "learning_rate": 5.345773874862788e-07, + "loss": 1.0146, + "mean_token_accuracy": 0.6885890960693359, + "num_tokens": 24922519.0, + "step": 975 + }, + { + "epoch": 0.10718207775093345, + "grad_norm": 2.554095506668091, + "learning_rate": 5.35126234906696e-07, + "loss": 0.9664, + "mean_token_accuracy": 0.7053772211074829, + "num_tokens": 24944118.0, + "step": 976 + }, + { + "epoch": 0.10729189545354711, + "grad_norm": 2.1737446784973145, + "learning_rate": 5.35675082327113e-07, + "loss": 1.0814, + "mean_token_accuracy": 0.6804926991462708, + "num_tokens": 24972689.0, + "step": 977 + }, + { + "epoch": 0.10740171315616077, + "grad_norm": 2.4476754665374756, + "learning_rate": 5.362239297475302e-07, + "loss": 1.1029, + "mean_token_accuracy": 0.6719488501548767, + "num_tokens": 24994896.0, + "step": 978 + }, + { + "epoch": 0.10751153085877443, + "grad_norm": 2.3766167163848877, + "learning_rate": 5.367727771679473e-07, + "loss": 1.0299, + "mean_token_accuracy": 0.6960015296936035, + "num_tokens": 25021022.0, + "step": 979 + }, + { + "epoch": 0.1076213485613881, + "grad_norm": 2.40889573097229, + "learning_rate": 5.373216245883643e-07, + "loss": 0.9859, + "mean_token_accuracy": 0.7000265121459961, + "num_tokens": 25044927.0, + "step": 980 + }, + { + "epoch": 0.10773116626400175, + "grad_norm": 2.166449785232544, + "learning_rate": 5.378704720087815e-07, + "loss": 1.1354, + "mean_token_accuracy": 0.6753066182136536, + "num_tokens": 25075328.0, + "step": 981 + }, + { + "epoch": 0.10784098396661541, + "grad_norm": 2.048584222793579, + "learning_rate": 5.384193194291986e-07, + "loss": 1.0598, + "mean_token_accuracy": 0.6943631172180176, + "num_tokens": 25105253.0, + "step": 982 + }, + { + "epoch": 0.10795080166922907, + "grad_norm": 2.2371182441711426, + "learning_rate": 5.389681668496157e-07, + "loss": 1.1712, + "mean_token_accuracy": 0.66316819190979, + "num_tokens": 25134880.0, + "step": 983 + }, + { + "epoch": 0.10806061937184275, + "grad_norm": 2.139439344406128, + "learning_rate": 5.395170142700329e-07, + "loss": 1.0692, + "mean_token_accuracy": 0.677188515663147, + "num_tokens": 25163796.0, + "step": 984 + }, + { + "epoch": 0.10817043707445641, + "grad_norm": 2.416607618331909, + "learning_rate": 5.4006586169045e-07, + "loss": 1.0435, + "mean_token_accuracy": 0.6921741366386414, + "num_tokens": 25189039.0, + "step": 985 + }, + { + "epoch": 0.10828025477707007, + "grad_norm": 2.50161075592041, + "learning_rate": 5.406147091108671e-07, + "loss": 1.0715, + "mean_token_accuracy": 0.6823917031288147, + "num_tokens": 25211698.0, + "step": 986 + }, + { + "epoch": 0.10839007247968373, + "grad_norm": 2.4113574028015137, + "learning_rate": 5.411635565312843e-07, + "loss": 1.0474, + "mean_token_accuracy": 0.6898902654647827, + "num_tokens": 25234334.0, + "step": 987 + }, + { + "epoch": 0.10849989018229739, + "grad_norm": 2.236767530441284, + "learning_rate": 5.417124039517014e-07, + "loss": 1.1565, + "mean_token_accuracy": 0.6661991477012634, + "num_tokens": 25263821.0, + "step": 988 + }, + { + "epoch": 0.10860970788491105, + "grad_norm": 2.3572564125061035, + "learning_rate": 5.422612513721185e-07, + "loss": 1.0711, + "mean_token_accuracy": 0.6848658323287964, + "num_tokens": 25290474.0, + "step": 989 + }, + { + "epoch": 0.1087195255875247, + "grad_norm": 2.32488751411438, + "learning_rate": 5.428100987925357e-07, + "loss": 1.0441, + "mean_token_accuracy": 0.6921615600585938, + "num_tokens": 25316439.0, + "step": 990 + }, + { + "epoch": 0.10882934329013837, + "grad_norm": 2.4524497985839844, + "learning_rate": 5.433589462129528e-07, + "loss": 1.0539, + "mean_token_accuracy": 0.7029678821563721, + "num_tokens": 25339836.0, + "step": 991 + }, + { + "epoch": 0.10893916099275203, + "grad_norm": 2.3071534633636475, + "learning_rate": 5.439077936333699e-07, + "loss": 1.0041, + "mean_token_accuracy": 0.6971914768218994, + "num_tokens": 25365294.0, + "step": 992 + }, + { + "epoch": 0.10904897869536569, + "grad_norm": 2.5616559982299805, + "learning_rate": 5.44456641053787e-07, + "loss": 1.0726, + "mean_token_accuracy": 0.6926769018173218, + "num_tokens": 25387381.0, + "step": 993 + }, + { + "epoch": 0.10915879639797936, + "grad_norm": 2.0409982204437256, + "learning_rate": 5.450054884742042e-07, + "loss": 1.1305, + "mean_token_accuracy": 0.6612395644187927, + "num_tokens": 25420203.0, + "step": 994 + }, + { + "epoch": 0.10926861410059302, + "grad_norm": 2.508613109588623, + "learning_rate": 5.455543358946213e-07, + "loss": 0.9986, + "mean_token_accuracy": 0.7031484842300415, + "num_tokens": 25442549.0, + "step": 995 + }, + { + "epoch": 0.10937843180320668, + "grad_norm": 2.486121654510498, + "learning_rate": 5.461031833150384e-07, + "loss": 0.9732, + "mean_token_accuracy": 0.7064170837402344, + "num_tokens": 25465131.0, + "step": 996 + }, + { + "epoch": 0.10948824950582034, + "grad_norm": 2.4710686206817627, + "learning_rate": 5.466520307354556e-07, + "loss": 1.0522, + "mean_token_accuracy": 0.6868271827697754, + "num_tokens": 25489436.0, + "step": 997 + }, + { + "epoch": 0.109598067208434, + "grad_norm": 2.460122585296631, + "learning_rate": 5.472008781558726e-07, + "loss": 1.0342, + "mean_token_accuracy": 0.6934085488319397, + "num_tokens": 25511973.0, + "step": 998 + }, + { + "epoch": 0.10970788491104766, + "grad_norm": 2.4071171283721924, + "learning_rate": 5.477497255762897e-07, + "loss": 1.1717, + "mean_token_accuracy": 0.6702781915664673, + "num_tokens": 25539274.0, + "step": 999 + }, + { + "epoch": 0.10981770261366132, + "grad_norm": 2.1122219562530518, + "learning_rate": 5.482985729967069e-07, + "loss": 0.9899, + "mean_token_accuracy": 0.6998860239982605, + "num_tokens": 25566667.0, + "step": 1000 + }, + { + "epoch": 0.10992752031627498, + "grad_norm": 2.363929510116577, + "learning_rate": 5.48847420417124e-07, + "loss": 1.0427, + "mean_token_accuracy": 0.6855452656745911, + "num_tokens": 25592877.0, + "step": 1001 + }, + { + "epoch": 0.11003733801888864, + "grad_norm": 2.159090280532837, + "learning_rate": 5.493962678375411e-07, + "loss": 1.1567, + "mean_token_accuracy": 0.6642252206802368, + "num_tokens": 25623109.0, + "step": 1002 + }, + { + "epoch": 0.11014715572150231, + "grad_norm": 2.2549362182617188, + "learning_rate": 5.499451152579583e-07, + "loss": 1.0923, + "mean_token_accuracy": 0.6809996366500854, + "num_tokens": 25650814.0, + "step": 1003 + }, + { + "epoch": 0.11025697342411597, + "grad_norm": 2.242125988006592, + "learning_rate": 5.504939626783753e-07, + "loss": 1.0137, + "mean_token_accuracy": 0.6936848163604736, + "num_tokens": 25677842.0, + "step": 1004 + }, + { + "epoch": 0.11036679112672963, + "grad_norm": 2.3709583282470703, + "learning_rate": 5.510428100987925e-07, + "loss": 1.1555, + "mean_token_accuracy": 0.656548261642456, + "num_tokens": 25703110.0, + "step": 1005 + }, + { + "epoch": 0.11047660882934329, + "grad_norm": 2.755208730697632, + "learning_rate": 5.515916575192097e-07, + "loss": 0.9923, + "mean_token_accuracy": 0.7013557553291321, + "num_tokens": 25721247.0, + "step": 1006 + }, + { + "epoch": 0.11058642653195695, + "grad_norm": 2.3578453063964844, + "learning_rate": 5.521405049396267e-07, + "loss": 1.0742, + "mean_token_accuracy": 0.6858857870101929, + "num_tokens": 25746627.0, + "step": 1007 + }, + { + "epoch": 0.11069624423457061, + "grad_norm": 2.395209312438965, + "learning_rate": 5.526893523600439e-07, + "loss": 0.9657, + "mean_token_accuracy": 0.7065850496292114, + "num_tokens": 25769925.0, + "step": 1008 + }, + { + "epoch": 0.11080606193718427, + "grad_norm": 2.851872205734253, + "learning_rate": 5.532381997804611e-07, + "loss": 0.9584, + "mean_token_accuracy": 0.7083455324172974, + "num_tokens": 25787252.0, + "step": 1009 + }, + { + "epoch": 0.11091587963979793, + "grad_norm": 2.357118844985962, + "learning_rate": 5.537870472008781e-07, + "loss": 1.0889, + "mean_token_accuracy": 0.6774642467498779, + "num_tokens": 25812124.0, + "step": 1010 + }, + { + "epoch": 0.11102569734241159, + "grad_norm": 2.4497618675231934, + "learning_rate": 5.543358946212953e-07, + "loss": 1.0044, + "mean_token_accuracy": 0.7045637369155884, + "num_tokens": 25834718.0, + "step": 1011 + }, + { + "epoch": 0.11113551504502525, + "grad_norm": 2.3211774826049805, + "learning_rate": 5.548847420417125e-07, + "loss": 1.0974, + "mean_token_accuracy": 0.6719145774841309, + "num_tokens": 25861382.0, + "step": 1012 + }, + { + "epoch": 0.11124533274763893, + "grad_norm": 2.0416464805603027, + "learning_rate": 5.554335894621295e-07, + "loss": 1.0347, + "mean_token_accuracy": 0.692787766456604, + "num_tokens": 25893429.0, + "step": 1013 + }, + { + "epoch": 0.11135515045025259, + "grad_norm": 2.1931262016296387, + "learning_rate": 5.559824368825467e-07, + "loss": 1.1286, + "mean_token_accuracy": 0.6704572439193726, + "num_tokens": 25923080.0, + "step": 1014 + }, + { + "epoch": 0.11146496815286625, + "grad_norm": 2.2871439456939697, + "learning_rate": 5.565312843029637e-07, + "loss": 1.0869, + "mean_token_accuracy": 0.6846202611923218, + "num_tokens": 25949141.0, + "step": 1015 + }, + { + "epoch": 0.1115747858554799, + "grad_norm": 2.662323236465454, + "learning_rate": 5.570801317233809e-07, + "loss": 0.9943, + "mean_token_accuracy": 0.701807975769043, + "num_tokens": 25968003.0, + "step": 1016 + }, + { + "epoch": 0.11168460355809356, + "grad_norm": 2.365321397781372, + "learning_rate": 5.57628979143798e-07, + "loss": 1.059, + "mean_token_accuracy": 0.6865823268890381, + "num_tokens": 25994706.0, + "step": 1017 + }, + { + "epoch": 0.11179442126070722, + "grad_norm": 2.4106507301330566, + "learning_rate": 5.58177826564215e-07, + "loss": 0.9944, + "mean_token_accuracy": 0.6939229965209961, + "num_tokens": 26018478.0, + "step": 1018 + }, + { + "epoch": 0.11190423896332088, + "grad_norm": 2.3687567710876465, + "learning_rate": 5.587266739846322e-07, + "loss": 0.9896, + "mean_token_accuracy": 0.701260507106781, + "num_tokens": 26042770.0, + "step": 1019 + }, + { + "epoch": 0.11201405666593454, + "grad_norm": 2.208139181137085, + "learning_rate": 5.592755214050494e-07, + "loss": 1.1307, + "mean_token_accuracy": 0.6652946472167969, + "num_tokens": 26071287.0, + "step": 1020 + }, + { + "epoch": 0.1121238743685482, + "grad_norm": 2.3043174743652344, + "learning_rate": 5.598243688254664e-07, + "loss": 1.0917, + "mean_token_accuracy": 0.6839613914489746, + "num_tokens": 26097687.0, + "step": 1021 + }, + { + "epoch": 0.11223369207116188, + "grad_norm": 2.7185606956481934, + "learning_rate": 5.603732162458836e-07, + "loss": 1.0449, + "mean_token_accuracy": 0.6900323033332825, + "num_tokens": 26117594.0, + "step": 1022 + }, + { + "epoch": 0.11234350977377554, + "grad_norm": 2.1774322986602783, + "learning_rate": 5.609220636663008e-07, + "loss": 0.9639, + "mean_token_accuracy": 0.709465503692627, + "num_tokens": 26144223.0, + "step": 1023 + }, + { + "epoch": 0.1124533274763892, + "grad_norm": 2.3149781227111816, + "learning_rate": 5.614709110867178e-07, + "loss": 1.0805, + "mean_token_accuracy": 0.6769900918006897, + "num_tokens": 26169962.0, + "step": 1024 + }, + { + "epoch": 0.11256314517900286, + "grad_norm": 2.4132275581359863, + "learning_rate": 5.62019758507135e-07, + "loss": 1.0328, + "mean_token_accuracy": 0.6910185813903809, + "num_tokens": 26194680.0, + "step": 1025 + }, + { + "epoch": 0.11267296288161652, + "grad_norm": 2.2334606647491455, + "learning_rate": 5.625686059275521e-07, + "loss": 1.0832, + "mean_token_accuracy": 0.6787500381469727, + "num_tokens": 26222438.0, + "step": 1026 + }, + { + "epoch": 0.11278278058423018, + "grad_norm": 2.4306066036224365, + "learning_rate": 5.631174533479692e-07, + "loss": 1.0948, + "mean_token_accuracy": 0.6828962564468384, + "num_tokens": 26246299.0, + "step": 1027 + }, + { + "epoch": 0.11289259828684384, + "grad_norm": 2.4965577125549316, + "learning_rate": 5.636663007683864e-07, + "loss": 1.1224, + "mean_token_accuracy": 0.6669524312019348, + "num_tokens": 26269139.0, + "step": 1028 + }, + { + "epoch": 0.1130024159894575, + "grad_norm": 2.3189871311187744, + "learning_rate": 5.642151481888035e-07, + "loss": 1.1773, + "mean_token_accuracy": 0.6568245887756348, + "num_tokens": 26297629.0, + "step": 1029 + }, + { + "epoch": 0.11311223369207116, + "grad_norm": 2.223025321960449, + "learning_rate": 5.647639956092206e-07, + "loss": 1.0875, + "mean_token_accuracy": 0.6772985458374023, + "num_tokens": 26324921.0, + "step": 1030 + }, + { + "epoch": 0.11322205139468482, + "grad_norm": 2.552905559539795, + "learning_rate": 5.653128430296378e-07, + "loss": 1.032, + "mean_token_accuracy": 0.6991462111473083, + "num_tokens": 26347459.0, + "step": 1031 + }, + { + "epoch": 0.11333186909729849, + "grad_norm": 2.3205947875976562, + "learning_rate": 5.658616904500549e-07, + "loss": 1.0798, + "mean_token_accuracy": 0.6777387261390686, + "num_tokens": 26374503.0, + "step": 1032 + }, + { + "epoch": 0.11344168679991215, + "grad_norm": 2.123811960220337, + "learning_rate": 5.66410537870472e-07, + "loss": 1.127, + "mean_token_accuracy": 0.6717268228530884, + "num_tokens": 26405366.0, + "step": 1033 + }, + { + "epoch": 0.11355150450252581, + "grad_norm": 2.140223979949951, + "learning_rate": 5.669593852908892e-07, + "loss": 1.1471, + "mean_token_accuracy": 0.6628724932670593, + "num_tokens": 26434260.0, + "step": 1034 + }, + { + "epoch": 0.11366132220513947, + "grad_norm": 2.3844563961029053, + "learning_rate": 5.675082327113063e-07, + "loss": 1.0105, + "mean_token_accuracy": 0.6952896118164062, + "num_tokens": 26458066.0, + "step": 1035 + }, + { + "epoch": 0.11377113990775313, + "grad_norm": 2.644216775894165, + "learning_rate": 5.680570801317233e-07, + "loss": 1.0784, + "mean_token_accuracy": 0.6927073001861572, + "num_tokens": 26479264.0, + "step": 1036 + }, + { + "epoch": 0.11388095761036679, + "grad_norm": 2.3328073024749756, + "learning_rate": 5.686059275521404e-07, + "loss": 1.0323, + "mean_token_accuracy": 0.6919130086898804, + "num_tokens": 26504746.0, + "step": 1037 + }, + { + "epoch": 0.11399077531298045, + "grad_norm": 2.5497241020202637, + "learning_rate": 5.691547749725576e-07, + "loss": 1.036, + "mean_token_accuracy": 0.6910390853881836, + "num_tokens": 26527820.0, + "step": 1038 + }, + { + "epoch": 0.11410059301559411, + "grad_norm": 2.3317747116088867, + "learning_rate": 5.697036223929747e-07, + "loss": 1.1601, + "mean_token_accuracy": 0.6595733165740967, + "num_tokens": 26555090.0, + "step": 1039 + }, + { + "epoch": 0.11421041071820777, + "grad_norm": 2.2607038021087646, + "learning_rate": 5.702524698133918e-07, + "loss": 1.0074, + "mean_token_accuracy": 0.6991442441940308, + "num_tokens": 26580017.0, + "step": 1040 + }, + { + "epoch": 0.11432022842082144, + "grad_norm": 2.128394842147827, + "learning_rate": 5.70801317233809e-07, + "loss": 1.1746, + "mean_token_accuracy": 0.6489684581756592, + "num_tokens": 26613083.0, + "step": 1041 + }, + { + "epoch": 0.1144300461234351, + "grad_norm": 2.2163493633270264, + "learning_rate": 5.713501646542261e-07, + "loss": 1.0942, + "mean_token_accuracy": 0.6710448265075684, + "num_tokens": 26641003.0, + "step": 1042 + }, + { + "epoch": 0.11453986382604876, + "grad_norm": 1.9805797338485718, + "learning_rate": 5.718990120746432e-07, + "loss": 1.0862, + "mean_token_accuracy": 0.6758020520210266, + "num_tokens": 26673407.0, + "step": 1043 + }, + { + "epoch": 0.11464968152866242, + "grad_norm": 2.6884334087371826, + "learning_rate": 5.724478594950604e-07, + "loss": 0.9454, + "mean_token_accuracy": 0.710923969745636, + "num_tokens": 26691891.0, + "step": 1044 + }, + { + "epoch": 0.11475949923127608, + "grad_norm": 2.7043957710266113, + "learning_rate": 5.729967069154775e-07, + "loss": 1.1323, + "mean_token_accuracy": 0.6656020879745483, + "num_tokens": 26715320.0, + "step": 1045 + }, + { + "epoch": 0.11486931693388974, + "grad_norm": 2.322812557220459, + "learning_rate": 5.735455543358946e-07, + "loss": 0.9982, + "mean_token_accuracy": 0.7005172967910767, + "num_tokens": 26740019.0, + "step": 1046 + }, + { + "epoch": 0.1149791346365034, + "grad_norm": 2.2204442024230957, + "learning_rate": 5.740944017563118e-07, + "loss": 1.0267, + "mean_token_accuracy": 0.6883218288421631, + "num_tokens": 26766894.0, + "step": 1047 + }, + { + "epoch": 0.11508895233911706, + "grad_norm": 2.1310062408447266, + "learning_rate": 5.746432491767288e-07, + "loss": 1.1373, + "mean_token_accuracy": 0.6618829965591431, + "num_tokens": 26795939.0, + "step": 1048 + }, + { + "epoch": 0.11519877004173072, + "grad_norm": 2.2261393070220947, + "learning_rate": 5.75192096597146e-07, + "loss": 1.0957, + "mean_token_accuracy": 0.672803521156311, + "num_tokens": 26825463.0, + "step": 1049 + }, + { + "epoch": 0.11530858774434438, + "grad_norm": 2.299870491027832, + "learning_rate": 5.757409440175632e-07, + "loss": 1.025, + "mean_token_accuracy": 0.6887872219085693, + "num_tokens": 26851753.0, + "step": 1050 + }, + { + "epoch": 0.11541840544695806, + "grad_norm": 2.3332903385162354, + "learning_rate": 5.762897914379802e-07, + "loss": 1.0583, + "mean_token_accuracy": 0.6953582763671875, + "num_tokens": 26877676.0, + "step": 1051 + }, + { + "epoch": 0.11552822314957172, + "grad_norm": 2.457249164581299, + "learning_rate": 5.768386388583974e-07, + "loss": 1.0051, + "mean_token_accuracy": 0.7012569904327393, + "num_tokens": 26902904.0, + "step": 1052 + }, + { + "epoch": 0.11563804085218538, + "grad_norm": 2.418948173522949, + "learning_rate": 5.773874862788145e-07, + "loss": 1.1724, + "mean_token_accuracy": 0.6605175733566284, + "num_tokens": 26928902.0, + "step": 1053 + }, + { + "epoch": 0.11574785855479904, + "grad_norm": 2.196483612060547, + "learning_rate": 5.779363336992316e-07, + "loss": 1.1103, + "mean_token_accuracy": 0.6638792753219604, + "num_tokens": 26958399.0, + "step": 1054 + }, + { + "epoch": 0.1158576762574127, + "grad_norm": 2.2111892700195312, + "learning_rate": 5.784851811196487e-07, + "loss": 1.1009, + "mean_token_accuracy": 0.6720449924468994, + "num_tokens": 26988068.0, + "step": 1055 + }, + { + "epoch": 0.11596749396002635, + "grad_norm": 1.9413789510726929, + "learning_rate": 5.790340285400658e-07, + "loss": 1.0728, + "mean_token_accuracy": 0.6872462034225464, + "num_tokens": 27021978.0, + "step": 1056 + }, + { + "epoch": 0.11607731166264001, + "grad_norm": 2.5438318252563477, + "learning_rate": 5.795828759604829e-07, + "loss": 1.1897, + "mean_token_accuracy": 0.6555063724517822, + "num_tokens": 27044318.0, + "step": 1057 + }, + { + "epoch": 0.11618712936525367, + "grad_norm": 2.3024258613586426, + "learning_rate": 5.801317233809001e-07, + "loss": 1.1075, + "mean_token_accuracy": 0.6854103207588196, + "num_tokens": 27070007.0, + "step": 1058 + }, + { + "epoch": 0.11629694706786733, + "grad_norm": 2.633547306060791, + "learning_rate": 5.806805708013171e-07, + "loss": 1.0716, + "mean_token_accuracy": 0.6826178431510925, + "num_tokens": 27092263.0, + "step": 1059 + }, + { + "epoch": 0.11640676477048101, + "grad_norm": 2.382767677307129, + "learning_rate": 5.812294182217343e-07, + "loss": 1.0508, + "mean_token_accuracy": 0.689628541469574, + "num_tokens": 27118373.0, + "step": 1060 + }, + { + "epoch": 0.11651658247309467, + "grad_norm": 2.3107755184173584, + "learning_rate": 5.817782656421515e-07, + "loss": 1.062, + "mean_token_accuracy": 0.6881730556488037, + "num_tokens": 27143402.0, + "step": 1061 + }, + { + "epoch": 0.11662640017570833, + "grad_norm": 2.617924928665161, + "learning_rate": 5.823271130625685e-07, + "loss": 1.0491, + "mean_token_accuracy": 0.6887562274932861, + "num_tokens": 27164756.0, + "step": 1062 + }, + { + "epoch": 0.11673621787832199, + "grad_norm": 2.405938148498535, + "learning_rate": 5.828759604829857e-07, + "loss": 1.0331, + "mean_token_accuracy": 0.6914063096046448, + "num_tokens": 27189857.0, + "step": 1063 + }, + { + "epoch": 0.11684603558093565, + "grad_norm": 2.4180846214294434, + "learning_rate": 5.834248079034029e-07, + "loss": 1.0648, + "mean_token_accuracy": 0.6859700679779053, + "num_tokens": 27214493.0, + "step": 1064 + }, + { + "epoch": 0.11695585328354931, + "grad_norm": 2.1460659503936768, + "learning_rate": 5.839736553238199e-07, + "loss": 1.1513, + "mean_token_accuracy": 0.6761540770530701, + "num_tokens": 27245433.0, + "step": 1065 + }, + { + "epoch": 0.11706567098616297, + "grad_norm": 2.1041407585144043, + "learning_rate": 5.845225027442371e-07, + "loss": 1.1528, + "mean_token_accuracy": 0.658236026763916, + "num_tokens": 27275538.0, + "step": 1066 + }, + { + "epoch": 0.11717548868877663, + "grad_norm": 2.5851709842681885, + "learning_rate": 5.850713501646543e-07, + "loss": 1.0104, + "mean_token_accuracy": 0.6959775686264038, + "num_tokens": 27298092.0, + "step": 1067 + }, + { + "epoch": 0.11728530639139029, + "grad_norm": 2.1364574432373047, + "learning_rate": 5.856201975850713e-07, + "loss": 1.0998, + "mean_token_accuracy": 0.6790059208869934, + "num_tokens": 27327598.0, + "step": 1068 + }, + { + "epoch": 0.11739512409400395, + "grad_norm": 2.074652671813965, + "learning_rate": 5.861690450054885e-07, + "loss": 1.1596, + "mean_token_accuracy": 0.6566696166992188, + "num_tokens": 27359945.0, + "step": 1069 + }, + { + "epoch": 0.11750494179661762, + "grad_norm": 2.2638912200927734, + "learning_rate": 5.867178924259056e-07, + "loss": 1.0395, + "mean_token_accuracy": 0.6953232884407043, + "num_tokens": 27386144.0, + "step": 1070 + }, + { + "epoch": 0.11761475949923128, + "grad_norm": 2.5813519954681396, + "learning_rate": 5.872667398463227e-07, + "loss": 1.0115, + "mean_token_accuracy": 0.6946108341217041, + "num_tokens": 27406574.0, + "step": 1071 + }, + { + "epoch": 0.11772457720184494, + "grad_norm": 2.1815688610076904, + "learning_rate": 5.878155872667399e-07, + "loss": 1.0926, + "mean_token_accuracy": 0.6769355535507202, + "num_tokens": 27435083.0, + "step": 1072 + }, + { + "epoch": 0.1178343949044586, + "grad_norm": 2.671551465988159, + "learning_rate": 5.88364434687157e-07, + "loss": 1.0455, + "mean_token_accuracy": 0.6892681121826172, + "num_tokens": 27455973.0, + "step": 1073 + }, + { + "epoch": 0.11794421260707226, + "grad_norm": 2.545584201812744, + "learning_rate": 5.88913282107574e-07, + "loss": 1.1503, + "mean_token_accuracy": 0.6763935685157776, + "num_tokens": 27478414.0, + "step": 1074 + }, + { + "epoch": 0.11805403030968592, + "grad_norm": 2.2146072387695312, + "learning_rate": 5.894621295279912e-07, + "loss": 1.0381, + "mean_token_accuracy": 0.69078528881073, + "num_tokens": 27506242.0, + "step": 1075 + }, + { + "epoch": 0.11816384801229958, + "grad_norm": 2.138453483581543, + "learning_rate": 5.900109769484083e-07, + "loss": 1.0638, + "mean_token_accuracy": 0.681011438369751, + "num_tokens": 27534161.0, + "step": 1076 + }, + { + "epoch": 0.11827366571491324, + "grad_norm": 2.3485846519470215, + "learning_rate": 5.905598243688254e-07, + "loss": 1.0677, + "mean_token_accuracy": 0.678607702255249, + "num_tokens": 27560079.0, + "step": 1077 + }, + { + "epoch": 0.1183834834175269, + "grad_norm": 2.6422805786132812, + "learning_rate": 5.911086717892426e-07, + "loss": 1.1539, + "mean_token_accuracy": 0.6665201187133789, + "num_tokens": 27583493.0, + "step": 1078 + }, + { + "epoch": 0.11849330112014057, + "grad_norm": 2.447913408279419, + "learning_rate": 5.916575192096597e-07, + "loss": 1.1505, + "mean_token_accuracy": 0.6597746014595032, + "num_tokens": 27608366.0, + "step": 1079 + }, + { + "epoch": 0.11860311882275423, + "grad_norm": 2.057044267654419, + "learning_rate": 5.922063666300768e-07, + "loss": 1.0826, + "mean_token_accuracy": 0.6759148836135864, + "num_tokens": 27640846.0, + "step": 1080 + }, + { + "epoch": 0.1187129365253679, + "grad_norm": 2.267632484436035, + "learning_rate": 5.927552140504939e-07, + "loss": 1.0185, + "mean_token_accuracy": 0.698656439781189, + "num_tokens": 27665982.0, + "step": 1081 + }, + { + "epoch": 0.11882275422798155, + "grad_norm": 2.5414044857025146, + "learning_rate": 5.93304061470911e-07, + "loss": 1.0683, + "mean_token_accuracy": 0.6803206205368042, + "num_tokens": 27688762.0, + "step": 1082 + }, + { + "epoch": 0.11893257193059521, + "grad_norm": 2.4636361598968506, + "learning_rate": 5.938529088913282e-07, + "loss": 1.0863, + "mean_token_accuracy": 0.6826382875442505, + "num_tokens": 27712344.0, + "step": 1083 + }, + { + "epoch": 0.11904238963320887, + "grad_norm": 2.2675468921661377, + "learning_rate": 5.944017563117453e-07, + "loss": 1.0959, + "mean_token_accuracy": 0.6778266429901123, + "num_tokens": 27739671.0, + "step": 1084 + }, + { + "epoch": 0.11915220733582253, + "grad_norm": 2.5186445713043213, + "learning_rate": 5.949506037321624e-07, + "loss": 1.1446, + "mean_token_accuracy": 0.6723453402519226, + "num_tokens": 27763677.0, + "step": 1085 + }, + { + "epoch": 0.11926202503843619, + "grad_norm": 2.4534785747528076, + "learning_rate": 5.954994511525796e-07, + "loss": 1.0756, + "mean_token_accuracy": 0.6912261843681335, + "num_tokens": 27787531.0, + "step": 1086 + }, + { + "epoch": 0.11937184274104985, + "grad_norm": 2.2552216053009033, + "learning_rate": 5.960482985729967e-07, + "loss": 1.1205, + "mean_token_accuracy": 0.6678779721260071, + "num_tokens": 27814782.0, + "step": 1087 + }, + { + "epoch": 0.11948166044366351, + "grad_norm": 2.2859060764312744, + "learning_rate": 5.965971459934138e-07, + "loss": 1.088, + "mean_token_accuracy": 0.6732913255691528, + "num_tokens": 27841340.0, + "step": 1088 + }, + { + "epoch": 0.11959147814627719, + "grad_norm": 2.596658706665039, + "learning_rate": 5.971459934138309e-07, + "loss": 0.9892, + "mean_token_accuracy": 0.7014905214309692, + "num_tokens": 27862855.0, + "step": 1089 + }, + { + "epoch": 0.11970129584889085, + "grad_norm": 2.2058653831481934, + "learning_rate": 5.976948408342481e-07, + "loss": 1.0836, + "mean_token_accuracy": 0.6771237850189209, + "num_tokens": 27892009.0, + "step": 1090 + }, + { + "epoch": 0.1198111135515045, + "grad_norm": 2.5108675956726074, + "learning_rate": 5.982436882546652e-07, + "loss": 0.9922, + "mean_token_accuracy": 0.6998687982559204, + "num_tokens": 27914884.0, + "step": 1091 + }, + { + "epoch": 0.11992093125411817, + "grad_norm": 2.2816200256347656, + "learning_rate": 5.987925356750822e-07, + "loss": 1.1258, + "mean_token_accuracy": 0.6682695150375366, + "num_tokens": 27944044.0, + "step": 1092 + }, + { + "epoch": 0.12003074895673183, + "grad_norm": 2.3103127479553223, + "learning_rate": 5.993413830954994e-07, + "loss": 1.0229, + "mean_token_accuracy": 0.697665810585022, + "num_tokens": 27970407.0, + "step": 1093 + }, + { + "epoch": 0.12014056665934549, + "grad_norm": 2.305299758911133, + "learning_rate": 5.998902305159165e-07, + "loss": 1.0671, + "mean_token_accuracy": 0.6924338936805725, + "num_tokens": 27993930.0, + "step": 1094 + }, + { + "epoch": 0.12025038436195915, + "grad_norm": 2.2564141750335693, + "learning_rate": 6.004390779363336e-07, + "loss": 1.0897, + "mean_token_accuracy": 0.6798684597015381, + "num_tokens": 28023469.0, + "step": 1095 + }, + { + "epoch": 0.1203602020645728, + "grad_norm": 2.5243709087371826, + "learning_rate": 6.009879253567508e-07, + "loss": 1.044, + "mean_token_accuracy": 0.696258544921875, + "num_tokens": 28047250.0, + "step": 1096 + }, + { + "epoch": 0.12047001976718646, + "grad_norm": 2.4145662784576416, + "learning_rate": 6.015367727771679e-07, + "loss": 1.0705, + "mean_token_accuracy": 0.6936265230178833, + "num_tokens": 28072120.0, + "step": 1097 + }, + { + "epoch": 0.12057983746980014, + "grad_norm": 2.37949275970459, + "learning_rate": 6.02085620197585e-07, + "loss": 1.0446, + "mean_token_accuracy": 0.6917973160743713, + "num_tokens": 28097514.0, + "step": 1098 + }, + { + "epoch": 0.1206896551724138, + "grad_norm": 2.205156087875366, + "learning_rate": 6.026344676180022e-07, + "loss": 1.1211, + "mean_token_accuracy": 0.6786275506019592, + "num_tokens": 28126049.0, + "step": 1099 + }, + { + "epoch": 0.12079947287502746, + "grad_norm": 2.2819674015045166, + "learning_rate": 6.031833150384192e-07, + "loss": 1.0665, + "mean_token_accuracy": 0.6897584199905396, + "num_tokens": 28152285.0, + "step": 1100 + }, + { + "epoch": 0.12090929057764112, + "grad_norm": 2.3862063884735107, + "learning_rate": 6.037321624588364e-07, + "loss": 1.1369, + "mean_token_accuracy": 0.6697108745574951, + "num_tokens": 28178049.0, + "step": 1101 + }, + { + "epoch": 0.12101910828025478, + "grad_norm": 2.2956461906433105, + "learning_rate": 6.042810098792536e-07, + "loss": 1.0783, + "mean_token_accuracy": 0.6836432218551636, + "num_tokens": 28204687.0, + "step": 1102 + }, + { + "epoch": 0.12112892598286844, + "grad_norm": 2.67421293258667, + "learning_rate": 6.048298572996706e-07, + "loss": 0.9984, + "mean_token_accuracy": 0.7007089257240295, + "num_tokens": 28224762.0, + "step": 1103 + }, + { + "epoch": 0.1212387436854821, + "grad_norm": 2.490821599960327, + "learning_rate": 6.053787047200878e-07, + "loss": 1.0324, + "mean_token_accuracy": 0.6928340792655945, + "num_tokens": 28247070.0, + "step": 1104 + }, + { + "epoch": 0.12134856138809576, + "grad_norm": 2.6959738731384277, + "learning_rate": 6.05927552140505e-07, + "loss": 0.9631, + "mean_token_accuracy": 0.7175220251083374, + "num_tokens": 28266012.0, + "step": 1105 + }, + { + "epoch": 0.12145837909070942, + "grad_norm": 2.3387365341186523, + "learning_rate": 6.06476399560922e-07, + "loss": 1.0576, + "mean_token_accuracy": 0.684715747833252, + "num_tokens": 28290906.0, + "step": 1106 + }, + { + "epoch": 0.12156819679332308, + "grad_norm": 2.5674428939819336, + "learning_rate": 6.070252469813392e-07, + "loss": 1.0543, + "mean_token_accuracy": 0.6807563900947571, + "num_tokens": 28313733.0, + "step": 1107 + }, + { + "epoch": 0.12167801449593675, + "grad_norm": 2.4669995307922363, + "learning_rate": 6.075740944017564e-07, + "loss": 1.1068, + "mean_token_accuracy": 0.6812207102775574, + "num_tokens": 28339761.0, + "step": 1108 + }, + { + "epoch": 0.12178783219855041, + "grad_norm": 2.331376075744629, + "learning_rate": 6.081229418221734e-07, + "loss": 1.0523, + "mean_token_accuracy": 0.6854852437973022, + "num_tokens": 28365237.0, + "step": 1109 + }, + { + "epoch": 0.12189764990116407, + "grad_norm": 2.4751551151275635, + "learning_rate": 6.086717892425906e-07, + "loss": 1.0731, + "mean_token_accuracy": 0.6853611469268799, + "num_tokens": 28391592.0, + "step": 1110 + }, + { + "epoch": 0.12200746760377773, + "grad_norm": 2.3469386100769043, + "learning_rate": 6.092206366630076e-07, + "loss": 1.0219, + "mean_token_accuracy": 0.6969906687736511, + "num_tokens": 28413977.0, + "step": 1111 + }, + { + "epoch": 0.12211728530639139, + "grad_norm": 2.406385660171509, + "learning_rate": 6.097694840834247e-07, + "loss": 1.0119, + "mean_token_accuracy": 0.6932421326637268, + "num_tokens": 28437183.0, + "step": 1112 + }, + { + "epoch": 0.12222710300900505, + "grad_norm": 2.207448959350586, + "learning_rate": 6.103183315038419e-07, + "loss": 1.0377, + "mean_token_accuracy": 0.7023807764053345, + "num_tokens": 28463307.0, + "step": 1113 + }, + { + "epoch": 0.12233692071161871, + "grad_norm": 2.3642754554748535, + "learning_rate": 6.10867178924259e-07, + "loss": 1.1001, + "mean_token_accuracy": 0.6765137314796448, + "num_tokens": 28489535.0, + "step": 1114 + }, + { + "epoch": 0.12244673841423237, + "grad_norm": 2.575465679168701, + "learning_rate": 6.114160263446761e-07, + "loss": 0.9881, + "mean_token_accuracy": 0.7051224708557129, + "num_tokens": 28510599.0, + "step": 1115 + }, + { + "epoch": 0.12255655611684603, + "grad_norm": 2.223978042602539, + "learning_rate": 6.119648737650933e-07, + "loss": 1.0267, + "mean_token_accuracy": 0.6966384649276733, + "num_tokens": 28537199.0, + "step": 1116 + }, + { + "epoch": 0.1226663738194597, + "grad_norm": 2.4306676387786865, + "learning_rate": 6.125137211855103e-07, + "loss": 1.003, + "mean_token_accuracy": 0.6935455203056335, + "num_tokens": 28560355.0, + "step": 1117 + }, + { + "epoch": 0.12277619152207336, + "grad_norm": 2.7614166736602783, + "learning_rate": 6.130625686059275e-07, + "loss": 1.1202, + "mean_token_accuracy": 0.6688264608383179, + "num_tokens": 28581888.0, + "step": 1118 + }, + { + "epoch": 0.12288600922468702, + "grad_norm": 2.2378556728363037, + "learning_rate": 6.136114160263447e-07, + "loss": 0.9893, + "mean_token_accuracy": 0.7020886540412903, + "num_tokens": 28606420.0, + "step": 1119 + }, + { + "epoch": 0.12299582692730068, + "grad_norm": 2.237828493118286, + "learning_rate": 6.141602634467617e-07, + "loss": 1.0297, + "mean_token_accuracy": 0.6938437223434448, + "num_tokens": 28632836.0, + "step": 1120 + }, + { + "epoch": 0.12310564462991434, + "grad_norm": 2.1348485946655273, + "learning_rate": 6.147091108671789e-07, + "loss": 1.0865, + "mean_token_accuracy": 0.6795476675033569, + "num_tokens": 28660659.0, + "step": 1121 + }, + { + "epoch": 0.123215462332528, + "grad_norm": 2.4643044471740723, + "learning_rate": 6.15257958287596e-07, + "loss": 1.0327, + "mean_token_accuracy": 0.6905616521835327, + "num_tokens": 28684242.0, + "step": 1122 + }, + { + "epoch": 0.12332528003514166, + "grad_norm": 2.397108554840088, + "learning_rate": 6.158068057080131e-07, + "loss": 1.0619, + "mean_token_accuracy": 0.6760594844818115, + "num_tokens": 28710461.0, + "step": 1123 + }, + { + "epoch": 0.12343509773775532, + "grad_norm": 2.3774807453155518, + "learning_rate": 6.163556531284303e-07, + "loss": 1.0365, + "mean_token_accuracy": 0.6852315664291382, + "num_tokens": 28735199.0, + "step": 1124 + }, + { + "epoch": 0.12354491544036898, + "grad_norm": 2.2008578777313232, + "learning_rate": 6.169045005488474e-07, + "loss": 1.0493, + "mean_token_accuracy": 0.6864997148513794, + "num_tokens": 28762490.0, + "step": 1125 + }, + { + "epoch": 0.12365473314298264, + "grad_norm": 2.3524701595306396, + "learning_rate": 6.174533479692645e-07, + "loss": 0.9955, + "mean_token_accuracy": 0.7036919593811035, + "num_tokens": 28785542.0, + "step": 1126 + }, + { + "epoch": 0.12376455084559632, + "grad_norm": 2.272047758102417, + "learning_rate": 6.180021953896817e-07, + "loss": 1.0692, + "mean_token_accuracy": 0.681554913520813, + "num_tokens": 28811622.0, + "step": 1127 + }, + { + "epoch": 0.12387436854820998, + "grad_norm": 2.1488864421844482, + "learning_rate": 6.185510428100988e-07, + "loss": 1.1362, + "mean_token_accuracy": 0.6602373123168945, + "num_tokens": 28841459.0, + "step": 1128 + }, + { + "epoch": 0.12398418625082364, + "grad_norm": 2.139072895050049, + "learning_rate": 6.190998902305159e-07, + "loss": 1.0469, + "mean_token_accuracy": 0.6922773122787476, + "num_tokens": 28869431.0, + "step": 1129 + }, + { + "epoch": 0.1240940039534373, + "grad_norm": 2.3254218101501465, + "learning_rate": 6.196487376509331e-07, + "loss": 1.0213, + "mean_token_accuracy": 0.6949234008789062, + "num_tokens": 28892444.0, + "step": 1130 + }, + { + "epoch": 0.12420382165605096, + "grad_norm": 2.362672805786133, + "learning_rate": 6.201975850713501e-07, + "loss": 1.1249, + "mean_token_accuracy": 0.6648603081703186, + "num_tokens": 28916728.0, + "step": 1131 + }, + { + "epoch": 0.12431363935866462, + "grad_norm": 2.5285308361053467, + "learning_rate": 6.207464324917672e-07, + "loss": 1.0871, + "mean_token_accuracy": 0.6850054264068604, + "num_tokens": 28942012.0, + "step": 1132 + }, + { + "epoch": 0.12442345706127828, + "grad_norm": 2.371985912322998, + "learning_rate": 6.212952799121843e-07, + "loss": 1.0534, + "mean_token_accuracy": 0.6854907870292664, + "num_tokens": 28965961.0, + "step": 1133 + }, + { + "epoch": 0.12453327476389194, + "grad_norm": 2.4247982501983643, + "learning_rate": 6.218441273326015e-07, + "loss": 1.0641, + "mean_token_accuracy": 0.6888430118560791, + "num_tokens": 28990318.0, + "step": 1134 + }, + { + "epoch": 0.1246430924665056, + "grad_norm": 2.3700242042541504, + "learning_rate": 6.223929747530186e-07, + "loss": 1.0719, + "mean_token_accuracy": 0.6803445816040039, + "num_tokens": 29015773.0, + "step": 1135 + }, + { + "epoch": 0.12475291016911927, + "grad_norm": 2.4432427883148193, + "learning_rate": 6.229418221734357e-07, + "loss": 0.9875, + "mean_token_accuracy": 0.6993682384490967, + "num_tokens": 29037564.0, + "step": 1136 + }, + { + "epoch": 0.12486272787173293, + "grad_norm": 2.453930377960205, + "learning_rate": 6.234906695938529e-07, + "loss": 1.1133, + "mean_token_accuracy": 0.670090913772583, + "num_tokens": 29061924.0, + "step": 1137 + }, + { + "epoch": 0.12497254557434659, + "grad_norm": 2.511662006378174, + "learning_rate": 6.2403951701427e-07, + "loss": 1.044, + "mean_token_accuracy": 0.6939488649368286, + "num_tokens": 29085067.0, + "step": 1138 + }, + { + "epoch": 0.12508236327696023, + "grad_norm": 2.291383743286133, + "learning_rate": 6.245883644346871e-07, + "loss": 1.0023, + "mean_token_accuracy": 0.6986606121063232, + "num_tokens": 29111172.0, + "step": 1139 + }, + { + "epoch": 0.1251921809795739, + "grad_norm": 2.193422317504883, + "learning_rate": 6.251372118551043e-07, + "loss": 1.0297, + "mean_token_accuracy": 0.6959479451179504, + "num_tokens": 29137946.0, + "step": 1140 + }, + { + "epoch": 0.12530199868218758, + "grad_norm": 2.267160415649414, + "learning_rate": 6.256860592755214e-07, + "loss": 1.1291, + "mean_token_accuracy": 0.666365385055542, + "num_tokens": 29164113.0, + "step": 1141 + }, + { + "epoch": 0.12541181638480123, + "grad_norm": 2.6167690753936768, + "learning_rate": 6.262349066959385e-07, + "loss": 1.1077, + "mean_token_accuracy": 0.6798285245895386, + "num_tokens": 29187004.0, + "step": 1142 + }, + { + "epoch": 0.1255216340874149, + "grad_norm": 2.347716808319092, + "learning_rate": 6.267837541163557e-07, + "loss": 1.0354, + "mean_token_accuracy": 0.6814848184585571, + "num_tokens": 29212325.0, + "step": 1143 + }, + { + "epoch": 0.12563145179002855, + "grad_norm": 2.609480619430542, + "learning_rate": 6.273326015367727e-07, + "loss": 1.032, + "mean_token_accuracy": 0.6907092332839966, + "num_tokens": 29233593.0, + "step": 1144 + }, + { + "epoch": 0.12574126949264222, + "grad_norm": 2.572300672531128, + "learning_rate": 6.278814489571899e-07, + "loss": 1.0542, + "mean_token_accuracy": 0.6904224157333374, + "num_tokens": 29255675.0, + "step": 1145 + }, + { + "epoch": 0.12585108719525587, + "grad_norm": 2.329184055328369, + "learning_rate": 6.284302963776071e-07, + "loss": 1.1209, + "mean_token_accuracy": 0.6734004020690918, + "num_tokens": 29281094.0, + "step": 1146 + }, + { + "epoch": 0.12596090489786954, + "grad_norm": 2.28016996383667, + "learning_rate": 6.289791437980241e-07, + "loss": 0.9153, + "mean_token_accuracy": 0.7259137034416199, + "num_tokens": 29307907.0, + "step": 1147 + }, + { + "epoch": 0.1260707226004832, + "grad_norm": 2.410409927368164, + "learning_rate": 6.295279912184413e-07, + "loss": 1.0919, + "mean_token_accuracy": 0.6743306517601013, + "num_tokens": 29330871.0, + "step": 1148 + }, + { + "epoch": 0.12618054030309686, + "grad_norm": 2.718230962753296, + "learning_rate": 6.300768386388585e-07, + "loss": 1.0396, + "mean_token_accuracy": 0.6898641586303711, + "num_tokens": 29350865.0, + "step": 1149 + }, + { + "epoch": 0.1262903580057105, + "grad_norm": 1.878237009048462, + "learning_rate": 6.306256860592754e-07, + "loss": 1.0032, + "mean_token_accuracy": 0.6978763341903687, + "num_tokens": 29386525.0, + "step": 1150 + }, + { + "epoch": 0.12640017570832418, + "grad_norm": 2.1850521564483643, + "learning_rate": 6.311745334796926e-07, + "loss": 1.1152, + "mean_token_accuracy": 0.6719352006912231, + "num_tokens": 29416207.0, + "step": 1151 + }, + { + "epoch": 0.12650999341093785, + "grad_norm": 2.3905858993530273, + "learning_rate": 6.317233809001098e-07, + "loss": 1.044, + "mean_token_accuracy": 0.6999943256378174, + "num_tokens": 29441518.0, + "step": 1152 + }, + { + "epoch": 0.1266198111135515, + "grad_norm": 2.233829975128174, + "learning_rate": 6.322722283205268e-07, + "loss": 1.1163, + "mean_token_accuracy": 0.6699068546295166, + "num_tokens": 29469307.0, + "step": 1153 + }, + { + "epoch": 0.12672962881616517, + "grad_norm": 2.893907070159912, + "learning_rate": 6.32821075740944e-07, + "loss": 0.9805, + "mean_token_accuracy": 0.6983928084373474, + "num_tokens": 29488227.0, + "step": 1154 + }, + { + "epoch": 0.12683944651877882, + "grad_norm": 2.158937454223633, + "learning_rate": 6.33369923161361e-07, + "loss": 1.0348, + "mean_token_accuracy": 0.6882863640785217, + "num_tokens": 29517041.0, + "step": 1155 + }, + { + "epoch": 0.1269492642213925, + "grad_norm": 2.14481258392334, + "learning_rate": 6.339187705817782e-07, + "loss": 1.0488, + "mean_token_accuracy": 0.6912362575531006, + "num_tokens": 29545817.0, + "step": 1156 + }, + { + "epoch": 0.12705908192400614, + "grad_norm": 2.249232530593872, + "learning_rate": 6.344676180021954e-07, + "loss": 0.9423, + "mean_token_accuracy": 0.7152783870697021, + "num_tokens": 29570748.0, + "step": 1157 + }, + { + "epoch": 0.1271688996266198, + "grad_norm": 2.3923768997192383, + "learning_rate": 6.350164654226124e-07, + "loss": 1.1087, + "mean_token_accuracy": 0.6725513935089111, + "num_tokens": 29595134.0, + "step": 1158 + }, + { + "epoch": 0.12727871732923346, + "grad_norm": 2.3654820919036865, + "learning_rate": 6.355653128430296e-07, + "loss": 1.0335, + "mean_token_accuracy": 0.6907879114151001, + "num_tokens": 29619004.0, + "step": 1159 + }, + { + "epoch": 0.12738853503184713, + "grad_norm": 2.241605281829834, + "learning_rate": 6.361141602634468e-07, + "loss": 1.025, + "mean_token_accuracy": 0.6927070617675781, + "num_tokens": 29645066.0, + "step": 1160 + }, + { + "epoch": 0.1274983527344608, + "grad_norm": 2.1334569454193115, + "learning_rate": 6.366630076838638e-07, + "loss": 0.9824, + "mean_token_accuracy": 0.7023938894271851, + "num_tokens": 29673093.0, + "step": 1161 + }, + { + "epoch": 0.12760817043707445, + "grad_norm": 2.2691023349761963, + "learning_rate": 6.37211855104281e-07, + "loss": 1.1007, + "mean_token_accuracy": 0.6670941114425659, + "num_tokens": 29700744.0, + "step": 1162 + }, + { + "epoch": 0.12771798813968813, + "grad_norm": 2.2989284992218018, + "learning_rate": 6.377607025246982e-07, + "loss": 1.082, + "mean_token_accuracy": 0.6763409972190857, + "num_tokens": 29726790.0, + "step": 1163 + }, + { + "epoch": 0.12782780584230177, + "grad_norm": 2.2655022144317627, + "learning_rate": 6.383095499451152e-07, + "loss": 1.0985, + "mean_token_accuracy": 0.6846755743026733, + "num_tokens": 29755354.0, + "step": 1164 + }, + { + "epoch": 0.12793762354491545, + "grad_norm": 2.2192208766937256, + "learning_rate": 6.388583973655324e-07, + "loss": 1.0567, + "mean_token_accuracy": 0.6883320808410645, + "num_tokens": 29784779.0, + "step": 1165 + }, + { + "epoch": 0.1280474412475291, + "grad_norm": 2.32658052444458, + "learning_rate": 6.394072447859495e-07, + "loss": 1.0445, + "mean_token_accuracy": 0.6861882209777832, + "num_tokens": 29809774.0, + "step": 1166 + }, + { + "epoch": 0.12815725895014277, + "grad_norm": 2.185602903366089, + "learning_rate": 6.399560922063666e-07, + "loss": 0.9621, + "mean_token_accuracy": 0.7104849219322205, + "num_tokens": 29837016.0, + "step": 1167 + }, + { + "epoch": 0.1282670766527564, + "grad_norm": 2.3272359371185303, + "learning_rate": 6.405049396267838e-07, + "loss": 1.005, + "mean_token_accuracy": 0.7016311883926392, + "num_tokens": 29863104.0, + "step": 1168 + }, + { + "epoch": 0.12837689435537009, + "grad_norm": 2.40974497795105, + "learning_rate": 6.410537870472008e-07, + "loss": 1.0318, + "mean_token_accuracy": 0.6926003098487854, + "num_tokens": 29886454.0, + "step": 1169 + }, + { + "epoch": 0.12848671205798376, + "grad_norm": 2.7563226222991943, + "learning_rate": 6.416026344676179e-07, + "loss": 1.0545, + "mean_token_accuracy": 0.6822482347488403, + "num_tokens": 29906687.0, + "step": 1170 + }, + { + "epoch": 0.1285965297605974, + "grad_norm": 2.151181697845459, + "learning_rate": 6.421514818880351e-07, + "loss": 1.0042, + "mean_token_accuracy": 0.7034266591072083, + "num_tokens": 29935153.0, + "step": 1171 + }, + { + "epoch": 0.12870634746321108, + "grad_norm": 2.3818390369415283, + "learning_rate": 6.427003293084522e-07, + "loss": 1.0033, + "mean_token_accuracy": 0.6917464733123779, + "num_tokens": 29957860.0, + "step": 1172 + }, + { + "epoch": 0.12881616516582473, + "grad_norm": 2.1935336589813232, + "learning_rate": 6.432491767288693e-07, + "loss": 1.0582, + "mean_token_accuracy": 0.6854526996612549, + "num_tokens": 29984309.0, + "step": 1173 + }, + { + "epoch": 0.1289259828684384, + "grad_norm": 2.01192045211792, + "learning_rate": 6.437980241492865e-07, + "loss": 1.0897, + "mean_token_accuracy": 0.677416205406189, + "num_tokens": 30015975.0, + "step": 1174 + }, + { + "epoch": 0.12903580057105205, + "grad_norm": 2.474642038345337, + "learning_rate": 6.443468715697036e-07, + "loss": 1.0503, + "mean_token_accuracy": 0.6974145770072937, + "num_tokens": 30039407.0, + "step": 1175 + }, + { + "epoch": 0.12914561827366572, + "grad_norm": 2.332347869873047, + "learning_rate": 6.448957189901207e-07, + "loss": 1.0569, + "mean_token_accuracy": 0.7019035220146179, + "num_tokens": 30063718.0, + "step": 1176 + }, + { + "epoch": 0.12925543597627936, + "grad_norm": 2.4443304538726807, + "learning_rate": 6.454445664105378e-07, + "loss": 1.08, + "mean_token_accuracy": 0.6859581470489502, + "num_tokens": 30088927.0, + "step": 1177 + }, + { + "epoch": 0.12936525367889304, + "grad_norm": 2.541144371032715, + "learning_rate": 6.45993413830955e-07, + "loss": 1.0302, + "mean_token_accuracy": 0.7000258564949036, + "num_tokens": 30111153.0, + "step": 1178 + }, + { + "epoch": 0.1294750713815067, + "grad_norm": 2.667220115661621, + "learning_rate": 6.465422612513721e-07, + "loss": 1.039, + "mean_token_accuracy": 0.6886048913002014, + "num_tokens": 30131780.0, + "step": 1179 + }, + { + "epoch": 0.12958488908412036, + "grad_norm": 2.544753313064575, + "learning_rate": 6.470911086717892e-07, + "loss": 1.0799, + "mean_token_accuracy": 0.6820244193077087, + "num_tokens": 30154646.0, + "step": 1180 + }, + { + "epoch": 0.12969470678673403, + "grad_norm": 2.1568830013275146, + "learning_rate": 6.476399560922064e-07, + "loss": 1.1485, + "mean_token_accuracy": 0.6643336415290833, + "num_tokens": 30185633.0, + "step": 1181 + }, + { + "epoch": 0.12980452448934768, + "grad_norm": 2.71431565284729, + "learning_rate": 6.481888035126235e-07, + "loss": 1.0124, + "mean_token_accuracy": 0.7032834887504578, + "num_tokens": 30206483.0, + "step": 1182 + }, + { + "epoch": 0.12991434219196135, + "grad_norm": 2.466132164001465, + "learning_rate": 6.487376509330406e-07, + "loss": 0.9908, + "mean_token_accuracy": 0.7075790762901306, + "num_tokens": 30229412.0, + "step": 1183 + }, + { + "epoch": 0.130024159894575, + "grad_norm": 2.1819372177124023, + "learning_rate": 6.492864983534578e-07, + "loss": 1.1627, + "mean_token_accuracy": 0.6690676808357239, + "num_tokens": 30259826.0, + "step": 1184 + }, + { + "epoch": 0.13013397759718867, + "grad_norm": 2.245145559310913, + "learning_rate": 6.498353457738749e-07, + "loss": 1.0052, + "mean_token_accuracy": 0.6974976062774658, + "num_tokens": 30285037.0, + "step": 1185 + }, + { + "epoch": 0.13024379529980232, + "grad_norm": 2.115730047225952, + "learning_rate": 6.50384193194292e-07, + "loss": 1.0479, + "mean_token_accuracy": 0.6943193078041077, + "num_tokens": 30313976.0, + "step": 1186 + }, + { + "epoch": 0.130353613002416, + "grad_norm": 2.4973833560943604, + "learning_rate": 6.509330406147092e-07, + "loss": 1.0482, + "mean_token_accuracy": 0.6835758090019226, + "num_tokens": 30338102.0, + "step": 1187 + }, + { + "epoch": 0.13046343070502964, + "grad_norm": 3.177109718322754, + "learning_rate": 6.514818880351261e-07, + "loss": 0.9339, + "mean_token_accuracy": 0.7174923419952393, + "num_tokens": 30353637.0, + "step": 1188 + }, + { + "epoch": 0.1305732484076433, + "grad_norm": 2.733936071395874, + "learning_rate": 6.520307354555433e-07, + "loss": 1.1287, + "mean_token_accuracy": 0.6789090037345886, + "num_tokens": 30373941.0, + "step": 1189 + }, + { + "epoch": 0.13068306611025698, + "grad_norm": 2.452324390411377, + "learning_rate": 6.525795828759604e-07, + "loss": 1.0031, + "mean_token_accuracy": 0.6967952251434326, + "num_tokens": 30396478.0, + "step": 1190 + }, + { + "epoch": 0.13079288381287063, + "grad_norm": 2.4944403171539307, + "learning_rate": 6.531284302963775e-07, + "loss": 1.076, + "mean_token_accuracy": 0.6901614665985107, + "num_tokens": 30418370.0, + "step": 1191 + }, + { + "epoch": 0.1309027015154843, + "grad_norm": 2.4639973640441895, + "learning_rate": 6.536772777167947e-07, + "loss": 0.9888, + "mean_token_accuracy": 0.7036996483802795, + "num_tokens": 30441187.0, + "step": 1192 + }, + { + "epoch": 0.13101251921809795, + "grad_norm": 2.3344688415527344, + "learning_rate": 6.542261251372118e-07, + "loss": 1.0583, + "mean_token_accuracy": 0.6925820112228394, + "num_tokens": 30466139.0, + "step": 1193 + }, + { + "epoch": 0.13112233692071162, + "grad_norm": 2.360260248184204, + "learning_rate": 6.547749725576289e-07, + "loss": 1.0398, + "mean_token_accuracy": 0.6975589990615845, + "num_tokens": 30490430.0, + "step": 1194 + }, + { + "epoch": 0.13123215462332527, + "grad_norm": 2.308945894241333, + "learning_rate": 6.553238199780461e-07, + "loss": 1.1117, + "mean_token_accuracy": 0.6794428825378418, + "num_tokens": 30517683.0, + "step": 1195 + }, + { + "epoch": 0.13134197232593894, + "grad_norm": 2.4571683406829834, + "learning_rate": 6.558726673984632e-07, + "loss": 1.0224, + "mean_token_accuracy": 0.6963696479797363, + "num_tokens": 30542038.0, + "step": 1196 + }, + { + "epoch": 0.1314517900285526, + "grad_norm": 2.3630874156951904, + "learning_rate": 6.564215148188803e-07, + "loss": 1.0492, + "mean_token_accuracy": 0.689022421836853, + "num_tokens": 30566258.0, + "step": 1197 + }, + { + "epoch": 0.13156160773116626, + "grad_norm": 2.506849765777588, + "learning_rate": 6.569703622392975e-07, + "loss": 1.0282, + "mean_token_accuracy": 0.6885770559310913, + "num_tokens": 30588555.0, + "step": 1198 + }, + { + "epoch": 0.13167142543377994, + "grad_norm": 2.3105764389038086, + "learning_rate": 6.575192096597145e-07, + "loss": 0.966, + "mean_token_accuracy": 0.7088468074798584, + "num_tokens": 30611978.0, + "step": 1199 + }, + { + "epoch": 0.13178124313639358, + "grad_norm": 2.4296090602874756, + "learning_rate": 6.580680570801317e-07, + "loss": 1.0066, + "mean_token_accuracy": 0.6952579021453857, + "num_tokens": 30635031.0, + "step": 1200 + }, + { + "epoch": 0.13189106083900726, + "grad_norm": 2.120295763015747, + "learning_rate": 6.586169045005489e-07, + "loss": 0.9823, + "mean_token_accuracy": 0.7025829553604126, + "num_tokens": 30661959.0, + "step": 1201 + }, + { + "epoch": 0.1320008785416209, + "grad_norm": 2.261159896850586, + "learning_rate": 6.591657519209659e-07, + "loss": 1.0127, + "mean_token_accuracy": 0.7045356035232544, + "num_tokens": 30687074.0, + "step": 1202 + }, + { + "epoch": 0.13211069624423458, + "grad_norm": 2.1071441173553467, + "learning_rate": 6.597145993413831e-07, + "loss": 1.128, + "mean_token_accuracy": 0.6650514602661133, + "num_tokens": 30717553.0, + "step": 1203 + }, + { + "epoch": 0.13222051394684822, + "grad_norm": 2.2233810424804688, + "learning_rate": 6.602634467618003e-07, + "loss": 1.0296, + "mean_token_accuracy": 0.692791759967804, + "num_tokens": 30745825.0, + "step": 1204 + }, + { + "epoch": 0.1323303316494619, + "grad_norm": 1.9670255184173584, + "learning_rate": 6.608122941822173e-07, + "loss": 1.0712, + "mean_token_accuracy": 0.6772178411483765, + "num_tokens": 30783456.0, + "step": 1205 + }, + { + "epoch": 0.13244014935207554, + "grad_norm": 2.3234527111053467, + "learning_rate": 6.613611416026345e-07, + "loss": 1.0313, + "mean_token_accuracy": 0.6850552558898926, + "num_tokens": 30808386.0, + "step": 1206 + }, + { + "epoch": 0.13254996705468922, + "grad_norm": 2.222641944885254, + "learning_rate": 6.619099890230515e-07, + "loss": 1.0537, + "mean_token_accuracy": 0.6837196946144104, + "num_tokens": 30836670.0, + "step": 1207 + }, + { + "epoch": 0.1326597847573029, + "grad_norm": 2.241655111312866, + "learning_rate": 6.624588364434686e-07, + "loss": 1.1079, + "mean_token_accuracy": 0.6777092218399048, + "num_tokens": 30865153.0, + "step": 1208 + }, + { + "epoch": 0.13276960245991654, + "grad_norm": 2.467984199523926, + "learning_rate": 6.630076838638858e-07, + "loss": 1.0674, + "mean_token_accuracy": 0.6825209856033325, + "num_tokens": 30888638.0, + "step": 1209 + }, + { + "epoch": 0.1328794201625302, + "grad_norm": 2.187246799468994, + "learning_rate": 6.635565312843029e-07, + "loss": 1.0704, + "mean_token_accuracy": 0.6953966617584229, + "num_tokens": 30915939.0, + "step": 1210 + }, + { + "epoch": 0.13298923786514386, + "grad_norm": 2.0390007495880127, + "learning_rate": 6.6410537870472e-07, + "loss": 1.0821, + "mean_token_accuracy": 0.6820299625396729, + "num_tokens": 30948340.0, + "step": 1211 + }, + { + "epoch": 0.13309905556775753, + "grad_norm": 2.0703227519989014, + "learning_rate": 6.646542261251372e-07, + "loss": 1.0867, + "mean_token_accuracy": 0.6791175603866577, + "num_tokens": 30980349.0, + "step": 1212 + }, + { + "epoch": 0.13320887327037118, + "grad_norm": 2.2604944705963135, + "learning_rate": 6.652030735455543e-07, + "loss": 1.0049, + "mean_token_accuracy": 0.6976976990699768, + "num_tokens": 31006534.0, + "step": 1213 + }, + { + "epoch": 0.13331869097298485, + "grad_norm": 2.261094331741333, + "learning_rate": 6.657519209659714e-07, + "loss": 1.0727, + "mean_token_accuracy": 0.686668872833252, + "num_tokens": 31032778.0, + "step": 1214 + }, + { + "epoch": 0.1334285086755985, + "grad_norm": 2.574907064437866, + "learning_rate": 6.663007683863886e-07, + "loss": 1.0317, + "mean_token_accuracy": 0.6980312466621399, + "num_tokens": 31054474.0, + "step": 1215 + }, + { + "epoch": 0.13353832637821217, + "grad_norm": 2.6807873249053955, + "learning_rate": 6.668496158068057e-07, + "loss": 1.0705, + "mean_token_accuracy": 0.6819486618041992, + "num_tokens": 31077607.0, + "step": 1216 + }, + { + "epoch": 0.13364814408082584, + "grad_norm": 2.3285648822784424, + "learning_rate": 6.673984632272228e-07, + "loss": 0.9758, + "mean_token_accuracy": 0.6997992396354675, + "num_tokens": 31101481.0, + "step": 1217 + }, + { + "epoch": 0.1337579617834395, + "grad_norm": 2.416393995285034, + "learning_rate": 6.679473106476399e-07, + "loss": 0.9987, + "mean_token_accuracy": 0.7012567520141602, + "num_tokens": 31124529.0, + "step": 1218 + }, + { + "epoch": 0.13386777948605316, + "grad_norm": 2.1786179542541504, + "learning_rate": 6.684961580680571e-07, + "loss": 1.0661, + "mean_token_accuracy": 0.6763957142829895, + "num_tokens": 31153059.0, + "step": 1219 + }, + { + "epoch": 0.1339775971886668, + "grad_norm": 2.1867151260375977, + "learning_rate": 6.690450054884742e-07, + "loss": 0.963, + "mean_token_accuracy": 0.711469829082489, + "num_tokens": 31181362.0, + "step": 1220 + }, + { + "epoch": 0.13408741489128048, + "grad_norm": 2.2449092864990234, + "learning_rate": 6.695938529088913e-07, + "loss": 1.0306, + "mean_token_accuracy": 0.6889147758483887, + "num_tokens": 31207618.0, + "step": 1221 + }, + { + "epoch": 0.13419723259389413, + "grad_norm": 2.466148853302002, + "learning_rate": 6.701427003293085e-07, + "loss": 1.0357, + "mean_token_accuracy": 0.6940021514892578, + "num_tokens": 31231361.0, + "step": 1222 + }, + { + "epoch": 0.1343070502965078, + "grad_norm": 2.135155200958252, + "learning_rate": 6.706915477497256e-07, + "loss": 0.985, + "mean_token_accuracy": 0.7018053531646729, + "num_tokens": 31261454.0, + "step": 1223 + }, + { + "epoch": 0.13441686799912145, + "grad_norm": 2.4702537059783936, + "learning_rate": 6.712403951701427e-07, + "loss": 1.1411, + "mean_token_accuracy": 0.665968656539917, + "num_tokens": 31285362.0, + "step": 1224 + }, + { + "epoch": 0.13452668570173512, + "grad_norm": 2.239283800125122, + "learning_rate": 6.717892425905599e-07, + "loss": 1.1361, + "mean_token_accuracy": 0.661713719367981, + "num_tokens": 31314504.0, + "step": 1225 + }, + { + "epoch": 0.13463650340434877, + "grad_norm": 2.702317714691162, + "learning_rate": 6.723380900109769e-07, + "loss": 0.924, + "mean_token_accuracy": 0.7256633639335632, + "num_tokens": 31333348.0, + "step": 1226 + }, + { + "epoch": 0.13474632110696244, + "grad_norm": 2.6889727115631104, + "learning_rate": 6.72886937431394e-07, + "loss": 1.0516, + "mean_token_accuracy": 0.6918603181838989, + "num_tokens": 31356136.0, + "step": 1227 + }, + { + "epoch": 0.13485613880957611, + "grad_norm": 2.1080689430236816, + "learning_rate": 6.734357848518111e-07, + "loss": 1.0541, + "mean_token_accuracy": 0.6891375184059143, + "num_tokens": 31386137.0, + "step": 1228 + }, + { + "epoch": 0.13496595651218976, + "grad_norm": 2.23156476020813, + "learning_rate": 6.739846322722282e-07, + "loss": 1.0372, + "mean_token_accuracy": 0.6853164434432983, + "num_tokens": 31415987.0, + "step": 1229 + }, + { + "epoch": 0.13507577421480343, + "grad_norm": 2.287879228591919, + "learning_rate": 6.745334796926454e-07, + "loss": 1.0722, + "mean_token_accuracy": 0.6819713711738586, + "num_tokens": 31441785.0, + "step": 1230 + }, + { + "epoch": 0.13518559191741708, + "grad_norm": 2.0594887733459473, + "learning_rate": 6.750823271130625e-07, + "loss": 1.12, + "mean_token_accuracy": 0.6651239395141602, + "num_tokens": 31473599.0, + "step": 1231 + }, + { + "epoch": 0.13529540962003075, + "grad_norm": 2.019110679626465, + "learning_rate": 6.756311745334796e-07, + "loss": 1.0451, + "mean_token_accuracy": 0.6872040033340454, + "num_tokens": 31504433.0, + "step": 1232 + }, + { + "epoch": 0.1354052273226444, + "grad_norm": 2.413830041885376, + "learning_rate": 6.761800219538968e-07, + "loss": 1.1149, + "mean_token_accuracy": 0.6789494752883911, + "num_tokens": 31528710.0, + "step": 1233 + }, + { + "epoch": 0.13551504502525807, + "grad_norm": 2.1716208457946777, + "learning_rate": 6.767288693743139e-07, + "loss": 1.0725, + "mean_token_accuracy": 0.6883137822151184, + "num_tokens": 31556311.0, + "step": 1234 + }, + { + "epoch": 0.13562486272787172, + "grad_norm": 2.3207132816314697, + "learning_rate": 6.77277716794731e-07, + "loss": 0.9675, + "mean_token_accuracy": 0.7134361267089844, + "num_tokens": 31581721.0, + "step": 1235 + }, + { + "epoch": 0.1357346804304854, + "grad_norm": 2.365119218826294, + "learning_rate": 6.778265642151482e-07, + "loss": 1.0749, + "mean_token_accuracy": 0.6844238042831421, + "num_tokens": 31606471.0, + "step": 1236 + }, + { + "epoch": 0.13584449813309907, + "grad_norm": 2.749690055847168, + "learning_rate": 6.783754116355653e-07, + "loss": 1.0749, + "mean_token_accuracy": 0.6830126047134399, + "num_tokens": 31635489.0, + "step": 1237 + }, + { + "epoch": 0.1359543158357127, + "grad_norm": 2.2916359901428223, + "learning_rate": 6.789242590559824e-07, + "loss": 1.122, + "mean_token_accuracy": 0.6694114804267883, + "num_tokens": 31662860.0, + "step": 1238 + }, + { + "epoch": 0.1360641335383264, + "grad_norm": 2.458491563796997, + "learning_rate": 6.794731064763996e-07, + "loss": 1.0327, + "mean_token_accuracy": 0.6891790628433228, + "num_tokens": 31686430.0, + "step": 1239 + }, + { + "epoch": 0.13617395124094003, + "grad_norm": 2.4843342304229736, + "learning_rate": 6.800219538968166e-07, + "loss": 0.997, + "mean_token_accuracy": 0.696845531463623, + "num_tokens": 31708453.0, + "step": 1240 + }, + { + "epoch": 0.1362837689435537, + "grad_norm": 2.162937641143799, + "learning_rate": 6.805708013172338e-07, + "loss": 0.9788, + "mean_token_accuracy": 0.7024153470993042, + "num_tokens": 31736053.0, + "step": 1241 + }, + { + "epoch": 0.13639358664616735, + "grad_norm": 2.5933494567871094, + "learning_rate": 6.81119648737651e-07, + "loss": 1.0337, + "mean_token_accuracy": 0.6909090876579285, + "num_tokens": 31757064.0, + "step": 1242 + }, + { + "epoch": 0.13650340434878103, + "grad_norm": 2.386564016342163, + "learning_rate": 6.81668496158068e-07, + "loss": 1.0568, + "mean_token_accuracy": 0.6826043128967285, + "num_tokens": 31783470.0, + "step": 1243 + }, + { + "epoch": 0.13661322205139467, + "grad_norm": 2.2309389114379883, + "learning_rate": 6.822173435784852e-07, + "loss": 1.0386, + "mean_token_accuracy": 0.6902261972427368, + "num_tokens": 31809999.0, + "step": 1244 + }, + { + "epoch": 0.13672303975400835, + "grad_norm": 2.3515865802764893, + "learning_rate": 6.827661909989023e-07, + "loss": 1.0198, + "mean_token_accuracy": 0.6941908597946167, + "num_tokens": 31833579.0, + "step": 1245 + }, + { + "epoch": 0.13683285745662202, + "grad_norm": 2.3050129413604736, + "learning_rate": 6.833150384193193e-07, + "loss": 1.1286, + "mean_token_accuracy": 0.6809172034263611, + "num_tokens": 31862689.0, + "step": 1246 + }, + { + "epoch": 0.13694267515923567, + "grad_norm": 2.2516815662384033, + "learning_rate": 6.838638858397365e-07, + "loss": 1.105, + "mean_token_accuracy": 0.676328718662262, + "num_tokens": 31891658.0, + "step": 1247 + }, + { + "epoch": 0.13705249286184934, + "grad_norm": 2.789647102355957, + "learning_rate": 6.844127332601537e-07, + "loss": 0.9485, + "mean_token_accuracy": 0.7107663750648499, + "num_tokens": 31910488.0, + "step": 1248 + }, + { + "epoch": 0.13716231056446299, + "grad_norm": 2.1391942501068115, + "learning_rate": 6.849615806805707e-07, + "loss": 1.1511, + "mean_token_accuracy": 0.670203447341919, + "num_tokens": 31943625.0, + "step": 1249 + }, + { + "epoch": 0.13727212826707666, + "grad_norm": 2.3091864585876465, + "learning_rate": 6.855104281009879e-07, + "loss": 1.0006, + "mean_token_accuracy": 0.6984648704528809, + "num_tokens": 31971344.0, + "step": 1250 + }, + { + "epoch": 0.1373819459696903, + "grad_norm": 2.5095601081848145, + "learning_rate": 6.86059275521405e-07, + "loss": 1.0895, + "mean_token_accuracy": 0.6691877841949463, + "num_tokens": 31994278.0, + "step": 1251 + }, + { + "epoch": 0.13749176367230398, + "grad_norm": 2.364649772644043, + "learning_rate": 6.866081229418221e-07, + "loss": 1.0707, + "mean_token_accuracy": 0.6826422214508057, + "num_tokens": 32019248.0, + "step": 1252 + }, + { + "epoch": 0.13760158137491763, + "grad_norm": 2.2945921421051025, + "learning_rate": 6.871569703622393e-07, + "loss": 0.9415, + "mean_token_accuracy": 0.711524248123169, + "num_tokens": 32042846.0, + "step": 1253 + }, + { + "epoch": 0.1377113990775313, + "grad_norm": 2.081045150756836, + "learning_rate": 6.877058177826564e-07, + "loss": 1.0875, + "mean_token_accuracy": 0.6837338209152222, + "num_tokens": 32074410.0, + "step": 1254 + }, + { + "epoch": 0.13782121678014497, + "grad_norm": 2.3178529739379883, + "learning_rate": 6.882546652030735e-07, + "loss": 0.9468, + "mean_token_accuracy": 0.7170696258544922, + "num_tokens": 32100580.0, + "step": 1255 + }, + { + "epoch": 0.13793103448275862, + "grad_norm": 2.196028470993042, + "learning_rate": 6.888035126234907e-07, + "loss": 1.0906, + "mean_token_accuracy": 0.6834166049957275, + "num_tokens": 32127428.0, + "step": 1256 + }, + { + "epoch": 0.1380408521853723, + "grad_norm": 1.9426381587982178, + "learning_rate": 6.893523600439078e-07, + "loss": 1.0477, + "mean_token_accuracy": 0.6882951259613037, + "num_tokens": 32159964.0, + "step": 1257 + }, + { + "epoch": 0.13815066988798594, + "grad_norm": 2.364988327026367, + "learning_rate": 6.899012074643249e-07, + "loss": 1.0655, + "mean_token_accuracy": 0.6841843128204346, + "num_tokens": 32186894.0, + "step": 1258 + }, + { + "epoch": 0.1382604875905996, + "grad_norm": 2.5277838706970215, + "learning_rate": 6.904500548847421e-07, + "loss": 1.058, + "mean_token_accuracy": 0.6864731907844543, + "num_tokens": 32211838.0, + "step": 1259 + }, + { + "epoch": 0.13837030529321326, + "grad_norm": 2.404265880584717, + "learning_rate": 6.909989023051592e-07, + "loss": 1.1157, + "mean_token_accuracy": 0.6663843393325806, + "num_tokens": 32236616.0, + "step": 1260 + }, + { + "epoch": 0.13848012299582693, + "grad_norm": 2.5680551528930664, + "learning_rate": 6.915477497255763e-07, + "loss": 1.0683, + "mean_token_accuracy": 0.6791365146636963, + "num_tokens": 32259147.0, + "step": 1261 + }, + { + "epoch": 0.13858994069844058, + "grad_norm": 2.51232647895813, + "learning_rate": 6.920965971459934e-07, + "loss": 1.0938, + "mean_token_accuracy": 0.6734046936035156, + "num_tokens": 32283279.0, + "step": 1262 + }, + { + "epoch": 0.13869975840105425, + "grad_norm": 2.2515740394592285, + "learning_rate": 6.926454445664105e-07, + "loss": 1.0294, + "mean_token_accuracy": 0.6890336275100708, + "num_tokens": 32311165.0, + "step": 1263 + }, + { + "epoch": 0.1388095761036679, + "grad_norm": 2.4086499214172363, + "learning_rate": 6.931942919868276e-07, + "loss": 0.9885, + "mean_token_accuracy": 0.6982499361038208, + "num_tokens": 32334543.0, + "step": 1264 + }, + { + "epoch": 0.13891939380628157, + "grad_norm": 2.2818126678466797, + "learning_rate": 6.937431394072447e-07, + "loss": 1.0847, + "mean_token_accuracy": 0.6882123351097107, + "num_tokens": 32362637.0, + "step": 1265 + }, + { + "epoch": 0.13902921150889525, + "grad_norm": 2.4158411026000977, + "learning_rate": 6.942919868276618e-07, + "loss": 1.0049, + "mean_token_accuracy": 0.70120769739151, + "num_tokens": 32387266.0, + "step": 1266 + }, + { + "epoch": 0.1391390292115089, + "grad_norm": 2.5345304012298584, + "learning_rate": 6.94840834248079e-07, + "loss": 1.0775, + "mean_token_accuracy": 0.6801779270172119, + "num_tokens": 32410402.0, + "step": 1267 + }, + { + "epoch": 0.13924884691412256, + "grad_norm": 2.1187427043914795, + "learning_rate": 6.953896816684961e-07, + "loss": 1.0644, + "mean_token_accuracy": 0.6874958276748657, + "num_tokens": 32440729.0, + "step": 1268 + }, + { + "epoch": 0.1393586646167362, + "grad_norm": 2.378587484359741, + "learning_rate": 6.959385290889132e-07, + "loss": 0.9292, + "mean_token_accuracy": 0.7186192870140076, + "num_tokens": 32464723.0, + "step": 1269 + }, + { + "epoch": 0.13946848231934988, + "grad_norm": 2.2482640743255615, + "learning_rate": 6.964873765093304e-07, + "loss": 1.1013, + "mean_token_accuracy": 0.6747531890869141, + "num_tokens": 32494062.0, + "step": 1270 + }, + { + "epoch": 0.13957830002196353, + "grad_norm": 2.327904224395752, + "learning_rate": 6.970362239297475e-07, + "loss": 0.9393, + "mean_token_accuracy": 0.7226214408874512, + "num_tokens": 32517003.0, + "step": 1271 + }, + { + "epoch": 0.1396881177245772, + "grad_norm": 2.3066823482513428, + "learning_rate": 6.975850713501646e-07, + "loss": 1.0936, + "mean_token_accuracy": 0.6823464632034302, + "num_tokens": 32544459.0, + "step": 1272 + }, + { + "epoch": 0.13979793542719085, + "grad_norm": 2.2770802974700928, + "learning_rate": 6.981339187705817e-07, + "loss": 1.0422, + "mean_token_accuracy": 0.6957260370254517, + "num_tokens": 32572069.0, + "step": 1273 + }, + { + "epoch": 0.13990775312980452, + "grad_norm": 1.9297000169754028, + "learning_rate": 6.986827661909989e-07, + "loss": 1.1055, + "mean_token_accuracy": 0.6669110059738159, + "num_tokens": 32608742.0, + "step": 1274 + }, + { + "epoch": 0.1400175708324182, + "grad_norm": 2.577317476272583, + "learning_rate": 6.99231613611416e-07, + "loss": 1.0246, + "mean_token_accuracy": 0.6969162821769714, + "num_tokens": 32629895.0, + "step": 1275 + }, + { + "epoch": 0.14012738853503184, + "grad_norm": 2.486680269241333, + "learning_rate": 6.997804610318331e-07, + "loss": 1.0602, + "mean_token_accuracy": 0.698925793170929, + "num_tokens": 32652586.0, + "step": 1276 + }, + { + "epoch": 0.14023720623764552, + "grad_norm": 2.2639036178588867, + "learning_rate": 7.003293084522503e-07, + "loss": 1.138, + "mean_token_accuracy": 0.6660692691802979, + "num_tokens": 32680588.0, + "step": 1277 + }, + { + "epoch": 0.14034702394025916, + "grad_norm": 2.685483932495117, + "learning_rate": 7.008781558726674e-07, + "loss": 1.0504, + "mean_token_accuracy": 0.6837084293365479, + "num_tokens": 32700009.0, + "step": 1278 + }, + { + "epoch": 0.14045684164287284, + "grad_norm": 2.3001441955566406, + "learning_rate": 7.014270032930845e-07, + "loss": 1.0586, + "mean_token_accuracy": 0.6894928216934204, + "num_tokens": 32725617.0, + "step": 1279 + }, + { + "epoch": 0.14056665934548648, + "grad_norm": 2.462036609649658, + "learning_rate": 7.019758507135017e-07, + "loss": 1.0074, + "mean_token_accuracy": 0.693831205368042, + "num_tokens": 32747613.0, + "step": 1280 + }, + { + "epoch": 0.14067647704810016, + "grad_norm": 2.2492594718933105, + "learning_rate": 7.025246981339188e-07, + "loss": 0.9942, + "mean_token_accuracy": 0.7049951553344727, + "num_tokens": 32772443.0, + "step": 1281 + }, + { + "epoch": 0.1407862947507138, + "grad_norm": 2.191500425338745, + "learning_rate": 7.030735455543358e-07, + "loss": 0.9734, + "mean_token_accuracy": 0.7097636461257935, + "num_tokens": 32797614.0, + "step": 1282 + }, + { + "epoch": 0.14089611245332748, + "grad_norm": 2.4377145767211914, + "learning_rate": 7.03622392974753e-07, + "loss": 1.0188, + "mean_token_accuracy": 0.6971884965896606, + "num_tokens": 32820788.0, + "step": 1283 + }, + { + "epoch": 0.14100593015594115, + "grad_norm": 2.378610134124756, + "learning_rate": 7.0417124039517e-07, + "loss": 1.0187, + "mean_token_accuracy": 0.6997827291488647, + "num_tokens": 32844114.0, + "step": 1284 + }, + { + "epoch": 0.1411157478585548, + "grad_norm": 2.5428242683410645, + "learning_rate": 7.047200878155872e-07, + "loss": 1.0978, + "mean_token_accuracy": 0.6851783990859985, + "num_tokens": 32865285.0, + "step": 1285 + }, + { + "epoch": 0.14122556556116847, + "grad_norm": 2.5657551288604736, + "learning_rate": 7.052689352360044e-07, + "loss": 0.991, + "mean_token_accuracy": 0.6988951563835144, + "num_tokens": 32887613.0, + "step": 1286 + }, + { + "epoch": 0.14133538326378212, + "grad_norm": 2.249654531478882, + "learning_rate": 7.058177826564214e-07, + "loss": 1.0771, + "mean_token_accuracy": 0.6897719502449036, + "num_tokens": 32913980.0, + "step": 1287 + }, + { + "epoch": 0.1414452009663958, + "grad_norm": 2.349440097808838, + "learning_rate": 7.063666300768386e-07, + "loss": 1.1432, + "mean_token_accuracy": 0.661970853805542, + "num_tokens": 32942849.0, + "step": 1288 + }, + { + "epoch": 0.14155501866900944, + "grad_norm": 2.1611223220825195, + "learning_rate": 7.069154774972558e-07, + "loss": 1.0891, + "mean_token_accuracy": 0.6776362061500549, + "num_tokens": 32972872.0, + "step": 1289 + }, + { + "epoch": 0.1416648363716231, + "grad_norm": 2.328237771987915, + "learning_rate": 7.074643249176728e-07, + "loss": 1.006, + "mean_token_accuracy": 0.6977503895759583, + "num_tokens": 32997153.0, + "step": 1290 + }, + { + "epoch": 0.14177465407423676, + "grad_norm": 2.4661529064178467, + "learning_rate": 7.0801317233809e-07, + "loss": 0.9634, + "mean_token_accuracy": 0.7147178053855896, + "num_tokens": 33018495.0, + "step": 1291 + }, + { + "epoch": 0.14188447177685043, + "grad_norm": 2.489710569381714, + "learning_rate": 7.085620197585072e-07, + "loss": 1.1088, + "mean_token_accuracy": 0.6750771403312683, + "num_tokens": 33044244.0, + "step": 1292 + }, + { + "epoch": 0.1419942894794641, + "grad_norm": 1.9317710399627686, + "learning_rate": 7.091108671789242e-07, + "loss": 1.1058, + "mean_token_accuracy": 0.6743650436401367, + "num_tokens": 33078369.0, + "step": 1293 + }, + { + "epoch": 0.14210410718207775, + "grad_norm": 2.2339820861816406, + "learning_rate": 7.096597145993414e-07, + "loss": 1.0133, + "mean_token_accuracy": 0.6926794052124023, + "num_tokens": 33104597.0, + "step": 1294 + }, + { + "epoch": 0.14221392488469142, + "grad_norm": 2.4692559242248535, + "learning_rate": 7.102085620197584e-07, + "loss": 1.0725, + "mean_token_accuracy": 0.6711868047714233, + "num_tokens": 33127556.0, + "step": 1295 + }, + { + "epoch": 0.14232374258730507, + "grad_norm": 2.2038283348083496, + "learning_rate": 7.107574094401756e-07, + "loss": 1.0146, + "mean_token_accuracy": 0.7060924172401428, + "num_tokens": 33152377.0, + "step": 1296 + }, + { + "epoch": 0.14243356028991874, + "grad_norm": 2.402479887008667, + "learning_rate": 7.113062568605928e-07, + "loss": 1.0255, + "mean_token_accuracy": 0.6881752014160156, + "num_tokens": 33175384.0, + "step": 1297 + }, + { + "epoch": 0.1425433779925324, + "grad_norm": 2.1322543621063232, + "learning_rate": 7.118551042810098e-07, + "loss": 1.0568, + "mean_token_accuracy": 0.683943510055542, + "num_tokens": 33202227.0, + "step": 1298 + }, + { + "epoch": 0.14265319569514606, + "grad_norm": 2.373429536819458, + "learning_rate": 7.12403951701427e-07, + "loss": 1.1412, + "mean_token_accuracy": 0.6624051928520203, + "num_tokens": 33227928.0, + "step": 1299 + }, + { + "epoch": 0.1427630133977597, + "grad_norm": 2.2120320796966553, + "learning_rate": 7.129527991218442e-07, + "loss": 1.0418, + "mean_token_accuracy": 0.6851433515548706, + "num_tokens": 33253317.0, + "step": 1300 + }, + { + "epoch": 0.14287283110037338, + "grad_norm": 2.3247225284576416, + "learning_rate": 7.135016465422611e-07, + "loss": 1.0448, + "mean_token_accuracy": 0.7003815770149231, + "num_tokens": 33278399.0, + "step": 1301 + }, + { + "epoch": 0.14298264880298703, + "grad_norm": 2.3174057006835938, + "learning_rate": 7.140504939626783e-07, + "loss": 0.9106, + "mean_token_accuracy": 0.7240427732467651, + "num_tokens": 33302136.0, + "step": 1302 + }, + { + "epoch": 0.1430924665056007, + "grad_norm": 2.1406092643737793, + "learning_rate": 7.145993413830955e-07, + "loss": 1.0296, + "mean_token_accuracy": 0.697746992111206, + "num_tokens": 33330914.0, + "step": 1303 + }, + { + "epoch": 0.14320228420821438, + "grad_norm": 2.10473370552063, + "learning_rate": 7.151481888035125e-07, + "loss": 0.9973, + "mean_token_accuracy": 0.7015065550804138, + "num_tokens": 33359508.0, + "step": 1304 + }, + { + "epoch": 0.14331210191082802, + "grad_norm": 2.2708346843719482, + "learning_rate": 7.156970362239297e-07, + "loss": 0.9856, + "mean_token_accuracy": 0.7061212658882141, + "num_tokens": 33384110.0, + "step": 1305 + }, + { + "epoch": 0.1434219196134417, + "grad_norm": 1.9711776971817017, + "learning_rate": 7.162458836443468e-07, + "loss": 1.0767, + "mean_token_accuracy": 0.6848935484886169, + "num_tokens": 33417588.0, + "step": 1306 + }, + { + "epoch": 0.14353173731605534, + "grad_norm": 2.2821695804595947, + "learning_rate": 7.167947310647639e-07, + "loss": 0.9748, + "mean_token_accuracy": 0.7278985381126404, + "num_tokens": 33441557.0, + "step": 1307 + }, + { + "epoch": 0.14364155501866901, + "grad_norm": 1.994834303855896, + "learning_rate": 7.173435784851811e-07, + "loss": 1.0967, + "mean_token_accuracy": 0.6787920594215393, + "num_tokens": 33472791.0, + "step": 1308 + }, + { + "epoch": 0.14375137272128266, + "grad_norm": 2.276148557662964, + "learning_rate": 7.178924259055982e-07, + "loss": 1.0568, + "mean_token_accuracy": 0.684862494468689, + "num_tokens": 33498357.0, + "step": 1309 + }, + { + "epoch": 0.14386119042389633, + "grad_norm": 2.256948947906494, + "learning_rate": 7.184412733260153e-07, + "loss": 0.9266, + "mean_token_accuracy": 0.7171564698219299, + "num_tokens": 33525858.0, + "step": 1310 + }, + { + "epoch": 0.14397100812650998, + "grad_norm": 2.33355712890625, + "learning_rate": 7.189901207464325e-07, + "loss": 1.042, + "mean_token_accuracy": 0.6870474815368652, + "num_tokens": 33551823.0, + "step": 1311 + }, + { + "epoch": 0.14408082582912365, + "grad_norm": 2.3002383708953857, + "learning_rate": 7.195389681668496e-07, + "loss": 1.0695, + "mean_token_accuracy": 0.6807148456573486, + "num_tokens": 33576622.0, + "step": 1312 + }, + { + "epoch": 0.14419064353173733, + "grad_norm": 2.63264536857605, + "learning_rate": 7.200878155872667e-07, + "loss": 0.9274, + "mean_token_accuracy": 0.7153497338294983, + "num_tokens": 33594269.0, + "step": 1313 + }, + { + "epoch": 0.14430046123435097, + "grad_norm": 2.4036896228790283, + "learning_rate": 7.206366630076838e-07, + "loss": 1.0533, + "mean_token_accuracy": 0.6856850385665894, + "num_tokens": 33617873.0, + "step": 1314 + }, + { + "epoch": 0.14441027893696465, + "grad_norm": 2.3204920291900635, + "learning_rate": 7.21185510428101e-07, + "loss": 1.0204, + "mean_token_accuracy": 0.6969514489173889, + "num_tokens": 33644990.0, + "step": 1315 + }, + { + "epoch": 0.1445200966395783, + "grad_norm": 2.091507911682129, + "learning_rate": 7.217343578485181e-07, + "loss": 0.9995, + "mean_token_accuracy": 0.7012112140655518, + "num_tokens": 33675451.0, + "step": 1316 + }, + { + "epoch": 0.14462991434219197, + "grad_norm": 2.2125790119171143, + "learning_rate": 7.222832052689352e-07, + "loss": 1.0025, + "mean_token_accuracy": 0.7056171894073486, + "num_tokens": 33702247.0, + "step": 1317 + }, + { + "epoch": 0.1447397320448056, + "grad_norm": 2.3502726554870605, + "learning_rate": 7.228320526893524e-07, + "loss": 1.0486, + "mean_token_accuracy": 0.6974613666534424, + "num_tokens": 33727216.0, + "step": 1318 + }, + { + "epoch": 0.1448495497474193, + "grad_norm": 2.0723743438720703, + "learning_rate": 7.233809001097695e-07, + "loss": 1.0755, + "mean_token_accuracy": 0.6792881488800049, + "num_tokens": 33757018.0, + "step": 1319 + }, + { + "epoch": 0.14495936745003293, + "grad_norm": 2.2987632751464844, + "learning_rate": 7.239297475301865e-07, + "loss": 1.1224, + "mean_token_accuracy": 0.6692900657653809, + "num_tokens": 33782381.0, + "step": 1320 + }, + { + "epoch": 0.1450691851526466, + "grad_norm": 1.8504960536956787, + "learning_rate": 7.244785949506037e-07, + "loss": 1.1332, + "mean_token_accuracy": 0.6668559312820435, + "num_tokens": 33820094.0, + "step": 1321 + }, + { + "epoch": 0.14517900285526028, + "grad_norm": 2.536787271499634, + "learning_rate": 7.250274423710208e-07, + "loss": 1.1068, + "mean_token_accuracy": 0.6757258176803589, + "num_tokens": 33843332.0, + "step": 1322 + }, + { + "epoch": 0.14528882055787393, + "grad_norm": 2.286051034927368, + "learning_rate": 7.255762897914379e-07, + "loss": 1.0659, + "mean_token_accuracy": 0.6853057146072388, + "num_tokens": 33870124.0, + "step": 1323 + }, + { + "epoch": 0.1453986382604876, + "grad_norm": 1.9955652952194214, + "learning_rate": 7.261251372118551e-07, + "loss": 1.1167, + "mean_token_accuracy": 0.6799165606498718, + "num_tokens": 33903801.0, + "step": 1324 + }, + { + "epoch": 0.14550845596310125, + "grad_norm": 2.2695164680480957, + "learning_rate": 7.266739846322721e-07, + "loss": 1.1094, + "mean_token_accuracy": 0.6667909622192383, + "num_tokens": 33931543.0, + "step": 1325 + }, + { + "epoch": 0.14561827366571492, + "grad_norm": 2.3519341945648193, + "learning_rate": 7.272228320526893e-07, + "loss": 0.896, + "mean_token_accuracy": 0.7263771891593933, + "num_tokens": 33955171.0, + "step": 1326 + }, + { + "epoch": 0.14572809136832857, + "grad_norm": 2.518554210662842, + "learning_rate": 7.277716794731065e-07, + "loss": 0.9775, + "mean_token_accuracy": 0.7033058404922485, + "num_tokens": 33976518.0, + "step": 1327 + }, + { + "epoch": 0.14583790907094224, + "grad_norm": 2.192896604537964, + "learning_rate": 7.283205268935235e-07, + "loss": 1.0133, + "mean_token_accuracy": 0.696617603302002, + "num_tokens": 34003249.0, + "step": 1328 + }, + { + "epoch": 0.14594772677355589, + "grad_norm": 2.492169141769409, + "learning_rate": 7.288693743139407e-07, + "loss": 0.9976, + "mean_token_accuracy": 0.6957709193229675, + "num_tokens": 34025314.0, + "step": 1329 + }, + { + "epoch": 0.14605754447616956, + "grad_norm": 2.4463696479797363, + "learning_rate": 7.294182217343579e-07, + "loss": 0.9962, + "mean_token_accuracy": 0.6968315243721008, + "num_tokens": 34048891.0, + "step": 1330 + }, + { + "epoch": 0.14616736217878323, + "grad_norm": 2.2113070487976074, + "learning_rate": 7.299670691547749e-07, + "loss": 1.0876, + "mean_token_accuracy": 0.6829423904418945, + "num_tokens": 34074590.0, + "step": 1331 + }, + { + "epoch": 0.14627717988139688, + "grad_norm": 2.498845338821411, + "learning_rate": 7.305159165751921e-07, + "loss": 1.0161, + "mean_token_accuracy": 0.6963306665420532, + "num_tokens": 34097270.0, + "step": 1332 + }, + { + "epoch": 0.14638699758401055, + "grad_norm": 2.340437173843384, + "learning_rate": 7.310647639956093e-07, + "loss": 1.0957, + "mean_token_accuracy": 0.6753911375999451, + "num_tokens": 34124553.0, + "step": 1333 + }, + { + "epoch": 0.1464968152866242, + "grad_norm": 2.4008939266204834, + "learning_rate": 7.316136114160263e-07, + "loss": 1.0622, + "mean_token_accuracy": 0.6840792894363403, + "num_tokens": 34149084.0, + "step": 1334 + }, + { + "epoch": 0.14660663298923787, + "grad_norm": 2.1405282020568848, + "learning_rate": 7.321624588364435e-07, + "loss": 0.9912, + "mean_token_accuracy": 0.7040966749191284, + "num_tokens": 34176894.0, + "step": 1335 + }, + { + "epoch": 0.14671645069185152, + "grad_norm": 2.1917731761932373, + "learning_rate": 7.327113062568605e-07, + "loss": 1.1343, + "mean_token_accuracy": 0.6680232286453247, + "num_tokens": 34205207.0, + "step": 1336 + }, + { + "epoch": 0.1468262683944652, + "grad_norm": 2.6403558254241943, + "learning_rate": 7.332601536772777e-07, + "loss": 1.1021, + "mean_token_accuracy": 0.681955099105835, + "num_tokens": 34227580.0, + "step": 1337 + }, + { + "epoch": 0.14693608609707884, + "grad_norm": 2.2966010570526123, + "learning_rate": 7.338090010976949e-07, + "loss": 1.0375, + "mean_token_accuracy": 0.6919702291488647, + "num_tokens": 34253589.0, + "step": 1338 + }, + { + "epoch": 0.1470459037996925, + "grad_norm": 2.4379212856292725, + "learning_rate": 7.343578485181118e-07, + "loss": 1.0882, + "mean_token_accuracy": 0.6780909299850464, + "num_tokens": 34278078.0, + "step": 1339 + }, + { + "epoch": 0.14715572150230616, + "grad_norm": 2.516477584838867, + "learning_rate": 7.34906695938529e-07, + "loss": 1.0764, + "mean_token_accuracy": 0.6756150126457214, + "num_tokens": 34298594.0, + "step": 1340 + }, + { + "epoch": 0.14726553920491983, + "grad_norm": 2.2875783443450928, + "learning_rate": 7.354555433589462e-07, + "loss": 0.9304, + "mean_token_accuracy": 0.7139771580696106, + "num_tokens": 34320528.0, + "step": 1341 + }, + { + "epoch": 0.1473753569075335, + "grad_norm": 2.240898609161377, + "learning_rate": 7.360043907793632e-07, + "loss": 1.0515, + "mean_token_accuracy": 0.684171736240387, + "num_tokens": 34346542.0, + "step": 1342 + }, + { + "epoch": 0.14748517461014715, + "grad_norm": 2.221475601196289, + "learning_rate": 7.365532381997804e-07, + "loss": 1.0988, + "mean_token_accuracy": 0.6684052348136902, + "num_tokens": 34374764.0, + "step": 1343 + }, + { + "epoch": 0.14759499231276083, + "grad_norm": 2.465275287628174, + "learning_rate": 7.371020856201976e-07, + "loss": 0.9299, + "mean_token_accuracy": 0.7105779647827148, + "num_tokens": 34396354.0, + "step": 1344 + }, + { + "epoch": 0.14770481001537447, + "grad_norm": 2.2361974716186523, + "learning_rate": 7.376509330406146e-07, + "loss": 1.0183, + "mean_token_accuracy": 0.7005178928375244, + "num_tokens": 34423119.0, + "step": 1345 + }, + { + "epoch": 0.14781462771798815, + "grad_norm": 2.361527442932129, + "learning_rate": 7.381997804610318e-07, + "loss": 1.0805, + "mean_token_accuracy": 0.6887157559394836, + "num_tokens": 34447502.0, + "step": 1346 + }, + { + "epoch": 0.1479244454206018, + "grad_norm": 2.3615221977233887, + "learning_rate": 7.387486278814489e-07, + "loss": 1.0446, + "mean_token_accuracy": 0.6899142861366272, + "num_tokens": 34471812.0, + "step": 1347 + }, + { + "epoch": 0.14803426312321546, + "grad_norm": 2.3949358463287354, + "learning_rate": 7.39297475301866e-07, + "loss": 1.1041, + "mean_token_accuracy": 0.6717970371246338, + "num_tokens": 34496679.0, + "step": 1348 + }, + { + "epoch": 0.1481440808258291, + "grad_norm": 2.1916120052337646, + "learning_rate": 7.398463227222832e-07, + "loss": 1.062, + "mean_token_accuracy": 0.683023989200592, + "num_tokens": 34524948.0, + "step": 1349 + }, + { + "epoch": 0.14825389852844278, + "grad_norm": 2.552324056625366, + "learning_rate": 7.403951701427003e-07, + "loss": 1.0149, + "mean_token_accuracy": 0.6931304931640625, + "num_tokens": 34547161.0, + "step": 1350 + }, + { + "epoch": 0.14836371623105646, + "grad_norm": 2.1980385780334473, + "learning_rate": 7.409440175631174e-07, + "loss": 1.0972, + "mean_token_accuracy": 0.6788745522499084, + "num_tokens": 34575281.0, + "step": 1351 + }, + { + "epoch": 0.1484735339336701, + "grad_norm": 2.4990243911743164, + "learning_rate": 7.414928649835346e-07, + "loss": 0.9786, + "mean_token_accuracy": 0.7033196687698364, + "num_tokens": 34595820.0, + "step": 1352 + }, + { + "epoch": 0.14858335163628378, + "grad_norm": 2.9267513751983643, + "learning_rate": 7.420417124039517e-07, + "loss": 1.0065, + "mean_token_accuracy": 0.6931508779525757, + "num_tokens": 34623495.0, + "step": 1353 + }, + { + "epoch": 0.14869316933889742, + "grad_norm": 2.266777753829956, + "learning_rate": 7.425905598243688e-07, + "loss": 0.9558, + "mean_token_accuracy": 0.7093894481658936, + "num_tokens": 34647340.0, + "step": 1354 + }, + { + "epoch": 0.1488029870415111, + "grad_norm": 2.398285388946533, + "learning_rate": 7.43139407244786e-07, + "loss": 1.0876, + "mean_token_accuracy": 0.6784405708312988, + "num_tokens": 34672008.0, + "step": 1355 + }, + { + "epoch": 0.14891280474412474, + "grad_norm": 2.4099478721618652, + "learning_rate": 7.436882546652031e-07, + "loss": 1.005, + "mean_token_accuracy": 0.698257565498352, + "num_tokens": 34696275.0, + "step": 1356 + }, + { + "epoch": 0.14902262244673842, + "grad_norm": 2.171253204345703, + "learning_rate": 7.442371020856202e-07, + "loss": 1.0332, + "mean_token_accuracy": 0.6911093592643738, + "num_tokens": 34724874.0, + "step": 1357 + }, + { + "epoch": 0.14913244014935206, + "grad_norm": 2.438288450241089, + "learning_rate": 7.447859495060372e-07, + "loss": 1.0593, + "mean_token_accuracy": 0.6897088885307312, + "num_tokens": 34747688.0, + "step": 1358 + }, + { + "epoch": 0.14924225785196574, + "grad_norm": 2.151923894882202, + "learning_rate": 7.453347969264544e-07, + "loss": 0.9965, + "mean_token_accuracy": 0.6947108507156372, + "num_tokens": 34774221.0, + "step": 1359 + }, + { + "epoch": 0.1493520755545794, + "grad_norm": 2.36879825592041, + "learning_rate": 7.458836443468715e-07, + "loss": 1.0363, + "mean_token_accuracy": 0.6932151317596436, + "num_tokens": 34798725.0, + "step": 1360 + }, + { + "epoch": 0.14946189325719306, + "grad_norm": 2.269984483718872, + "learning_rate": 7.464324917672886e-07, + "loss": 1.0399, + "mean_token_accuracy": 0.6848617196083069, + "num_tokens": 34827932.0, + "step": 1361 + }, + { + "epoch": 0.14957171095980673, + "grad_norm": 2.3656415939331055, + "learning_rate": 7.469813391877058e-07, + "loss": 1.0711, + "mean_token_accuracy": 0.6790351271629333, + "num_tokens": 34854085.0, + "step": 1362 + }, + { + "epoch": 0.14968152866242038, + "grad_norm": 2.100402593612671, + "learning_rate": 7.475301866081229e-07, + "loss": 1.0927, + "mean_token_accuracy": 0.6759006977081299, + "num_tokens": 34883859.0, + "step": 1363 + }, + { + "epoch": 0.14979134636503405, + "grad_norm": 2.165574073791504, + "learning_rate": 7.4807903402854e-07, + "loss": 0.9863, + "mean_token_accuracy": 0.7028398513793945, + "num_tokens": 34912747.0, + "step": 1364 + }, + { + "epoch": 0.1499011640676477, + "grad_norm": 2.107786178588867, + "learning_rate": 7.486278814489572e-07, + "loss": 1.0514, + "mean_token_accuracy": 0.6835331916809082, + "num_tokens": 34943286.0, + "step": 1365 + }, + { + "epoch": 0.15001098177026137, + "grad_norm": 2.201176404953003, + "learning_rate": 7.491767288693743e-07, + "loss": 1.1182, + "mean_token_accuracy": 0.6792147159576416, + "num_tokens": 34972438.0, + "step": 1366 + }, + { + "epoch": 0.15012079947287502, + "grad_norm": 2.47337007522583, + "learning_rate": 7.497255762897914e-07, + "loss": 0.9761, + "mean_token_accuracy": 0.7052915096282959, + "num_tokens": 34995462.0, + "step": 1367 + }, + { + "epoch": 0.1502306171754887, + "grad_norm": 2.240569829940796, + "learning_rate": 7.502744237102086e-07, + "loss": 1.0178, + "mean_token_accuracy": 0.6985488533973694, + "num_tokens": 35023267.0, + "step": 1368 + }, + { + "epoch": 0.15034043487810236, + "grad_norm": 2.3921992778778076, + "learning_rate": 7.508232711306256e-07, + "loss": 1.0501, + "mean_token_accuracy": 0.6850267648696899, + "num_tokens": 35046683.0, + "step": 1369 + }, + { + "epoch": 0.150450252580716, + "grad_norm": 2.394670248031616, + "learning_rate": 7.513721185510428e-07, + "loss": 1.0718, + "mean_token_accuracy": 0.6800662279129028, + "num_tokens": 35074180.0, + "step": 1370 + }, + { + "epoch": 0.15056007028332968, + "grad_norm": 2.357287645339966, + "learning_rate": 7.5192096597146e-07, + "loss": 1.0063, + "mean_token_accuracy": 0.7034634351730347, + "num_tokens": 35098797.0, + "step": 1371 + }, + { + "epoch": 0.15066988798594333, + "grad_norm": 2.4405698776245117, + "learning_rate": 7.52469813391877e-07, + "loss": 1.0675, + "mean_token_accuracy": 0.6823084354400635, + "num_tokens": 35121751.0, + "step": 1372 + }, + { + "epoch": 0.150779705688557, + "grad_norm": 2.1377265453338623, + "learning_rate": 7.530186608122942e-07, + "loss": 1.1681, + "mean_token_accuracy": 0.6612858176231384, + "num_tokens": 35153610.0, + "step": 1373 + }, + { + "epoch": 0.15088952339117065, + "grad_norm": 2.7084662914276123, + "learning_rate": 7.535675082327113e-07, + "loss": 1.0379, + "mean_token_accuracy": 0.6889113187789917, + "num_tokens": 35174720.0, + "step": 1374 + }, + { + "epoch": 0.15099934109378432, + "grad_norm": 2.4333417415618896, + "learning_rate": 7.541163556531284e-07, + "loss": 0.999, + "mean_token_accuracy": 0.7011537551879883, + "num_tokens": 35197451.0, + "step": 1375 + }, + { + "epoch": 0.15110915879639797, + "grad_norm": 2.427851915359497, + "learning_rate": 7.546652030735456e-07, + "loss": 1.1149, + "mean_token_accuracy": 0.6851558089256287, + "num_tokens": 35221491.0, + "step": 1376 + }, + { + "epoch": 0.15121897649901164, + "grad_norm": 2.3089654445648193, + "learning_rate": 7.552140504939627e-07, + "loss": 0.974, + "mean_token_accuracy": 0.7027887105941772, + "num_tokens": 35245085.0, + "step": 1377 + }, + { + "epoch": 0.1513287942016253, + "grad_norm": 2.2013344764709473, + "learning_rate": 7.557628979143797e-07, + "loss": 1.0822, + "mean_token_accuracy": 0.6826717853546143, + "num_tokens": 35271259.0, + "step": 1378 + }, + { + "epoch": 0.15143861190423896, + "grad_norm": 2.4194629192352295, + "learning_rate": 7.563117453347969e-07, + "loss": 1.0718, + "mean_token_accuracy": 0.6824245452880859, + "num_tokens": 35294547.0, + "step": 1379 + }, + { + "epoch": 0.15154842960685264, + "grad_norm": 2.1522316932678223, + "learning_rate": 7.568605927552139e-07, + "loss": 1.0521, + "mean_token_accuracy": 0.6866676807403564, + "num_tokens": 35322171.0, + "step": 1380 + }, + { + "epoch": 0.15165824730946628, + "grad_norm": 2.4162683486938477, + "learning_rate": 7.574094401756311e-07, + "loss": 0.972, + "mean_token_accuracy": 0.7064235806465149, + "num_tokens": 35345927.0, + "step": 1381 + }, + { + "epoch": 0.15176806501207996, + "grad_norm": 2.335517168045044, + "learning_rate": 7.579582875960483e-07, + "loss": 1.128, + "mean_token_accuracy": 0.6613811254501343, + "num_tokens": 35370276.0, + "step": 1382 + }, + { + "epoch": 0.1518778827146936, + "grad_norm": 2.2815017700195312, + "learning_rate": 7.585071350164653e-07, + "loss": 1.0807, + "mean_token_accuracy": 0.676624059677124, + "num_tokens": 35398068.0, + "step": 1383 + }, + { + "epoch": 0.15198770041730728, + "grad_norm": 2.203547716140747, + "learning_rate": 7.590559824368825e-07, + "loss": 0.9669, + "mean_token_accuracy": 0.7102487087249756, + "num_tokens": 35422598.0, + "step": 1384 + }, + { + "epoch": 0.15209751811992092, + "grad_norm": 2.11283802986145, + "learning_rate": 7.596048298572997e-07, + "loss": 1.1152, + "mean_token_accuracy": 0.6688425540924072, + "num_tokens": 35452205.0, + "step": 1385 + }, + { + "epoch": 0.1522073358225346, + "grad_norm": 2.474090099334717, + "learning_rate": 7.601536772777167e-07, + "loss": 1.0215, + "mean_token_accuracy": 0.6927486062049866, + "num_tokens": 35476687.0, + "step": 1386 + }, + { + "epoch": 0.15231715352514824, + "grad_norm": 2.0087411403656006, + "learning_rate": 7.607025246981339e-07, + "loss": 1.0875, + "mean_token_accuracy": 0.6789435148239136, + "num_tokens": 35507906.0, + "step": 1387 + }, + { + "epoch": 0.15242697122776191, + "grad_norm": 2.213932991027832, + "learning_rate": 7.612513721185511e-07, + "loss": 1.0174, + "mean_token_accuracy": 0.6962790489196777, + "num_tokens": 35534123.0, + "step": 1388 + }, + { + "epoch": 0.1525367889303756, + "grad_norm": 2.351181983947754, + "learning_rate": 7.618002195389681e-07, + "loss": 1.1028, + "mean_token_accuracy": 0.6732299327850342, + "num_tokens": 35560550.0, + "step": 1389 + }, + { + "epoch": 0.15264660663298923, + "grad_norm": 2.2623322010040283, + "learning_rate": 7.623490669593853e-07, + "loss": 1.0538, + "mean_token_accuracy": 0.6867098808288574, + "num_tokens": 35588398.0, + "step": 1390 + }, + { + "epoch": 0.1527564243356029, + "grad_norm": 2.4374067783355713, + "learning_rate": 7.628979143798024e-07, + "loss": 1.0212, + "mean_token_accuracy": 0.6924480199813843, + "num_tokens": 35611907.0, + "step": 1391 + }, + { + "epoch": 0.15286624203821655, + "grad_norm": 2.359165906906128, + "learning_rate": 7.634467618002195e-07, + "loss": 1.0564, + "mean_token_accuracy": 0.6863903403282166, + "num_tokens": 35637511.0, + "step": 1392 + }, + { + "epoch": 0.15297605974083023, + "grad_norm": 2.4319260120391846, + "learning_rate": 7.639956092206367e-07, + "loss": 1.0562, + "mean_token_accuracy": 0.6857137680053711, + "num_tokens": 35661367.0, + "step": 1393 + }, + { + "epoch": 0.15308587744344387, + "grad_norm": 2.409878730773926, + "learning_rate": 7.645444566410538e-07, + "loss": 1.1275, + "mean_token_accuracy": 0.6650971174240112, + "num_tokens": 35685848.0, + "step": 1394 + }, + { + "epoch": 0.15319569514605755, + "grad_norm": 2.19256591796875, + "learning_rate": 7.650933040614709e-07, + "loss": 0.9374, + "mean_token_accuracy": 0.7278707027435303, + "num_tokens": 35711883.0, + "step": 1395 + }, + { + "epoch": 0.1533055128486712, + "grad_norm": 2.171542167663574, + "learning_rate": 7.656421514818881e-07, + "loss": 1.0525, + "mean_token_accuracy": 0.6891616582870483, + "num_tokens": 35740512.0, + "step": 1396 + }, + { + "epoch": 0.15341533055128487, + "grad_norm": 2.333730459213257, + "learning_rate": 7.66190998902305e-07, + "loss": 1.0241, + "mean_token_accuracy": 0.7004289031028748, + "num_tokens": 35766631.0, + "step": 1397 + }, + { + "epoch": 0.15352514825389854, + "grad_norm": 1.9566134214401245, + "learning_rate": 7.667398463227222e-07, + "loss": 1.0491, + "mean_token_accuracy": 0.6843339204788208, + "num_tokens": 35800096.0, + "step": 1398 + }, + { + "epoch": 0.1536349659565122, + "grad_norm": 2.18318247795105, + "learning_rate": 7.672886937431394e-07, + "loss": 1.1281, + "mean_token_accuracy": 0.6651973724365234, + "num_tokens": 35829091.0, + "step": 1399 + }, + { + "epoch": 0.15374478365912586, + "grad_norm": 2.5336344242095947, + "learning_rate": 7.678375411635564e-07, + "loss": 1.0426, + "mean_token_accuracy": 0.689444899559021, + "num_tokens": 35851121.0, + "step": 1400 + }, + { + "epoch": 0.1538546013617395, + "grad_norm": 2.4422788619995117, + "learning_rate": 7.683863885839736e-07, + "loss": 1.0063, + "mean_token_accuracy": 0.699419379234314, + "num_tokens": 35872998.0, + "step": 1401 + }, + { + "epoch": 0.15396441906435318, + "grad_norm": 2.1176936626434326, + "learning_rate": 7.689352360043907e-07, + "loss": 1.0465, + "mean_token_accuracy": 0.6841087937355042, + "num_tokens": 35901839.0, + "step": 1402 + }, + { + "epoch": 0.15407423676696683, + "grad_norm": 2.300689935684204, + "learning_rate": 7.694840834248078e-07, + "loss": 1.0984, + "mean_token_accuracy": 0.6861339807510376, + "num_tokens": 35927359.0, + "step": 1403 + }, + { + "epoch": 0.1541840544695805, + "grad_norm": 2.409851551055908, + "learning_rate": 7.70032930845225e-07, + "loss": 1.1091, + "mean_token_accuracy": 0.6703145503997803, + "num_tokens": 35950373.0, + "step": 1404 + }, + { + "epoch": 0.15429387217219415, + "grad_norm": 2.1762757301330566, + "learning_rate": 7.705817782656421e-07, + "loss": 0.9663, + "mean_token_accuracy": 0.7073551416397095, + "num_tokens": 35976503.0, + "step": 1405 + }, + { + "epoch": 0.15440368987480782, + "grad_norm": 2.4222311973571777, + "learning_rate": 7.711306256860592e-07, + "loss": 1.0391, + "mean_token_accuracy": 0.6899356842041016, + "num_tokens": 35998990.0, + "step": 1406 + }, + { + "epoch": 0.1545135075774215, + "grad_norm": 2.7073893547058105, + "learning_rate": 7.716794731064764e-07, + "loss": 1.0404, + "mean_token_accuracy": 0.685185968875885, + "num_tokens": 36020238.0, + "step": 1407 + }, + { + "epoch": 0.15462332528003514, + "grad_norm": 2.542309284210205, + "learning_rate": 7.722283205268935e-07, + "loss": 1.0175, + "mean_token_accuracy": 0.6987132430076599, + "num_tokens": 36042472.0, + "step": 1408 + }, + { + "epoch": 0.1547331429826488, + "grad_norm": 2.39123797416687, + "learning_rate": 7.727771679473106e-07, + "loss": 1.0388, + "mean_token_accuracy": 0.6880583763122559, + "num_tokens": 36066792.0, + "step": 1409 + }, + { + "epoch": 0.15484296068526246, + "grad_norm": 2.403139114379883, + "learning_rate": 7.733260153677278e-07, + "loss": 1.114, + "mean_token_accuracy": 0.6782624125480652, + "num_tokens": 36093894.0, + "step": 1410 + }, + { + "epoch": 0.15495277838787613, + "grad_norm": 2.3975536823272705, + "learning_rate": 7.738748627881449e-07, + "loss": 1.0732, + "mean_token_accuracy": 0.6882598400115967, + "num_tokens": 36118444.0, + "step": 1411 + }, + { + "epoch": 0.15506259609048978, + "grad_norm": 2.3829686641693115, + "learning_rate": 7.74423710208562e-07, + "loss": 1.0247, + "mean_token_accuracy": 0.7038663625717163, + "num_tokens": 36144533.0, + "step": 1412 + }, + { + "epoch": 0.15517241379310345, + "grad_norm": 2.1775951385498047, + "learning_rate": 7.749725576289791e-07, + "loss": 1.0618, + "mean_token_accuracy": 0.6792048215866089, + "num_tokens": 36170853.0, + "step": 1413 + }, + { + "epoch": 0.1552822314957171, + "grad_norm": 2.4269628524780273, + "learning_rate": 7.755214050493963e-07, + "loss": 1.0482, + "mean_token_accuracy": 0.6881431341171265, + "num_tokens": 36192722.0, + "step": 1414 + }, + { + "epoch": 0.15539204919833077, + "grad_norm": 2.14896821975708, + "learning_rate": 7.760702524698134e-07, + "loss": 0.9432, + "mean_token_accuracy": 0.7119646072387695, + "num_tokens": 36217488.0, + "step": 1415 + }, + { + "epoch": 0.15550186690094442, + "grad_norm": 2.281989336013794, + "learning_rate": 7.766190998902304e-07, + "loss": 1.0038, + "mean_token_accuracy": 0.7052807807922363, + "num_tokens": 36242783.0, + "step": 1416 + }, + { + "epoch": 0.1556116846035581, + "grad_norm": 2.2688963413238525, + "learning_rate": 7.771679473106476e-07, + "loss": 0.9757, + "mean_token_accuracy": 0.7089031934738159, + "num_tokens": 36266545.0, + "step": 1417 + }, + { + "epoch": 0.15572150230617177, + "grad_norm": 2.2435624599456787, + "learning_rate": 7.777167947310647e-07, + "loss": 1.0752, + "mean_token_accuracy": 0.6881165504455566, + "num_tokens": 36293425.0, + "step": 1418 + }, + { + "epoch": 0.1558313200087854, + "grad_norm": 2.279756784439087, + "learning_rate": 7.782656421514818e-07, + "loss": 1.0254, + "mean_token_accuracy": 0.699083685874939, + "num_tokens": 36319359.0, + "step": 1419 + }, + { + "epoch": 0.15594113771139909, + "grad_norm": 2.0249745845794678, + "learning_rate": 7.78814489571899e-07, + "loss": 1.1102, + "mean_token_accuracy": 0.6726821660995483, + "num_tokens": 36354066.0, + "step": 1420 + }, + { + "epoch": 0.15605095541401273, + "grad_norm": 2.161832094192505, + "learning_rate": 7.793633369923161e-07, + "loss": 1.075, + "mean_token_accuracy": 0.6790690422058105, + "num_tokens": 36383569.0, + "step": 1421 + }, + { + "epoch": 0.1561607731166264, + "grad_norm": 2.543865203857422, + "learning_rate": 7.799121844127332e-07, + "loss": 1.131, + "mean_token_accuracy": 0.6659165620803833, + "num_tokens": 36406590.0, + "step": 1422 + }, + { + "epoch": 0.15627059081924005, + "grad_norm": 2.18678879737854, + "learning_rate": 7.804610318331504e-07, + "loss": 1.1366, + "mean_token_accuracy": 0.6620327234268188, + "num_tokens": 36436487.0, + "step": 1423 + }, + { + "epoch": 0.15638040852185373, + "grad_norm": 2.363878011703491, + "learning_rate": 7.810098792535674e-07, + "loss": 1.0224, + "mean_token_accuracy": 0.694301962852478, + "num_tokens": 36460530.0, + "step": 1424 + }, + { + "epoch": 0.15649022622446737, + "grad_norm": 2.35509991645813, + "learning_rate": 7.815587266739846e-07, + "loss": 1.034, + "mean_token_accuracy": 0.6978099346160889, + "num_tokens": 36484648.0, + "step": 1425 + }, + { + "epoch": 0.15660004392708105, + "grad_norm": 2.2582151889801025, + "learning_rate": 7.821075740944018e-07, + "loss": 1.0699, + "mean_token_accuracy": 0.6902559995651245, + "num_tokens": 36512182.0, + "step": 1426 + }, + { + "epoch": 0.15670986162969472, + "grad_norm": 2.046205997467041, + "learning_rate": 7.826564215148188e-07, + "loss": 1.0908, + "mean_token_accuracy": 0.6806096434593201, + "num_tokens": 36544812.0, + "step": 1427 + }, + { + "epoch": 0.15681967933230836, + "grad_norm": 2.40031361579895, + "learning_rate": 7.83205268935236e-07, + "loss": 1.0897, + "mean_token_accuracy": 0.6814260482788086, + "num_tokens": 36568379.0, + "step": 1428 + }, + { + "epoch": 0.15692949703492204, + "grad_norm": 2.1492884159088135, + "learning_rate": 7.837541163556532e-07, + "loss": 1.0065, + "mean_token_accuracy": 0.6975261569023132, + "num_tokens": 36595866.0, + "step": 1429 + }, + { + "epoch": 0.15703931473753568, + "grad_norm": 2.1926279067993164, + "learning_rate": 7.843029637760702e-07, + "loss": 0.9735, + "mean_token_accuracy": 0.7179089784622192, + "num_tokens": 36622139.0, + "step": 1430 + }, + { + "epoch": 0.15714913244014936, + "grad_norm": 2.3999438285827637, + "learning_rate": 7.848518111964874e-07, + "loss": 0.9115, + "mean_token_accuracy": 0.7171493768692017, + "num_tokens": 36643046.0, + "step": 1431 + }, + { + "epoch": 0.157258950142763, + "grad_norm": 2.4599931240081787, + "learning_rate": 7.854006586169045e-07, + "loss": 1.0522, + "mean_token_accuracy": 0.6847100257873535, + "num_tokens": 36665414.0, + "step": 1432 + }, + { + "epoch": 0.15736876784537668, + "grad_norm": 2.4719014167785645, + "learning_rate": 7.859495060373216e-07, + "loss": 1.1113, + "mean_token_accuracy": 0.6728964447975159, + "num_tokens": 36689941.0, + "step": 1433 + }, + { + "epoch": 0.15747858554799032, + "grad_norm": 2.2554895877838135, + "learning_rate": 7.864983534577388e-07, + "loss": 0.9686, + "mean_token_accuracy": 0.7083142995834351, + "num_tokens": 36715232.0, + "step": 1434 + }, + { + "epoch": 0.157588403250604, + "grad_norm": 2.533184766769409, + "learning_rate": 7.870472008781557e-07, + "loss": 1.0322, + "mean_token_accuracy": 0.6972748041152954, + "num_tokens": 36739283.0, + "step": 1435 + }, + { + "epoch": 0.15769822095321767, + "grad_norm": 2.322451591491699, + "learning_rate": 7.875960482985729e-07, + "loss": 0.9423, + "mean_token_accuracy": 0.7204725742340088, + "num_tokens": 36763007.0, + "step": 1436 + }, + { + "epoch": 0.15780803865583132, + "grad_norm": 2.2173755168914795, + "learning_rate": 7.881448957189901e-07, + "loss": 1.0947, + "mean_token_accuracy": 0.6716208457946777, + "num_tokens": 36791427.0, + "step": 1437 + }, + { + "epoch": 0.157917856358445, + "grad_norm": 2.178089141845703, + "learning_rate": 7.886937431394071e-07, + "loss": 1.0822, + "mean_token_accuracy": 0.6733913421630859, + "num_tokens": 36820360.0, + "step": 1438 + }, + { + "epoch": 0.15802767406105864, + "grad_norm": 2.375110149383545, + "learning_rate": 7.892425905598243e-07, + "loss": 0.991, + "mean_token_accuracy": 0.6982769966125488, + "num_tokens": 36844106.0, + "step": 1439 + }, + { + "epoch": 0.1581374917636723, + "grad_norm": 2.4676761627197266, + "learning_rate": 7.897914379802415e-07, + "loss": 0.939, + "mean_token_accuracy": 0.7144315242767334, + "num_tokens": 36867433.0, + "step": 1440 + }, + { + "epoch": 0.15824730946628596, + "grad_norm": 2.3624844551086426, + "learning_rate": 7.903402854006585e-07, + "loss": 0.9798, + "mean_token_accuracy": 0.706313967704773, + "num_tokens": 36891565.0, + "step": 1441 + }, + { + "epoch": 0.15835712716889963, + "grad_norm": 2.5611250400543213, + "learning_rate": 7.908891328210757e-07, + "loss": 1.0604, + "mean_token_accuracy": 0.6876197457313538, + "num_tokens": 36915341.0, + "step": 1442 + }, + { + "epoch": 0.15846694487151328, + "grad_norm": 2.466719388961792, + "learning_rate": 7.914379802414928e-07, + "loss": 0.9121, + "mean_token_accuracy": 0.7212693095207214, + "num_tokens": 36937544.0, + "step": 1443 + }, + { + "epoch": 0.15857676257412695, + "grad_norm": 2.075688123703003, + "learning_rate": 7.919868276619099e-07, + "loss": 0.9397, + "mean_token_accuracy": 0.711883544921875, + "num_tokens": 36967522.0, + "step": 1444 + }, + { + "epoch": 0.15868658027674062, + "grad_norm": 2.2801620960235596, + "learning_rate": 7.925356750823271e-07, + "loss": 1.0408, + "mean_token_accuracy": 0.6973063349723816, + "num_tokens": 36991885.0, + "step": 1445 + }, + { + "epoch": 0.15879639797935427, + "grad_norm": 1.9283339977264404, + "learning_rate": 7.930845225027442e-07, + "loss": 0.9815, + "mean_token_accuracy": 0.6996743679046631, + "num_tokens": 37022937.0, + "step": 1446 + }, + { + "epoch": 0.15890621568196794, + "grad_norm": 2.2146618366241455, + "learning_rate": 7.936333699231613e-07, + "loss": 1.0945, + "mean_token_accuracy": 0.6751668453216553, + "num_tokens": 37050625.0, + "step": 1447 + }, + { + "epoch": 0.1590160333845816, + "grad_norm": 2.5292434692382812, + "learning_rate": 7.941822173435785e-07, + "loss": 0.9984, + "mean_token_accuracy": 0.7004355788230896, + "num_tokens": 37072584.0, + "step": 1448 + }, + { + "epoch": 0.15912585108719526, + "grad_norm": 2.3118674755096436, + "learning_rate": 7.947310647639956e-07, + "loss": 1.0463, + "mean_token_accuracy": 0.6890068054199219, + "num_tokens": 37096815.0, + "step": 1449 + }, + { + "epoch": 0.1592356687898089, + "grad_norm": 2.2674665451049805, + "learning_rate": 7.952799121844127e-07, + "loss": 0.966, + "mean_token_accuracy": 0.7083832025527954, + "num_tokens": 37121396.0, + "step": 1450 + }, + { + "epoch": 0.15934548649242258, + "grad_norm": 2.4418118000030518, + "learning_rate": 7.958287596048299e-07, + "loss": 1.0309, + "mean_token_accuracy": 0.6856430768966675, + "num_tokens": 37143979.0, + "step": 1451 + }, + { + "epoch": 0.15945530419503623, + "grad_norm": 2.111048936843872, + "learning_rate": 7.96377607025247e-07, + "loss": 0.9655, + "mean_token_accuracy": 0.7110364437103271, + "num_tokens": 37172218.0, + "step": 1452 + }, + { + "epoch": 0.1595651218976499, + "grad_norm": 2.3659422397613525, + "learning_rate": 7.969264544456641e-07, + "loss": 1.0568, + "mean_token_accuracy": 0.6893777847290039, + "num_tokens": 37195769.0, + "step": 1453 + }, + { + "epoch": 0.15967493960026355, + "grad_norm": 2.393418312072754, + "learning_rate": 7.974753018660811e-07, + "loss": 1.1456, + "mean_token_accuracy": 0.6674643754959106, + "num_tokens": 37222382.0, + "step": 1454 + }, + { + "epoch": 0.15978475730287722, + "grad_norm": 2.1485018730163574, + "learning_rate": 7.980241492864983e-07, + "loss": 1.0502, + "mean_token_accuracy": 0.6928986310958862, + "num_tokens": 37250443.0, + "step": 1455 + }, + { + "epoch": 0.1598945750054909, + "grad_norm": 2.310121536254883, + "learning_rate": 7.985729967069154e-07, + "loss": 1.0315, + "mean_token_accuracy": 0.6877254247665405, + "num_tokens": 37278369.0, + "step": 1456 + }, + { + "epoch": 0.16000439270810454, + "grad_norm": 2.4250996112823486, + "learning_rate": 7.991218441273325e-07, + "loss": 1.0002, + "mean_token_accuracy": 0.6926387548446655, + "num_tokens": 37302702.0, + "step": 1457 + }, + { + "epoch": 0.16011421041071822, + "grad_norm": 2.461282253265381, + "learning_rate": 7.996706915477497e-07, + "loss": 1.0209, + "mean_token_accuracy": 0.697954535484314, + "num_tokens": 37325366.0, + "step": 1458 + }, + { + "epoch": 0.16022402811333186, + "grad_norm": 2.0658340454101562, + "learning_rate": 8.002195389681668e-07, + "loss": 1.0256, + "mean_token_accuracy": 0.695076584815979, + "num_tokens": 37356411.0, + "step": 1459 + }, + { + "epoch": 0.16033384581594554, + "grad_norm": 2.3327627182006836, + "learning_rate": 8.007683863885839e-07, + "loss": 1.0698, + "mean_token_accuracy": 0.6891471743583679, + "num_tokens": 37382504.0, + "step": 1460 + }, + { + "epoch": 0.16044366351855918, + "grad_norm": 2.1679420471191406, + "learning_rate": 8.013172338090011e-07, + "loss": 0.9803, + "mean_token_accuracy": 0.7095222473144531, + "num_tokens": 37409137.0, + "step": 1461 + }, + { + "epoch": 0.16055348122117286, + "grad_norm": 2.542983293533325, + "learning_rate": 8.018660812294182e-07, + "loss": 1.096, + "mean_token_accuracy": 0.6835992336273193, + "num_tokens": 37431336.0, + "step": 1462 + }, + { + "epoch": 0.1606632989237865, + "grad_norm": 2.2127132415771484, + "learning_rate": 8.024149286498353e-07, + "loss": 0.9901, + "mean_token_accuracy": 0.6993728280067444, + "num_tokens": 37460023.0, + "step": 1463 + }, + { + "epoch": 0.16077311662640018, + "grad_norm": 2.1983766555786133, + "learning_rate": 8.029637760702525e-07, + "loss": 1.0293, + "mean_token_accuracy": 0.6919887661933899, + "num_tokens": 37487790.0, + "step": 1464 + }, + { + "epoch": 0.16088293432901385, + "grad_norm": 2.1893603801727295, + "learning_rate": 8.035126234906695e-07, + "loss": 1.0341, + "mean_token_accuracy": 0.6870423555374146, + "num_tokens": 37515709.0, + "step": 1465 + }, + { + "epoch": 0.1609927520316275, + "grad_norm": 2.0602047443389893, + "learning_rate": 8.040614709110867e-07, + "loss": 1.0018, + "mean_token_accuracy": 0.7055994272232056, + "num_tokens": 37545478.0, + "step": 1466 + }, + { + "epoch": 0.16110256973424117, + "grad_norm": 2.2675390243530273, + "learning_rate": 8.046103183315039e-07, + "loss": 0.9667, + "mean_token_accuracy": 0.7069530487060547, + "num_tokens": 37571380.0, + "step": 1467 + }, + { + "epoch": 0.16121238743685481, + "grad_norm": 2.349660634994507, + "learning_rate": 8.051591657519209e-07, + "loss": 0.9921, + "mean_token_accuracy": 0.7023035883903503, + "num_tokens": 37595668.0, + "step": 1468 + }, + { + "epoch": 0.1613222051394685, + "grad_norm": 2.251199722290039, + "learning_rate": 8.057080131723381e-07, + "loss": 1.1439, + "mean_token_accuracy": 0.6587263941764832, + "num_tokens": 37625130.0, + "step": 1469 + }, + { + "epoch": 0.16143202284208213, + "grad_norm": 2.4719390869140625, + "learning_rate": 8.062568605927553e-07, + "loss": 1.0082, + "mean_token_accuracy": 0.6975260376930237, + "num_tokens": 37647663.0, + "step": 1470 + }, + { + "epoch": 0.1615418405446958, + "grad_norm": 2.4728593826293945, + "learning_rate": 8.068057080131723e-07, + "loss": 0.9819, + "mean_token_accuracy": 0.7026883959770203, + "num_tokens": 37668577.0, + "step": 1471 + }, + { + "epoch": 0.16165165824730945, + "grad_norm": 2.555386781692505, + "learning_rate": 8.073545554335894e-07, + "loss": 0.9938, + "mean_token_accuracy": 0.7035713195800781, + "num_tokens": 37689055.0, + "step": 1472 + }, + { + "epoch": 0.16176147594992313, + "grad_norm": 2.1941113471984863, + "learning_rate": 8.079034028540066e-07, + "loss": 1.122, + "mean_token_accuracy": 0.6671447157859802, + "num_tokens": 37720293.0, + "step": 1473 + }, + { + "epoch": 0.1618712936525368, + "grad_norm": 2.2796032428741455, + "learning_rate": 8.084522502744236e-07, + "loss": 0.9942, + "mean_token_accuracy": 0.7011559009552002, + "num_tokens": 37748425.0, + "step": 1474 + }, + { + "epoch": 0.16198111135515045, + "grad_norm": 2.0664236545562744, + "learning_rate": 8.090010976948408e-07, + "loss": 1.1183, + "mean_token_accuracy": 0.6791988611221313, + "num_tokens": 37779771.0, + "step": 1475 + }, + { + "epoch": 0.16209092905776412, + "grad_norm": 2.4047482013702393, + "learning_rate": 8.095499451152578e-07, + "loss": 1.1089, + "mean_token_accuracy": 0.674185037612915, + "num_tokens": 37803803.0, + "step": 1476 + }, + { + "epoch": 0.16220074676037777, + "grad_norm": 2.6977438926696777, + "learning_rate": 8.10098792535675e-07, + "loss": 0.9646, + "mean_token_accuracy": 0.7088260650634766, + "num_tokens": 37823112.0, + "step": 1477 + }, + { + "epoch": 0.16231056446299144, + "grad_norm": 2.4907894134521484, + "learning_rate": 8.106476399560922e-07, + "loss": 1.0274, + "mean_token_accuracy": 0.6939009428024292, + "num_tokens": 37845686.0, + "step": 1478 + }, + { + "epoch": 0.1624203821656051, + "grad_norm": 2.4781363010406494, + "learning_rate": 8.111964873765092e-07, + "loss": 1.0264, + "mean_token_accuracy": 0.6928783059120178, + "num_tokens": 37870510.0, + "step": 1479 + }, + { + "epoch": 0.16253019986821876, + "grad_norm": 2.585109233856201, + "learning_rate": 8.117453347969264e-07, + "loss": 0.9419, + "mean_token_accuracy": 0.7128769159317017, + "num_tokens": 37890854.0, + "step": 1480 + }, + { + "epoch": 0.1626400175708324, + "grad_norm": 2.376538038253784, + "learning_rate": 8.122941822173436e-07, + "loss": 1.0087, + "mean_token_accuracy": 0.6948882341384888, + "num_tokens": 37914882.0, + "step": 1481 + }, + { + "epoch": 0.16274983527344608, + "grad_norm": 2.0733485221862793, + "learning_rate": 8.128430296377606e-07, + "loss": 1.1098, + "mean_token_accuracy": 0.6682459115982056, + "num_tokens": 37947229.0, + "step": 1482 + }, + { + "epoch": 0.16285965297605975, + "grad_norm": 2.2588560581207275, + "learning_rate": 8.133918770581778e-07, + "loss": 0.9606, + "mean_token_accuracy": 0.7094703912734985, + "num_tokens": 37973570.0, + "step": 1483 + }, + { + "epoch": 0.1629694706786734, + "grad_norm": 2.3132102489471436, + "learning_rate": 8.13940724478595e-07, + "loss": 0.9859, + "mean_token_accuracy": 0.6999467611312866, + "num_tokens": 37998507.0, + "step": 1484 + }, + { + "epoch": 0.16307928838128707, + "grad_norm": 2.4198827743530273, + "learning_rate": 8.14489571899012e-07, + "loss": 1.0458, + "mean_token_accuracy": 0.6927634477615356, + "num_tokens": 38021960.0, + "step": 1485 + }, + { + "epoch": 0.16318910608390072, + "grad_norm": 2.4432175159454346, + "learning_rate": 8.150384193194292e-07, + "loss": 1.0954, + "mean_token_accuracy": 0.6824548244476318, + "num_tokens": 38046666.0, + "step": 1486 + }, + { + "epoch": 0.1632989237865144, + "grad_norm": 2.214531421661377, + "learning_rate": 8.155872667398463e-07, + "loss": 1.1358, + "mean_token_accuracy": 0.6820165514945984, + "num_tokens": 38074113.0, + "step": 1487 + }, + { + "epoch": 0.16340874148912804, + "grad_norm": 2.968079090118408, + "learning_rate": 8.161361141602634e-07, + "loss": 0.8703, + "mean_token_accuracy": 0.7319099307060242, + "num_tokens": 38090975.0, + "step": 1488 + }, + { + "epoch": 0.1635185591917417, + "grad_norm": 2.4088871479034424, + "learning_rate": 8.166849615806806e-07, + "loss": 1.069, + "mean_token_accuracy": 0.680095374584198, + "num_tokens": 38116333.0, + "step": 1489 + }, + { + "epoch": 0.16362837689435536, + "grad_norm": 2.233671188354492, + "learning_rate": 8.172338090010977e-07, + "loss": 1.0962, + "mean_token_accuracy": 0.6787694096565247, + "num_tokens": 38147265.0, + "step": 1490 + }, + { + "epoch": 0.16373819459696903, + "grad_norm": 2.5342888832092285, + "learning_rate": 8.177826564215147e-07, + "loss": 1.0982, + "mean_token_accuracy": 0.6863139867782593, + "num_tokens": 38169145.0, + "step": 1491 + }, + { + "epoch": 0.16384801229958268, + "grad_norm": 2.25319504737854, + "learning_rate": 8.183315038419319e-07, + "loss": 1.0102, + "mean_token_accuracy": 0.7003800868988037, + "num_tokens": 38195461.0, + "step": 1492 + }, + { + "epoch": 0.16395783000219635, + "grad_norm": 2.5547049045562744, + "learning_rate": 8.18880351262349e-07, + "loss": 0.93, + "mean_token_accuracy": 0.7116396427154541, + "num_tokens": 38215135.0, + "step": 1493 + }, + { + "epoch": 0.16406764770481003, + "grad_norm": 2.4761104583740234, + "learning_rate": 8.194291986827661e-07, + "loss": 1.0548, + "mean_token_accuracy": 0.6885881423950195, + "num_tokens": 38239934.0, + "step": 1494 + }, + { + "epoch": 0.16417746540742367, + "grad_norm": 2.6409294605255127, + "learning_rate": 8.199780461031833e-07, + "loss": 0.9527, + "mean_token_accuracy": 0.7120916247367859, + "num_tokens": 38262245.0, + "step": 1495 + }, + { + "epoch": 0.16428728311003735, + "grad_norm": 2.120588779449463, + "learning_rate": 8.205268935236004e-07, + "loss": 1.0076, + "mean_token_accuracy": 0.6949032545089722, + "num_tokens": 38290769.0, + "step": 1496 + }, + { + "epoch": 0.164397100812651, + "grad_norm": 2.615692615509033, + "learning_rate": 8.210757409440175e-07, + "loss": 1.0692, + "mean_token_accuracy": 0.679734468460083, + "num_tokens": 38313747.0, + "step": 1497 + }, + { + "epoch": 0.16450691851526467, + "grad_norm": 2.490731954574585, + "learning_rate": 8.216245883644346e-07, + "loss": 1.0577, + "mean_token_accuracy": 0.6881515979766846, + "num_tokens": 38337012.0, + "step": 1498 + }, + { + "epoch": 0.1646167362178783, + "grad_norm": 2.3545734882354736, + "learning_rate": 8.221734357848518e-07, + "loss": 1.0803, + "mean_token_accuracy": 0.6860709190368652, + "num_tokens": 38361893.0, + "step": 1499 + }, + { + "epoch": 0.16472655392049199, + "grad_norm": 2.310739278793335, + "learning_rate": 8.227222832052689e-07, + "loss": 0.9577, + "mean_token_accuracy": 0.7174622416496277, + "num_tokens": 38385043.0, + "step": 1500 + }, + { + "epoch": 0.16483637162310563, + "grad_norm": 2.1103992462158203, + "learning_rate": 8.23271130625686e-07, + "loss": 1.0439, + "mean_token_accuracy": 0.690568745136261, + "num_tokens": 38413918.0, + "step": 1501 + }, + { + "epoch": 0.1649461893257193, + "grad_norm": 2.2568154335021973, + "learning_rate": 8.238199780461032e-07, + "loss": 0.967, + "mean_token_accuracy": 0.7071869969367981, + "num_tokens": 38438883.0, + "step": 1502 + }, + { + "epoch": 0.16505600702833298, + "grad_norm": 2.3654682636260986, + "learning_rate": 8.243688254665203e-07, + "loss": 0.8402, + "mean_token_accuracy": 0.7356154322624207, + "num_tokens": 38459249.0, + "step": 1503 + }, + { + "epoch": 0.16516582473094663, + "grad_norm": 2.474609375, + "learning_rate": 8.249176728869374e-07, + "loss": 0.9641, + "mean_token_accuracy": 0.7202419638633728, + "num_tokens": 38480861.0, + "step": 1504 + }, + { + "epoch": 0.1652756424335603, + "grad_norm": 2.440260171890259, + "learning_rate": 8.254665203073546e-07, + "loss": 0.9382, + "mean_token_accuracy": 0.7172552347183228, + "num_tokens": 38503195.0, + "step": 1505 + }, + { + "epoch": 0.16538546013617395, + "grad_norm": 2.4179108142852783, + "learning_rate": 8.260153677277717e-07, + "loss": 0.9624, + "mean_token_accuracy": 0.7079695463180542, + "num_tokens": 38524890.0, + "step": 1506 + }, + { + "epoch": 0.16549527783878762, + "grad_norm": 2.596456527709961, + "learning_rate": 8.265642151481888e-07, + "loss": 0.9972, + "mean_token_accuracy": 0.6977109909057617, + "num_tokens": 38545824.0, + "step": 1507 + }, + { + "epoch": 0.16560509554140126, + "grad_norm": 2.419572114944458, + "learning_rate": 8.27113062568606e-07, + "loss": 0.9519, + "mean_token_accuracy": 0.7116296291351318, + "num_tokens": 38568310.0, + "step": 1508 + }, + { + "epoch": 0.16571491324401494, + "grad_norm": 2.501816987991333, + "learning_rate": 8.27661909989023e-07, + "loss": 1.0437, + "mean_token_accuracy": 0.6993807554244995, + "num_tokens": 38590682.0, + "step": 1509 + }, + { + "epoch": 0.16582473094662858, + "grad_norm": 2.4262967109680176, + "learning_rate": 8.282107574094401e-07, + "loss": 1.0318, + "mean_token_accuracy": 0.6881381273269653, + "num_tokens": 38613642.0, + "step": 1510 + }, + { + "epoch": 0.16593454864924226, + "grad_norm": 2.422300100326538, + "learning_rate": 8.287596048298572e-07, + "loss": 1.0081, + "mean_token_accuracy": 0.6943548917770386, + "num_tokens": 38637138.0, + "step": 1511 + }, + { + "epoch": 0.16604436635185593, + "grad_norm": 2.2370502948760986, + "learning_rate": 8.293084522502743e-07, + "loss": 1.0484, + "mean_token_accuracy": 0.6900200247764587, + "num_tokens": 38661432.0, + "step": 1512 + }, + { + "epoch": 0.16615418405446958, + "grad_norm": 2.115355968475342, + "learning_rate": 8.298572996706915e-07, + "loss": 1.0239, + "mean_token_accuracy": 0.6954762935638428, + "num_tokens": 38691378.0, + "step": 1513 + }, + { + "epoch": 0.16626400175708325, + "grad_norm": 2.240835428237915, + "learning_rate": 8.304061470911086e-07, + "loss": 1.1048, + "mean_token_accuracy": 0.6710975170135498, + "num_tokens": 38717530.0, + "step": 1514 + }, + { + "epoch": 0.1663738194596969, + "grad_norm": 2.4576923847198486, + "learning_rate": 8.309549945115257e-07, + "loss": 1.0089, + "mean_token_accuracy": 0.7007567882537842, + "num_tokens": 38740470.0, + "step": 1515 + }, + { + "epoch": 0.16648363716231057, + "grad_norm": 2.0762031078338623, + "learning_rate": 8.315038419319429e-07, + "loss": 0.9285, + "mean_token_accuracy": 0.7193874716758728, + "num_tokens": 38767976.0, + "step": 1516 + }, + { + "epoch": 0.16659345486492422, + "grad_norm": 2.026686906814575, + "learning_rate": 8.3205268935236e-07, + "loss": 1.036, + "mean_token_accuracy": 0.6960018873214722, + "num_tokens": 38798581.0, + "step": 1517 + }, + { + "epoch": 0.1667032725675379, + "grad_norm": 2.5423221588134766, + "learning_rate": 8.326015367727771e-07, + "loss": 1.1206, + "mean_token_accuracy": 0.6668636798858643, + "num_tokens": 38820397.0, + "step": 1518 + }, + { + "epoch": 0.16681309027015154, + "grad_norm": 2.276806592941284, + "learning_rate": 8.331503841931943e-07, + "loss": 1.0749, + "mean_token_accuracy": 0.6833056211471558, + "num_tokens": 38845222.0, + "step": 1519 + }, + { + "epoch": 0.1669229079727652, + "grad_norm": 2.242701768875122, + "learning_rate": 8.336992316136113e-07, + "loss": 1.0496, + "mean_token_accuracy": 0.6885234713554382, + "num_tokens": 38871982.0, + "step": 1520 + }, + { + "epoch": 0.16703272567537888, + "grad_norm": 2.7645843029022217, + "learning_rate": 8.342480790340285e-07, + "loss": 0.9812, + "mean_token_accuracy": 0.7080410718917847, + "num_tokens": 38890380.0, + "step": 1521 + }, + { + "epoch": 0.16714254337799253, + "grad_norm": 2.2615761756896973, + "learning_rate": 8.347969264544457e-07, + "loss": 1.0583, + "mean_token_accuracy": 0.6839885115623474, + "num_tokens": 38916800.0, + "step": 1522 + }, + { + "epoch": 0.1672523610806062, + "grad_norm": 2.3275630474090576, + "learning_rate": 8.353457738748627e-07, + "loss": 1.0379, + "mean_token_accuracy": 0.6926053762435913, + "num_tokens": 38940564.0, + "step": 1523 + }, + { + "epoch": 0.16736217878321985, + "grad_norm": 2.2238805294036865, + "learning_rate": 8.358946212952799e-07, + "loss": 0.9968, + "mean_token_accuracy": 0.7100387811660767, + "num_tokens": 38967034.0, + "step": 1524 + }, + { + "epoch": 0.16747199648583352, + "grad_norm": 2.0534722805023193, + "learning_rate": 8.364434687156971e-07, + "loss": 1.0521, + "mean_token_accuracy": 0.6843950748443604, + "num_tokens": 38998519.0, + "step": 1525 + }, + { + "epoch": 0.16758181418844717, + "grad_norm": 2.2360174655914307, + "learning_rate": 8.369923161361141e-07, + "loss": 1.0125, + "mean_token_accuracy": 0.6983431577682495, + "num_tokens": 39023397.0, + "step": 1526 + }, + { + "epoch": 0.16769163189106084, + "grad_norm": 2.224660634994507, + "learning_rate": 8.375411635565313e-07, + "loss": 1.0108, + "mean_token_accuracy": 0.6960116028785706, + "num_tokens": 39050040.0, + "step": 1527 + }, + { + "epoch": 0.1678014495936745, + "grad_norm": 2.3179919719696045, + "learning_rate": 8.380900109769485e-07, + "loss": 0.9884, + "mean_token_accuracy": 0.7063190340995789, + "num_tokens": 39074766.0, + "step": 1528 + }, + { + "epoch": 0.16791126729628816, + "grad_norm": 2.188610553741455, + "learning_rate": 8.386388583973654e-07, + "loss": 1.0667, + "mean_token_accuracy": 0.6973378658294678, + "num_tokens": 39102707.0, + "step": 1529 + }, + { + "epoch": 0.1680210849989018, + "grad_norm": 2.188512086868286, + "learning_rate": 8.391877058177826e-07, + "loss": 1.0884, + "mean_token_accuracy": 0.6796536445617676, + "num_tokens": 39132599.0, + "step": 1530 + }, + { + "epoch": 0.16813090270151548, + "grad_norm": 2.103876829147339, + "learning_rate": 8.397365532381997e-07, + "loss": 0.9157, + "mean_token_accuracy": 0.7306554317474365, + "num_tokens": 39159482.0, + "step": 1531 + }, + { + "epoch": 0.16824072040412916, + "grad_norm": 2.4249985218048096, + "learning_rate": 8.402854006586168e-07, + "loss": 1.0254, + "mean_token_accuracy": 0.6909936666488647, + "num_tokens": 39181709.0, + "step": 1532 + }, + { + "epoch": 0.1683505381067428, + "grad_norm": 2.172804355621338, + "learning_rate": 8.40834248079034e-07, + "loss": 1.0739, + "mean_token_accuracy": 0.6782289743423462, + "num_tokens": 39211464.0, + "step": 1533 + }, + { + "epoch": 0.16846035580935648, + "grad_norm": 2.431926727294922, + "learning_rate": 8.413830954994511e-07, + "loss": 1.0731, + "mean_token_accuracy": 0.6867292523384094, + "num_tokens": 39236726.0, + "step": 1534 + }, + { + "epoch": 0.16857017351197012, + "grad_norm": 2.6266109943389893, + "learning_rate": 8.419319429198682e-07, + "loss": 1.0443, + "mean_token_accuracy": 0.6883031129837036, + "num_tokens": 39259077.0, + "step": 1535 + }, + { + "epoch": 0.1686799912145838, + "grad_norm": 2.232654094696045, + "learning_rate": 8.424807903402854e-07, + "loss": 0.9733, + "mean_token_accuracy": 0.7076330780982971, + "num_tokens": 39285130.0, + "step": 1536 + }, + { + "epoch": 0.16878980891719744, + "grad_norm": 2.361870050430298, + "learning_rate": 8.430296377607025e-07, + "loss": 1.0229, + "mean_token_accuracy": 0.6934114694595337, + "num_tokens": 39310691.0, + "step": 1537 + }, + { + "epoch": 0.16889962661981112, + "grad_norm": 2.5161855220794678, + "learning_rate": 8.435784851811196e-07, + "loss": 1.0651, + "mean_token_accuracy": 0.6793939471244812, + "num_tokens": 39333661.0, + "step": 1538 + }, + { + "epoch": 0.16900944432242476, + "grad_norm": 2.3695123195648193, + "learning_rate": 8.441273326015367e-07, + "loss": 1.0535, + "mean_token_accuracy": 0.6945710182189941, + "num_tokens": 39359278.0, + "step": 1539 + }, + { + "epoch": 0.16911926202503844, + "grad_norm": 2.0882081985473633, + "learning_rate": 8.446761800219539e-07, + "loss": 1.122, + "mean_token_accuracy": 0.6634929180145264, + "num_tokens": 39391457.0, + "step": 1540 + }, + { + "epoch": 0.1692290797276521, + "grad_norm": 2.278157949447632, + "learning_rate": 8.45225027442371e-07, + "loss": 1.093, + "mean_token_accuracy": 0.6762073636054993, + "num_tokens": 39417228.0, + "step": 1541 + }, + { + "epoch": 0.16933889743026576, + "grad_norm": 2.3401851654052734, + "learning_rate": 8.457738748627881e-07, + "loss": 1.037, + "mean_token_accuracy": 0.6917052268981934, + "num_tokens": 39441067.0, + "step": 1542 + }, + { + "epoch": 0.16944871513287943, + "grad_norm": 2.472745418548584, + "learning_rate": 8.463227222832053e-07, + "loss": 0.9889, + "mean_token_accuracy": 0.7010893821716309, + "num_tokens": 39463326.0, + "step": 1543 + }, + { + "epoch": 0.16955853283549308, + "grad_norm": 2.2581043243408203, + "learning_rate": 8.468715697036224e-07, + "loss": 1.0492, + "mean_token_accuracy": 0.6887592673301697, + "num_tokens": 39487867.0, + "step": 1544 + }, + { + "epoch": 0.16966835053810675, + "grad_norm": 2.3996970653533936, + "learning_rate": 8.474204171240395e-07, + "loss": 1.0193, + "mean_token_accuracy": 0.6975046396255493, + "num_tokens": 39510865.0, + "step": 1545 + }, + { + "epoch": 0.1697781682407204, + "grad_norm": 2.2224740982055664, + "learning_rate": 8.479692645444567e-07, + "loss": 0.9639, + "mean_token_accuracy": 0.7064975500106812, + "num_tokens": 39534861.0, + "step": 1546 + }, + { + "epoch": 0.16988798594333407, + "grad_norm": 2.6295745372772217, + "learning_rate": 8.485181119648738e-07, + "loss": 1.0677, + "mean_token_accuracy": 0.686377763748169, + "num_tokens": 39557202.0, + "step": 1547 + }, + { + "epoch": 0.16999780364594771, + "grad_norm": 2.213609218597412, + "learning_rate": 8.490669593852908e-07, + "loss": 0.9932, + "mean_token_accuracy": 0.7015056610107422, + "num_tokens": 39583533.0, + "step": 1548 + }, + { + "epoch": 0.1701076213485614, + "grad_norm": 2.238704204559326, + "learning_rate": 8.496158068057079e-07, + "loss": 1.0196, + "mean_token_accuracy": 0.7006559371948242, + "num_tokens": 39610047.0, + "step": 1549 + }, + { + "epoch": 0.17021743905117506, + "grad_norm": 2.3416690826416016, + "learning_rate": 8.50164654226125e-07, + "loss": 0.984, + "mean_token_accuracy": 0.7024701833724976, + "num_tokens": 39634730.0, + "step": 1550 + }, + { + "epoch": 0.1703272567537887, + "grad_norm": 2.6367366313934326, + "learning_rate": 8.507135016465422e-07, + "loss": 1.0213, + "mean_token_accuracy": 0.6964163780212402, + "num_tokens": 39654935.0, + "step": 1551 + }, + { + "epoch": 0.17043707445640238, + "grad_norm": 2.3914847373962402, + "learning_rate": 8.512623490669593e-07, + "loss": 1.0247, + "mean_token_accuracy": 0.699015736579895, + "num_tokens": 39678194.0, + "step": 1552 + }, + { + "epoch": 0.17054689215901603, + "grad_norm": 2.243403911590576, + "learning_rate": 8.518111964873764e-07, + "loss": 1.0562, + "mean_token_accuracy": 0.683012843132019, + "num_tokens": 39704742.0, + "step": 1553 + }, + { + "epoch": 0.1706567098616297, + "grad_norm": 2.067636489868164, + "learning_rate": 8.523600439077936e-07, + "loss": 1.0023, + "mean_token_accuracy": 0.6938233375549316, + "num_tokens": 39732590.0, + "step": 1554 + }, + { + "epoch": 0.17076652756424335, + "grad_norm": 1.9961947202682495, + "learning_rate": 8.529088913282107e-07, + "loss": 1.0898, + "mean_token_accuracy": 0.6727994084358215, + "num_tokens": 39765715.0, + "step": 1555 + }, + { + "epoch": 0.17087634526685702, + "grad_norm": 2.128211736679077, + "learning_rate": 8.534577387486278e-07, + "loss": 1.0845, + "mean_token_accuracy": 0.670399010181427, + "num_tokens": 39795607.0, + "step": 1556 + }, + { + "epoch": 0.17098616296947067, + "grad_norm": 2.0723989009857178, + "learning_rate": 8.54006586169045e-07, + "loss": 0.9875, + "mean_token_accuracy": 0.6991793513298035, + "num_tokens": 39823650.0, + "step": 1557 + }, + { + "epoch": 0.17109598067208434, + "grad_norm": 2.1892502307891846, + "learning_rate": 8.545554335894621e-07, + "loss": 1.1231, + "mean_token_accuracy": 0.6695929765701294, + "num_tokens": 39852966.0, + "step": 1558 + }, + { + "epoch": 0.17120579837469801, + "grad_norm": 2.181138753890991, + "learning_rate": 8.551042810098792e-07, + "loss": 1.0697, + "mean_token_accuracy": 0.6823446154594421, + "num_tokens": 39881955.0, + "step": 1559 + }, + { + "epoch": 0.17131561607731166, + "grad_norm": 2.3221442699432373, + "learning_rate": 8.556531284302964e-07, + "loss": 1.1411, + "mean_token_accuracy": 0.6670087575912476, + "num_tokens": 39909359.0, + "step": 1560 + }, + { + "epoch": 0.17142543377992533, + "grad_norm": 2.365037441253662, + "learning_rate": 8.562019758507134e-07, + "loss": 1.0208, + "mean_token_accuracy": 0.7004137635231018, + "num_tokens": 39932850.0, + "step": 1561 + }, + { + "epoch": 0.17153525148253898, + "grad_norm": 2.2001662254333496, + "learning_rate": 8.567508232711306e-07, + "loss": 1.0637, + "mean_token_accuracy": 0.6834970712661743, + "num_tokens": 39962315.0, + "step": 1562 + }, + { + "epoch": 0.17164506918515265, + "grad_norm": 2.3364648818969727, + "learning_rate": 8.572996706915478e-07, + "loss": 0.9623, + "mean_token_accuracy": 0.7104230523109436, + "num_tokens": 39985916.0, + "step": 1563 + }, + { + "epoch": 0.1717548868877663, + "grad_norm": 2.622404098510742, + "learning_rate": 8.578485181119648e-07, + "loss": 0.9335, + "mean_token_accuracy": 0.7132030725479126, + "num_tokens": 40004498.0, + "step": 1564 + }, + { + "epoch": 0.17186470459037997, + "grad_norm": 2.143205404281616, + "learning_rate": 8.58397365532382e-07, + "loss": 1.0028, + "mean_token_accuracy": 0.6981576085090637, + "num_tokens": 40034562.0, + "step": 1565 + }, + { + "epoch": 0.17197452229299362, + "grad_norm": 2.250608205795288, + "learning_rate": 8.589462129527992e-07, + "loss": 1.0768, + "mean_token_accuracy": 0.6745154857635498, + "num_tokens": 40062808.0, + "step": 1566 + }, + { + "epoch": 0.1720843399956073, + "grad_norm": 2.072770357131958, + "learning_rate": 8.594950603732161e-07, + "loss": 1.035, + "mean_token_accuracy": 0.6922900080680847, + "num_tokens": 40094567.0, + "step": 1567 + }, + { + "epoch": 0.17219415769822094, + "grad_norm": 2.5115244388580322, + "learning_rate": 8.600439077936333e-07, + "loss": 0.9565, + "mean_token_accuracy": 0.7082608342170715, + "num_tokens": 40116854.0, + "step": 1568 + }, + { + "epoch": 0.1723039754008346, + "grad_norm": 2.288245677947998, + "learning_rate": 8.605927552140505e-07, + "loss": 0.9998, + "mean_token_accuracy": 0.7007331252098083, + "num_tokens": 40142451.0, + "step": 1569 + }, + { + "epoch": 0.1724137931034483, + "grad_norm": 2.5254642963409424, + "learning_rate": 8.611416026344675e-07, + "loss": 0.9925, + "mean_token_accuracy": 0.7015336751937866, + "num_tokens": 40163037.0, + "step": 1570 + }, + { + "epoch": 0.17252361080606193, + "grad_norm": 2.219202756881714, + "learning_rate": 8.616904500548847e-07, + "loss": 1.0564, + "mean_token_accuracy": 0.6859536170959473, + "num_tokens": 40190047.0, + "step": 1571 + }, + { + "epoch": 0.1726334285086756, + "grad_norm": 2.2257063388824463, + "learning_rate": 8.622392974753018e-07, + "loss": 0.9824, + "mean_token_accuracy": 0.7041842937469482, + "num_tokens": 40216176.0, + "step": 1572 + }, + { + "epoch": 0.17274324621128925, + "grad_norm": 2.204339027404785, + "learning_rate": 8.627881448957189e-07, + "loss": 1.103, + "mean_token_accuracy": 0.6730757355690002, + "num_tokens": 40247632.0, + "step": 1573 + }, + { + "epoch": 0.17285306391390293, + "grad_norm": 2.4181525707244873, + "learning_rate": 8.633369923161361e-07, + "loss": 1.0029, + "mean_token_accuracy": 0.6950483918190002, + "num_tokens": 40269428.0, + "step": 1574 + }, + { + "epoch": 0.17296288161651657, + "grad_norm": 2.0575456619262695, + "learning_rate": 8.638858397365532e-07, + "loss": 0.9892, + "mean_token_accuracy": 0.7028926014900208, + "num_tokens": 40299022.0, + "step": 1575 + }, + { + "epoch": 0.17307269931913025, + "grad_norm": 2.516101837158203, + "learning_rate": 8.644346871569703e-07, + "loss": 1.0444, + "mean_token_accuracy": 0.6856255531311035, + "num_tokens": 40321446.0, + "step": 1576 + }, + { + "epoch": 0.1731825170217439, + "grad_norm": 2.296029806137085, + "learning_rate": 8.649835345773875e-07, + "loss": 1.0204, + "mean_token_accuracy": 0.6943627595901489, + "num_tokens": 40345509.0, + "step": 1577 + }, + { + "epoch": 0.17329233472435757, + "grad_norm": 2.3279993534088135, + "learning_rate": 8.655323819978046e-07, + "loss": 1.0139, + "mean_token_accuracy": 0.7043192982673645, + "num_tokens": 40373072.0, + "step": 1578 + }, + { + "epoch": 0.17340215242697124, + "grad_norm": 2.21854567527771, + "learning_rate": 8.660812294182217e-07, + "loss": 1.0218, + "mean_token_accuracy": 0.6847776770591736, + "num_tokens": 40399681.0, + "step": 1579 + }, + { + "epoch": 0.17351197012958489, + "grad_norm": 2.3715758323669434, + "learning_rate": 8.666300768386389e-07, + "loss": 0.9363, + "mean_token_accuracy": 0.7159496545791626, + "num_tokens": 40422576.0, + "step": 1580 + }, + { + "epoch": 0.17362178783219856, + "grad_norm": 2.3201675415039062, + "learning_rate": 8.67178924259056e-07, + "loss": 1.059, + "mean_token_accuracy": 0.6838423609733582, + "num_tokens": 40446850.0, + "step": 1581 + }, + { + "epoch": 0.1737316055348122, + "grad_norm": 2.383200168609619, + "learning_rate": 8.677277716794731e-07, + "loss": 1.0247, + "mean_token_accuracy": 0.6912331581115723, + "num_tokens": 40471163.0, + "step": 1582 + }, + { + "epoch": 0.17384142323742588, + "grad_norm": 2.1672396659851074, + "learning_rate": 8.682766190998902e-07, + "loss": 1.1183, + "mean_token_accuracy": 0.6737644672393799, + "num_tokens": 40503425.0, + "step": 1583 + }, + { + "epoch": 0.17395124094003953, + "grad_norm": 2.0183660984039307, + "learning_rate": 8.688254665203073e-07, + "loss": 1.0539, + "mean_token_accuracy": 0.6867313981056213, + "num_tokens": 40536826.0, + "step": 1584 + }, + { + "epoch": 0.1740610586426532, + "grad_norm": 2.2910056114196777, + "learning_rate": 8.693743139407245e-07, + "loss": 1.1366, + "mean_token_accuracy": 0.6651262044906616, + "num_tokens": 40564628.0, + "step": 1585 + }, + { + "epoch": 0.17417087634526685, + "grad_norm": 2.4127540588378906, + "learning_rate": 8.699231613611415e-07, + "loss": 1.0135, + "mean_token_accuracy": 0.6972616910934448, + "num_tokens": 40588130.0, + "step": 1586 + }, + { + "epoch": 0.17428069404788052, + "grad_norm": 2.2368767261505127, + "learning_rate": 8.704720087815586e-07, + "loss": 1.0737, + "mean_token_accuracy": 0.6897912621498108, + "num_tokens": 40614561.0, + "step": 1587 + }, + { + "epoch": 0.1743905117504942, + "grad_norm": 2.418671131134033, + "learning_rate": 8.710208562019758e-07, + "loss": 1.0555, + "mean_token_accuracy": 0.7004085779190063, + "num_tokens": 40638759.0, + "step": 1588 + }, + { + "epoch": 0.17450032945310784, + "grad_norm": 2.167311429977417, + "learning_rate": 8.715697036223929e-07, + "loss": 1.0281, + "mean_token_accuracy": 0.6933973431587219, + "num_tokens": 40665717.0, + "step": 1589 + }, + { + "epoch": 0.1746101471557215, + "grad_norm": 2.284715175628662, + "learning_rate": 8.7211855104281e-07, + "loss": 1.0185, + "mean_token_accuracy": 0.7012127637863159, + "num_tokens": 40690490.0, + "step": 1590 + }, + { + "epoch": 0.17471996485833516, + "grad_norm": 2.3456027507781982, + "learning_rate": 8.726673984632272e-07, + "loss": 0.9921, + "mean_token_accuracy": 0.7029803991317749, + "num_tokens": 40715686.0, + "step": 1591 + }, + { + "epoch": 0.17482978256094883, + "grad_norm": 2.168998956680298, + "learning_rate": 8.732162458836443e-07, + "loss": 1.1033, + "mean_token_accuracy": 0.6776716113090515, + "num_tokens": 40743590.0, + "step": 1592 + }, + { + "epoch": 0.17493960026356248, + "grad_norm": 2.21360445022583, + "learning_rate": 8.737650933040614e-07, + "loss": 1.0289, + "mean_token_accuracy": 0.6951782703399658, + "num_tokens": 40772397.0, + "step": 1593 + }, + { + "epoch": 0.17504941796617615, + "grad_norm": 2.4040682315826416, + "learning_rate": 8.743139407244785e-07, + "loss": 0.9041, + "mean_token_accuracy": 0.7194291353225708, + "num_tokens": 40793191.0, + "step": 1594 + }, + { + "epoch": 0.1751592356687898, + "grad_norm": 2.275949239730835, + "learning_rate": 8.748627881448957e-07, + "loss": 1.0409, + "mean_token_accuracy": 0.6899837851524353, + "num_tokens": 40818548.0, + "step": 1595 + }, + { + "epoch": 0.17526905337140347, + "grad_norm": 2.6282002925872803, + "learning_rate": 8.754116355653128e-07, + "loss": 1.0422, + "mean_token_accuracy": 0.683754026889801, + "num_tokens": 40840220.0, + "step": 1596 + }, + { + "epoch": 0.17537887107401715, + "grad_norm": 2.403886318206787, + "learning_rate": 8.759604829857299e-07, + "loss": 0.9546, + "mean_token_accuracy": 0.7088642120361328, + "num_tokens": 40863813.0, + "step": 1597 + }, + { + "epoch": 0.1754886887766308, + "grad_norm": 2.299227714538574, + "learning_rate": 8.765093304061471e-07, + "loss": 1.0027, + "mean_token_accuracy": 0.6990155577659607, + "num_tokens": 40889201.0, + "step": 1598 + }, + { + "epoch": 0.17559850647924446, + "grad_norm": 2.086519718170166, + "learning_rate": 8.770581778265642e-07, + "loss": 1.0494, + "mean_token_accuracy": 0.6871469020843506, + "num_tokens": 40918233.0, + "step": 1599 + }, + { + "epoch": 0.1757083241818581, + "grad_norm": 1.9530404806137085, + "learning_rate": 8.776070252469813e-07, + "loss": 0.9877, + "mean_token_accuracy": 0.7008439302444458, + "num_tokens": 40947556.0, + "step": 1600 + }, + { + "epoch": 0.17581814188447178, + "grad_norm": 2.2869961261749268, + "learning_rate": 8.781558726673985e-07, + "loss": 1.0235, + "mean_token_accuracy": 0.6958431005477905, + "num_tokens": 40971882.0, + "step": 1601 + }, + { + "epoch": 0.17592795958708543, + "grad_norm": 2.770472526550293, + "learning_rate": 8.787047200878156e-07, + "loss": 0.9657, + "mean_token_accuracy": 0.707931399345398, + "num_tokens": 40990806.0, + "step": 1602 + }, + { + "epoch": 0.1760377772896991, + "grad_norm": 2.1451807022094727, + "learning_rate": 8.792535675082327e-07, + "loss": 1.1426, + "mean_token_accuracy": 0.665822446346283, + "num_tokens": 41018609.0, + "step": 1603 + }, + { + "epoch": 0.17614759499231275, + "grad_norm": 2.278790235519409, + "learning_rate": 8.798024149286499e-07, + "loss": 1.0314, + "mean_token_accuracy": 0.6947289705276489, + "num_tokens": 41044411.0, + "step": 1604 + }, + { + "epoch": 0.17625741269492642, + "grad_norm": 2.3867712020874023, + "learning_rate": 8.803512623490668e-07, + "loss": 0.9317, + "mean_token_accuracy": 0.71787428855896, + "num_tokens": 41067349.0, + "step": 1605 + }, + { + "epoch": 0.17636723039754007, + "grad_norm": 2.0930066108703613, + "learning_rate": 8.80900109769484e-07, + "loss": 1.047, + "mean_token_accuracy": 0.6912461519241333, + "num_tokens": 41097812.0, + "step": 1606 + }, + { + "epoch": 0.17647704810015374, + "grad_norm": 2.5711984634399414, + "learning_rate": 8.814489571899012e-07, + "loss": 1.0206, + "mean_token_accuracy": 0.6998223066329956, + "num_tokens": 41119564.0, + "step": 1607 + }, + { + "epoch": 0.17658686580276742, + "grad_norm": 2.616997003555298, + "learning_rate": 8.819978046103182e-07, + "loss": 1.0069, + "mean_token_accuracy": 0.6978645324707031, + "num_tokens": 41141144.0, + "step": 1608 + }, + { + "epoch": 0.17669668350538106, + "grad_norm": 2.312784433364868, + "learning_rate": 8.825466520307354e-07, + "loss": 1.0956, + "mean_token_accuracy": 0.6823378801345825, + "num_tokens": 41166481.0, + "step": 1609 + }, + { + "epoch": 0.17680650120799474, + "grad_norm": 2.7021820545196533, + "learning_rate": 8.830954994511526e-07, + "loss": 1.0151, + "mean_token_accuracy": 0.6922574043273926, + "num_tokens": 41185219.0, + "step": 1610 + }, + { + "epoch": 0.17691631891060838, + "grad_norm": 2.1574158668518066, + "learning_rate": 8.836443468715696e-07, + "loss": 1.0544, + "mean_token_accuracy": 0.6828898191452026, + "num_tokens": 41213727.0, + "step": 1611 + }, + { + "epoch": 0.17702613661322206, + "grad_norm": 2.0682661533355713, + "learning_rate": 8.841931942919868e-07, + "loss": 0.9741, + "mean_token_accuracy": 0.7033864855766296, + "num_tokens": 41242061.0, + "step": 1612 + }, + { + "epoch": 0.1771359543158357, + "grad_norm": 2.4379470348358154, + "learning_rate": 8.84742041712404e-07, + "loss": 0.9701, + "mean_token_accuracy": 0.7049189805984497, + "num_tokens": 41264465.0, + "step": 1613 + }, + { + "epoch": 0.17724577201844938, + "grad_norm": 2.3707234859466553, + "learning_rate": 8.85290889132821e-07, + "loss": 1.0565, + "mean_token_accuracy": 0.6847890615463257, + "num_tokens": 41290582.0, + "step": 1614 + }, + { + "epoch": 0.17735558972106302, + "grad_norm": 2.4270577430725098, + "learning_rate": 8.858397365532382e-07, + "loss": 1.0131, + "mean_token_accuracy": 0.6973550319671631, + "num_tokens": 41312874.0, + "step": 1615 + }, + { + "epoch": 0.1774654074236767, + "grad_norm": 2.350508451461792, + "learning_rate": 8.863885839736552e-07, + "loss": 1.0842, + "mean_token_accuracy": 0.6799224019050598, + "num_tokens": 41338158.0, + "step": 1616 + }, + { + "epoch": 0.17757522512629037, + "grad_norm": 2.31611704826355, + "learning_rate": 8.869374313940724e-07, + "loss": 0.9289, + "mean_token_accuracy": 0.7187235951423645, + "num_tokens": 41360323.0, + "step": 1617 + }, + { + "epoch": 0.17768504282890402, + "grad_norm": 2.919203996658325, + "learning_rate": 8.874862788144896e-07, + "loss": 0.8991, + "mean_token_accuracy": 0.7236604690551758, + "num_tokens": 41377348.0, + "step": 1618 + }, + { + "epoch": 0.1777948605315177, + "grad_norm": 2.38997220993042, + "learning_rate": 8.880351262349066e-07, + "loss": 1.0544, + "mean_token_accuracy": 0.6906465291976929, + "num_tokens": 41401395.0, + "step": 1619 + }, + { + "epoch": 0.17790467823413134, + "grad_norm": 2.424755811691284, + "learning_rate": 8.885839736553238e-07, + "loss": 1.023, + "mean_token_accuracy": 0.7000770568847656, + "num_tokens": 41424587.0, + "step": 1620 + }, + { + "epoch": 0.178014495936745, + "grad_norm": 2.1982922554016113, + "learning_rate": 8.89132821075741e-07, + "loss": 1.0576, + "mean_token_accuracy": 0.6837252378463745, + "num_tokens": 41453252.0, + "step": 1621 + }, + { + "epoch": 0.17812431363935866, + "grad_norm": 2.289029836654663, + "learning_rate": 8.89681668496158e-07, + "loss": 1.0792, + "mean_token_accuracy": 0.6766232848167419, + "num_tokens": 41480938.0, + "step": 1622 + }, + { + "epoch": 0.17823413134197233, + "grad_norm": 2.5588059425354004, + "learning_rate": 8.902305159165752e-07, + "loss": 0.9995, + "mean_token_accuracy": 0.6952502131462097, + "num_tokens": 41500989.0, + "step": 1623 + }, + { + "epoch": 0.17834394904458598, + "grad_norm": 2.1907029151916504, + "learning_rate": 8.907793633369924e-07, + "loss": 1.0689, + "mean_token_accuracy": 0.6872082948684692, + "num_tokens": 41529234.0, + "step": 1624 + }, + { + "epoch": 0.17845376674719965, + "grad_norm": 2.1816418170928955, + "learning_rate": 8.913282107574093e-07, + "loss": 1.0275, + "mean_token_accuracy": 0.6970343589782715, + "num_tokens": 41557002.0, + "step": 1625 + }, + { + "epoch": 0.17856358444981332, + "grad_norm": 2.8291561603546143, + "learning_rate": 8.918770581778265e-07, + "loss": 1.0066, + "mean_token_accuracy": 0.6949343085289001, + "num_tokens": 41575982.0, + "step": 1626 + }, + { + "epoch": 0.17867340215242697, + "grad_norm": 2.481424570083618, + "learning_rate": 8.924259055982436e-07, + "loss": 1.0821, + "mean_token_accuracy": 0.6830441951751709, + "num_tokens": 41599113.0, + "step": 1627 + }, + { + "epoch": 0.17878321985504064, + "grad_norm": 2.823976516723633, + "learning_rate": 8.929747530186607e-07, + "loss": 0.97, + "mean_token_accuracy": 0.705672025680542, + "num_tokens": 41617696.0, + "step": 1628 + }, + { + "epoch": 0.1788930375576543, + "grad_norm": 2.2707226276397705, + "learning_rate": 8.935236004390779e-07, + "loss": 1.0712, + "mean_token_accuracy": 0.6831728219985962, + "num_tokens": 41642030.0, + "step": 1629 + }, + { + "epoch": 0.17900285526026796, + "grad_norm": 2.225627899169922, + "learning_rate": 8.94072447859495e-07, + "loss": 1.0456, + "mean_token_accuracy": 0.6928434371948242, + "num_tokens": 41671024.0, + "step": 1630 + }, + { + "epoch": 0.1791126729628816, + "grad_norm": 2.3055508136749268, + "learning_rate": 8.946212952799121e-07, + "loss": 1.0709, + "mean_token_accuracy": 0.6980085372924805, + "num_tokens": 41695898.0, + "step": 1631 + }, + { + "epoch": 0.17922249066549528, + "grad_norm": 2.5693371295928955, + "learning_rate": 8.951701427003293e-07, + "loss": 1.0153, + "mean_token_accuracy": 0.6921200752258301, + "num_tokens": 41716099.0, + "step": 1632 + }, + { + "epoch": 0.17933230836810893, + "grad_norm": 2.0970335006713867, + "learning_rate": 8.957189901207464e-07, + "loss": 1.0064, + "mean_token_accuracy": 0.7004308104515076, + "num_tokens": 41743234.0, + "step": 1633 + }, + { + "epoch": 0.1794421260707226, + "grad_norm": 2.322857618331909, + "learning_rate": 8.962678375411635e-07, + "loss": 1.0585, + "mean_token_accuracy": 0.680281400680542, + "num_tokens": 41768384.0, + "step": 1634 + }, + { + "epoch": 0.17955194377333628, + "grad_norm": 2.364114284515381, + "learning_rate": 8.968166849615807e-07, + "loss": 0.9534, + "mean_token_accuracy": 0.7126083374023438, + "num_tokens": 41791246.0, + "step": 1635 + }, + { + "epoch": 0.17966176147594992, + "grad_norm": 2.2888216972351074, + "learning_rate": 8.973655323819978e-07, + "loss": 1.003, + "mean_token_accuracy": 0.7019110321998596, + "num_tokens": 41817171.0, + "step": 1636 + }, + { + "epoch": 0.1797715791785636, + "grad_norm": 2.3241732120513916, + "learning_rate": 8.979143798024149e-07, + "loss": 0.9293, + "mean_token_accuracy": 0.7174392938613892, + "num_tokens": 41840828.0, + "step": 1637 + }, + { + "epoch": 0.17988139688117724, + "grad_norm": 2.314108371734619, + "learning_rate": 8.98463227222832e-07, + "loss": 1.0026, + "mean_token_accuracy": 0.7001985907554626, + "num_tokens": 41864208.0, + "step": 1638 + }, + { + "epoch": 0.17999121458379091, + "grad_norm": 2.247187614440918, + "learning_rate": 8.990120746432492e-07, + "loss": 1.0358, + "mean_token_accuracy": 0.6960119009017944, + "num_tokens": 41889695.0, + "step": 1639 + }, + { + "epoch": 0.18010103228640456, + "grad_norm": 2.413201093673706, + "learning_rate": 8.995609220636663e-07, + "loss": 0.9429, + "mean_token_accuracy": 0.7136197090148926, + "num_tokens": 41913796.0, + "step": 1640 + }, + { + "epoch": 0.18021084998901823, + "grad_norm": 2.2713868618011475, + "learning_rate": 9.001097694840834e-07, + "loss": 1.0549, + "mean_token_accuracy": 0.6847524046897888, + "num_tokens": 41943821.0, + "step": 1641 + }, + { + "epoch": 0.18032066769163188, + "grad_norm": 2.5845754146575928, + "learning_rate": 9.006586169045006e-07, + "loss": 0.8871, + "mean_token_accuracy": 0.7230231761932373, + "num_tokens": 41964234.0, + "step": 1642 + }, + { + "epoch": 0.18043048539424555, + "grad_norm": 1.9851984977722168, + "learning_rate": 9.012074643249177e-07, + "loss": 0.9694, + "mean_token_accuracy": 0.7098394632339478, + "num_tokens": 41997959.0, + "step": 1643 + }, + { + "epoch": 0.1805403030968592, + "grad_norm": 2.1221518516540527, + "learning_rate": 9.017563117453347e-07, + "loss": 1.1184, + "mean_token_accuracy": 0.6695950031280518, + "num_tokens": 42028986.0, + "step": 1644 + }, + { + "epoch": 0.18065012079947287, + "grad_norm": 2.2859439849853516, + "learning_rate": 9.023051591657519e-07, + "loss": 1.0566, + "mean_token_accuracy": 0.6797901391983032, + "num_tokens": 42053746.0, + "step": 1645 + }, + { + "epoch": 0.18075993850208655, + "grad_norm": 2.2431480884552, + "learning_rate": 9.02854006586169e-07, + "loss": 1.0473, + "mean_token_accuracy": 0.6907745599746704, + "num_tokens": 42079961.0, + "step": 1646 + }, + { + "epoch": 0.1808697562047002, + "grad_norm": 2.4279494285583496, + "learning_rate": 9.034028540065861e-07, + "loss": 1.0859, + "mean_token_accuracy": 0.6728972792625427, + "num_tokens": 42103872.0, + "step": 1647 + }, + { + "epoch": 0.18097957390731387, + "grad_norm": 2.394460678100586, + "learning_rate": 9.039517014270033e-07, + "loss": 1.1367, + "mean_token_accuracy": 0.6619852781295776, + "num_tokens": 42129186.0, + "step": 1648 + }, + { + "epoch": 0.1810893916099275, + "grad_norm": 2.5521862506866455, + "learning_rate": 9.045005488474203e-07, + "loss": 1.0262, + "mean_token_accuracy": 0.6917643547058105, + "num_tokens": 42148892.0, + "step": 1649 + }, + { + "epoch": 0.1811992093125412, + "grad_norm": 2.0802528858184814, + "learning_rate": 9.050493962678375e-07, + "loss": 1.0107, + "mean_token_accuracy": 0.6951476335525513, + "num_tokens": 42178502.0, + "step": 1650 + }, + { + "epoch": 0.18130902701515483, + "grad_norm": 2.5104026794433594, + "learning_rate": 9.055982436882547e-07, + "loss": 1.0017, + "mean_token_accuracy": 0.6937829256057739, + "num_tokens": 42199951.0, + "step": 1651 + }, + { + "epoch": 0.1814188447177685, + "grad_norm": 2.2112388610839844, + "learning_rate": 9.061470911086717e-07, + "loss": 1.0718, + "mean_token_accuracy": 0.679183840751648, + "num_tokens": 42227002.0, + "step": 1652 + }, + { + "epoch": 0.18152866242038215, + "grad_norm": 2.2971904277801514, + "learning_rate": 9.066959385290889e-07, + "loss": 1.0823, + "mean_token_accuracy": 0.678779125213623, + "num_tokens": 42253937.0, + "step": 1653 + }, + { + "epoch": 0.18163848012299583, + "grad_norm": 2.1329843997955322, + "learning_rate": 9.07244785949506e-07, + "loss": 1.07, + "mean_token_accuracy": 0.6839348077774048, + "num_tokens": 42283179.0, + "step": 1654 + }, + { + "epoch": 0.1817482978256095, + "grad_norm": 2.2343783378601074, + "learning_rate": 9.077936333699231e-07, + "loss": 0.9878, + "mean_token_accuracy": 0.699802041053772, + "num_tokens": 42310295.0, + "step": 1655 + }, + { + "epoch": 0.18185811552822315, + "grad_norm": 2.2300446033477783, + "learning_rate": 9.083424807903403e-07, + "loss": 0.9376, + "mean_token_accuracy": 0.7166253328323364, + "num_tokens": 42336216.0, + "step": 1656 + }, + { + "epoch": 0.18196793323083682, + "grad_norm": 2.525783061981201, + "learning_rate": 9.088913282107573e-07, + "loss": 1.0893, + "mean_token_accuracy": 0.6736230850219727, + "num_tokens": 42362150.0, + "step": 1657 + }, + { + "epoch": 0.18207775093345047, + "grad_norm": 2.2948098182678223, + "learning_rate": 9.094401756311745e-07, + "loss": 1.0312, + "mean_token_accuracy": 0.6898394227027893, + "num_tokens": 42386280.0, + "step": 1658 + }, + { + "epoch": 0.18218756863606414, + "grad_norm": 2.55346941947937, + "learning_rate": 9.099890230515917e-07, + "loss": 1.0866, + "mean_token_accuracy": 0.6719538569450378, + "num_tokens": 42408793.0, + "step": 1659 + }, + { + "epoch": 0.18229738633867779, + "grad_norm": 2.4584693908691406, + "learning_rate": 9.105378704720087e-07, + "loss": 0.9953, + "mean_token_accuracy": 0.7089016437530518, + "num_tokens": 42431034.0, + "step": 1660 + }, + { + "epoch": 0.18240720404129146, + "grad_norm": 2.3822124004364014, + "learning_rate": 9.110867178924259e-07, + "loss": 1.086, + "mean_token_accuracy": 0.6837935447692871, + "num_tokens": 42454171.0, + "step": 1661 + }, + { + "epoch": 0.1825170217439051, + "grad_norm": 2.0113348960876465, + "learning_rate": 9.11635565312843e-07, + "loss": 1.0041, + "mean_token_accuracy": 0.6952787637710571, + "num_tokens": 42483357.0, + "step": 1662 + }, + { + "epoch": 0.18262683944651878, + "grad_norm": 2.6978464126586914, + "learning_rate": 9.1218441273326e-07, + "loss": 0.9874, + "mean_token_accuracy": 0.7042253017425537, + "num_tokens": 42502616.0, + "step": 1663 + }, + { + "epoch": 0.18273665714913245, + "grad_norm": 2.305274486541748, + "learning_rate": 9.127332601536772e-07, + "loss": 0.9986, + "mean_token_accuracy": 0.7021769881248474, + "num_tokens": 42526787.0, + "step": 1664 + }, + { + "epoch": 0.1828464748517461, + "grad_norm": 2.423718214035034, + "learning_rate": 9.132821075740944e-07, + "loss": 1.0545, + "mean_token_accuracy": 0.694096565246582, + "num_tokens": 42551431.0, + "step": 1665 + }, + { + "epoch": 0.18295629255435977, + "grad_norm": 2.4643654823303223, + "learning_rate": 9.138309549945114e-07, + "loss": 1.0447, + "mean_token_accuracy": 0.6891045570373535, + "num_tokens": 42573539.0, + "step": 1666 + }, + { + "epoch": 0.18306611025697342, + "grad_norm": 2.2501091957092285, + "learning_rate": 9.143798024149286e-07, + "loss": 0.9576, + "mean_token_accuracy": 0.7102280855178833, + "num_tokens": 42599647.0, + "step": 1667 + }, + { + "epoch": 0.1831759279595871, + "grad_norm": 2.21296763420105, + "learning_rate": 9.149286498353457e-07, + "loss": 1.0365, + "mean_token_accuracy": 0.7046021223068237, + "num_tokens": 42627307.0, + "step": 1668 + }, + { + "epoch": 0.18328574566220074, + "grad_norm": 2.42915415763855, + "learning_rate": 9.154774972557628e-07, + "loss": 1.0197, + "mean_token_accuracy": 0.6916933655738831, + "num_tokens": 42651441.0, + "step": 1669 + }, + { + "epoch": 0.1833955633648144, + "grad_norm": 2.3686108589172363, + "learning_rate": 9.1602634467618e-07, + "loss": 1.0538, + "mean_token_accuracy": 0.6892729997634888, + "num_tokens": 42676700.0, + "step": 1670 + }, + { + "epoch": 0.18350538106742806, + "grad_norm": 2.428591728210449, + "learning_rate": 9.165751920965971e-07, + "loss": 0.9665, + "mean_token_accuracy": 0.6996803283691406, + "num_tokens": 42698998.0, + "step": 1671 + }, + { + "epoch": 0.18361519877004173, + "grad_norm": 2.3208632469177246, + "learning_rate": 9.171240395170142e-07, + "loss": 1.0023, + "mean_token_accuracy": 0.7018932104110718, + "num_tokens": 42723174.0, + "step": 1672 + }, + { + "epoch": 0.1837250164726554, + "grad_norm": 2.2385904788970947, + "learning_rate": 9.176728869374314e-07, + "loss": 1.079, + "mean_token_accuracy": 0.6780858635902405, + "num_tokens": 42751723.0, + "step": 1673 + }, + { + "epoch": 0.18383483417526905, + "grad_norm": 2.5895538330078125, + "learning_rate": 9.182217343578485e-07, + "loss": 1.0743, + "mean_token_accuracy": 0.6902053356170654, + "num_tokens": 42772517.0, + "step": 1674 + }, + { + "epoch": 0.18394465187788273, + "grad_norm": 2.250617265701294, + "learning_rate": 9.187705817782656e-07, + "loss": 1.0152, + "mean_token_accuracy": 0.7029634714126587, + "num_tokens": 42797726.0, + "step": 1675 + }, + { + "epoch": 0.18405446958049637, + "grad_norm": 2.335696220397949, + "learning_rate": 9.193194291986828e-07, + "loss": 1.0772, + "mean_token_accuracy": 0.6828147172927856, + "num_tokens": 42822024.0, + "step": 1676 + }, + { + "epoch": 0.18416428728311005, + "grad_norm": 2.58333158493042, + "learning_rate": 9.198682766190999e-07, + "loss": 1.0149, + "mean_token_accuracy": 0.6999317407608032, + "num_tokens": 42843378.0, + "step": 1677 + }, + { + "epoch": 0.1842741049857237, + "grad_norm": 2.1563966274261475, + "learning_rate": 9.20417124039517e-07, + "loss": 1.1295, + "mean_token_accuracy": 0.668131411075592, + "num_tokens": 42872400.0, + "step": 1678 + }, + { + "epoch": 0.18438392268833736, + "grad_norm": 2.280522108078003, + "learning_rate": 9.209659714599341e-07, + "loss": 1.0769, + "mean_token_accuracy": 0.6819252967834473, + "num_tokens": 42898082.0, + "step": 1679 + }, + { + "epoch": 0.184493740390951, + "grad_norm": 1.9838122129440308, + "learning_rate": 9.215148188803513e-07, + "loss": 1.0843, + "mean_token_accuracy": 0.675693929195404, + "num_tokens": 42929971.0, + "step": 1680 + }, + { + "epoch": 0.18460355809356468, + "grad_norm": 1.8831619024276733, + "learning_rate": 9.220636663007683e-07, + "loss": 1.0143, + "mean_token_accuracy": 0.6995636820793152, + "num_tokens": 42964896.0, + "step": 1681 + }, + { + "epoch": 0.18471337579617833, + "grad_norm": 2.4596965312957764, + "learning_rate": 9.226125137211854e-07, + "loss": 1.0756, + "mean_token_accuracy": 0.6801037788391113, + "num_tokens": 42988584.0, + "step": 1682 + }, + { + "epoch": 0.184823193498792, + "grad_norm": 2.124675750732422, + "learning_rate": 9.231613611416026e-07, + "loss": 1.0558, + "mean_token_accuracy": 0.6884778738021851, + "num_tokens": 43016796.0, + "step": 1683 + }, + { + "epoch": 0.18493301120140568, + "grad_norm": 2.333085298538208, + "learning_rate": 9.237102085620197e-07, + "loss": 0.9779, + "mean_token_accuracy": 0.6987595558166504, + "num_tokens": 43039985.0, + "step": 1684 + }, + { + "epoch": 0.18504282890401932, + "grad_norm": 2.2525007724761963, + "learning_rate": 9.242590559824368e-07, + "loss": 0.9404, + "mean_token_accuracy": 0.7158507108688354, + "num_tokens": 43065110.0, + "step": 1685 + }, + { + "epoch": 0.185152646606633, + "grad_norm": 2.50325870513916, + "learning_rate": 9.24807903402854e-07, + "loss": 1.0083, + "mean_token_accuracy": 0.7052424550056458, + "num_tokens": 43087563.0, + "step": 1686 + }, + { + "epoch": 0.18526246430924664, + "grad_norm": 2.1211283206939697, + "learning_rate": 9.253567508232711e-07, + "loss": 1.0495, + "mean_token_accuracy": 0.6884695291519165, + "num_tokens": 43116791.0, + "step": 1687 + }, + { + "epoch": 0.18537228201186032, + "grad_norm": 2.3232438564300537, + "learning_rate": 9.259055982436882e-07, + "loss": 1.0419, + "mean_token_accuracy": 0.6893258094787598, + "num_tokens": 43140908.0, + "step": 1688 + }, + { + "epoch": 0.18548209971447396, + "grad_norm": 2.3318707942962646, + "learning_rate": 9.264544456641053e-07, + "loss": 0.8882, + "mean_token_accuracy": 0.7310359477996826, + "num_tokens": 43163763.0, + "step": 1689 + }, + { + "epoch": 0.18559191741708764, + "grad_norm": 2.545017719268799, + "learning_rate": 9.270032930845224e-07, + "loss": 1.0137, + "mean_token_accuracy": 0.6951171159744263, + "num_tokens": 43185067.0, + "step": 1690 + }, + { + "epoch": 0.18570173511970128, + "grad_norm": 2.261739730834961, + "learning_rate": 9.275521405049396e-07, + "loss": 1.034, + "mean_token_accuracy": 0.695175051689148, + "num_tokens": 43210044.0, + "step": 1691 + }, + { + "epoch": 0.18581155282231496, + "grad_norm": 2.5498428344726562, + "learning_rate": 9.281009879253567e-07, + "loss": 1.0117, + "mean_token_accuracy": 0.6931861042976379, + "num_tokens": 43232913.0, + "step": 1692 + }, + { + "epoch": 0.18592137052492863, + "grad_norm": 2.004796028137207, + "learning_rate": 9.286498353457738e-07, + "loss": 0.966, + "mean_token_accuracy": 0.7096947431564331, + "num_tokens": 43263495.0, + "step": 1693 + }, + { + "epoch": 0.18603118822754228, + "grad_norm": 2.1009161472320557, + "learning_rate": 9.29198682766191e-07, + "loss": 1.0214, + "mean_token_accuracy": 0.6929258108139038, + "num_tokens": 43291749.0, + "step": 1694 + }, + { + "epoch": 0.18614100593015595, + "grad_norm": 2.1886651515960693, + "learning_rate": 9.297475301866081e-07, + "loss": 1.0461, + "mean_token_accuracy": 0.6856807470321655, + "num_tokens": 43319473.0, + "step": 1695 + }, + { + "epoch": 0.1862508236327696, + "grad_norm": 2.502271890640259, + "learning_rate": 9.302963776070252e-07, + "loss": 1.0407, + "mean_token_accuracy": 0.6878035664558411, + "num_tokens": 43341550.0, + "step": 1696 + }, + { + "epoch": 0.18636064133538327, + "grad_norm": 2.4640939235687256, + "learning_rate": 9.308452250274424e-07, + "loss": 1.0442, + "mean_token_accuracy": 0.6919176578521729, + "num_tokens": 43363698.0, + "step": 1697 + }, + { + "epoch": 0.18647045903799692, + "grad_norm": 2.354450225830078, + "learning_rate": 9.313940724478595e-07, + "loss": 1.0788, + "mean_token_accuracy": 0.6746251583099365, + "num_tokens": 43387379.0, + "step": 1698 + }, + { + "epoch": 0.1865802767406106, + "grad_norm": 2.5071773529052734, + "learning_rate": 9.319429198682766e-07, + "loss": 0.9832, + "mean_token_accuracy": 0.7018500566482544, + "num_tokens": 43408087.0, + "step": 1699 + }, + { + "epoch": 0.18669009444322424, + "grad_norm": 2.1299116611480713, + "learning_rate": 9.324917672886937e-07, + "loss": 1.0897, + "mean_token_accuracy": 0.6708160638809204, + "num_tokens": 43436556.0, + "step": 1700 + }, + { + "epoch": 0.1867999121458379, + "grad_norm": 2.2706031799316406, + "learning_rate": 9.330406147091107e-07, + "loss": 1.0428, + "mean_token_accuracy": 0.681452751159668, + "num_tokens": 43462882.0, + "step": 1701 + }, + { + "epoch": 0.18690972984845158, + "grad_norm": 2.340949535369873, + "learning_rate": 9.335894621295279e-07, + "loss": 1.0483, + "mean_token_accuracy": 0.6986857652664185, + "num_tokens": 43487649.0, + "step": 1702 + }, + { + "epoch": 0.18701954755106523, + "grad_norm": 2.1984429359436035, + "learning_rate": 9.341383095499451e-07, + "loss": 1.0224, + "mean_token_accuracy": 0.6961643695831299, + "num_tokens": 43514613.0, + "step": 1703 + }, + { + "epoch": 0.1871293652536789, + "grad_norm": 2.452366590499878, + "learning_rate": 9.346871569703621e-07, + "loss": 0.9772, + "mean_token_accuracy": 0.7015194892883301, + "num_tokens": 43536575.0, + "step": 1704 + }, + { + "epoch": 0.18723918295629255, + "grad_norm": 2.349060535430908, + "learning_rate": 9.352360043907793e-07, + "loss": 1.0892, + "mean_token_accuracy": 0.6804033517837524, + "num_tokens": 43560114.0, + "step": 1705 + }, + { + "epoch": 0.18734900065890622, + "grad_norm": 2.362168073654175, + "learning_rate": 9.357848518111965e-07, + "loss": 0.9376, + "mean_token_accuracy": 0.7097727060317993, + "num_tokens": 43585482.0, + "step": 1706 + }, + { + "epoch": 0.18745881836151987, + "grad_norm": 2.1805546283721924, + "learning_rate": 9.363336992316135e-07, + "loss": 1.0343, + "mean_token_accuracy": 0.6839865446090698, + "num_tokens": 43614995.0, + "step": 1707 + }, + { + "epoch": 0.18756863606413354, + "grad_norm": 2.0412542819976807, + "learning_rate": 9.368825466520307e-07, + "loss": 0.9665, + "mean_token_accuracy": 0.7105109691619873, + "num_tokens": 43645923.0, + "step": 1708 + }, + { + "epoch": 0.1876784537667472, + "grad_norm": 2.277287483215332, + "learning_rate": 9.374313940724479e-07, + "loss": 1.1025, + "mean_token_accuracy": 0.6723883152008057, + "num_tokens": 43671436.0, + "step": 1709 + }, + { + "epoch": 0.18778827146936086, + "grad_norm": 2.9274439811706543, + "learning_rate": 9.379802414928649e-07, + "loss": 0.9801, + "mean_token_accuracy": 0.7016295194625854, + "num_tokens": 43690001.0, + "step": 1710 + }, + { + "epoch": 0.18789808917197454, + "grad_norm": 2.2760355472564697, + "learning_rate": 9.385290889132821e-07, + "loss": 1.1101, + "mean_token_accuracy": 0.6676499247550964, + "num_tokens": 43720123.0, + "step": 1711 + }, + { + "epoch": 0.18800790687458818, + "grad_norm": 2.3047125339508057, + "learning_rate": 9.390779363336992e-07, + "loss": 1.0556, + "mean_token_accuracy": 0.6854813098907471, + "num_tokens": 43746176.0, + "step": 1712 + }, + { + "epoch": 0.18811772457720186, + "grad_norm": 2.2868003845214844, + "learning_rate": 9.396267837541163e-07, + "loss": 1.0433, + "mean_token_accuracy": 0.6879456043243408, + "num_tokens": 43772124.0, + "step": 1713 + }, + { + "epoch": 0.1882275422798155, + "grad_norm": 2.0163938999176025, + "learning_rate": 9.401756311745335e-07, + "loss": 1.0469, + "mean_token_accuracy": 0.6939710974693298, + "num_tokens": 43805927.0, + "step": 1714 + }, + { + "epoch": 0.18833735998242918, + "grad_norm": 2.199021816253662, + "learning_rate": 9.407244785949506e-07, + "loss": 1.0195, + "mean_token_accuracy": 0.6966198086738586, + "num_tokens": 43832154.0, + "step": 1715 + }, + { + "epoch": 0.18844717768504282, + "grad_norm": 2.1687755584716797, + "learning_rate": 9.412733260153677e-07, + "loss": 1.0719, + "mean_token_accuracy": 0.6820591688156128, + "num_tokens": 43860402.0, + "step": 1716 + }, + { + "epoch": 0.1885569953876565, + "grad_norm": 2.264967441558838, + "learning_rate": 9.418221734357849e-07, + "loss": 0.9417, + "mean_token_accuracy": 0.7182168960571289, + "num_tokens": 43885020.0, + "step": 1717 + }, + { + "epoch": 0.18866681309027014, + "grad_norm": 2.0816421508789062, + "learning_rate": 9.42371020856202e-07, + "loss": 1.0203, + "mean_token_accuracy": 0.699084997177124, + "num_tokens": 43913827.0, + "step": 1718 + }, + { + "epoch": 0.18877663079288381, + "grad_norm": 2.5954954624176025, + "learning_rate": 9.42919868276619e-07, + "loss": 1.0298, + "mean_token_accuracy": 0.6855543851852417, + "num_tokens": 43932929.0, + "step": 1719 + }, + { + "epoch": 0.18888644849549746, + "grad_norm": 2.0308425426483154, + "learning_rate": 9.434687156970362e-07, + "loss": 0.9717, + "mean_token_accuracy": 0.6937326192855835, + "num_tokens": 43961512.0, + "step": 1720 + }, + { + "epoch": 0.18899626619811113, + "grad_norm": 2.144012689590454, + "learning_rate": 9.440175631174532e-07, + "loss": 1.011, + "mean_token_accuracy": 0.6967803239822388, + "num_tokens": 43988738.0, + "step": 1721 + }, + { + "epoch": 0.1891060839007248, + "grad_norm": 2.181427478790283, + "learning_rate": 9.445664105378704e-07, + "loss": 1.0569, + "mean_token_accuracy": 0.6802094578742981, + "num_tokens": 44016394.0, + "step": 1722 + }, + { + "epoch": 0.18921590160333845, + "grad_norm": 2.087789297103882, + "learning_rate": 9.451152579582875e-07, + "loss": 1.0322, + "mean_token_accuracy": 0.6952117085456848, + "num_tokens": 44045950.0, + "step": 1723 + }, + { + "epoch": 0.18932571930595213, + "grad_norm": 2.117680549621582, + "learning_rate": 9.456641053787046e-07, + "loss": 0.9304, + "mean_token_accuracy": 0.7064507007598877, + "num_tokens": 44071860.0, + "step": 1724 + }, + { + "epoch": 0.18943553700856577, + "grad_norm": 2.3822402954101562, + "learning_rate": 9.462129527991218e-07, + "loss": 1.0512, + "mean_token_accuracy": 0.6882939338684082, + "num_tokens": 44095608.0, + "step": 1725 + }, + { + "epoch": 0.18954535471117945, + "grad_norm": 2.145439863204956, + "learning_rate": 9.467618002195389e-07, + "loss": 1.0873, + "mean_token_accuracy": 0.6880396604537964, + "num_tokens": 44125312.0, + "step": 1726 + }, + { + "epoch": 0.1896551724137931, + "grad_norm": 2.3106510639190674, + "learning_rate": 9.47310647639956e-07, + "loss": 0.9397, + "mean_token_accuracy": 0.7144153714179993, + "num_tokens": 44151276.0, + "step": 1727 + }, + { + "epoch": 0.18976499011640677, + "grad_norm": 2.207639455795288, + "learning_rate": 9.478594950603732e-07, + "loss": 1.0514, + "mean_token_accuracy": 0.6832361221313477, + "num_tokens": 44177694.0, + "step": 1728 + }, + { + "epoch": 0.1898748078190204, + "grad_norm": 2.486145257949829, + "learning_rate": 9.484083424807903e-07, + "loss": 0.9842, + "mean_token_accuracy": 0.7008037567138672, + "num_tokens": 44199907.0, + "step": 1729 + }, + { + "epoch": 0.1899846255216341, + "grad_norm": 2.073145866394043, + "learning_rate": 9.489571899012074e-07, + "loss": 1.0514, + "mean_token_accuracy": 0.689173698425293, + "num_tokens": 44231137.0, + "step": 1730 + }, + { + "epoch": 0.19009444322424776, + "grad_norm": 2.002730369567871, + "learning_rate": 9.495060373216246e-07, + "loss": 0.9712, + "mean_token_accuracy": 0.7041199207305908, + "num_tokens": 44261504.0, + "step": 1731 + }, + { + "epoch": 0.1902042609268614, + "grad_norm": 2.226876735687256, + "learning_rate": 9.500548847420417e-07, + "loss": 0.9934, + "mean_token_accuracy": 0.6993594765663147, + "num_tokens": 44286201.0, + "step": 1732 + }, + { + "epoch": 0.19031407862947508, + "grad_norm": 2.4035487174987793, + "learning_rate": 9.506037321624588e-07, + "loss": 0.9479, + "mean_token_accuracy": 0.7074925303459167, + "num_tokens": 44308401.0, + "step": 1733 + }, + { + "epoch": 0.19042389633208873, + "grad_norm": 2.4752724170684814, + "learning_rate": 9.511525795828759e-07, + "loss": 1.0367, + "mean_token_accuracy": 0.6907608509063721, + "num_tokens": 44331531.0, + "step": 1734 + }, + { + "epoch": 0.1905337140347024, + "grad_norm": 2.006405830383301, + "learning_rate": 9.517014270032931e-07, + "loss": 1.1034, + "mean_token_accuracy": 0.6801983118057251, + "num_tokens": 44363358.0, + "step": 1735 + }, + { + "epoch": 0.19064353173731605, + "grad_norm": 2.50480055809021, + "learning_rate": 9.522502744237102e-07, + "loss": 0.9513, + "mean_token_accuracy": 0.7122413516044617, + "num_tokens": 44385082.0, + "step": 1736 + }, + { + "epoch": 0.19075334943992972, + "grad_norm": 2.0294911861419678, + "learning_rate": 9.527991218441273e-07, + "loss": 1.0722, + "mean_token_accuracy": 0.6798945069313049, + "num_tokens": 44418975.0, + "step": 1737 + }, + { + "epoch": 0.19086316714254337, + "grad_norm": 2.044265031814575, + "learning_rate": 9.533479692645444e-07, + "loss": 1.0927, + "mean_token_accuracy": 0.6729153990745544, + "num_tokens": 44451396.0, + "step": 1738 + }, + { + "epoch": 0.19097298484515704, + "grad_norm": 2.4162254333496094, + "learning_rate": 9.538968166849616e-07, + "loss": 0.9341, + "mean_token_accuracy": 0.7135027647018433, + "num_tokens": 44474477.0, + "step": 1739 + }, + { + "epoch": 0.1910828025477707, + "grad_norm": 2.634392261505127, + "learning_rate": 9.544456641053787e-07, + "loss": 0.9591, + "mean_token_accuracy": 0.7089351415634155, + "num_tokens": 44493860.0, + "step": 1740 + }, + { + "epoch": 0.19119262025038436, + "grad_norm": 2.233532667160034, + "learning_rate": 9.549945115257958e-07, + "loss": 1.0896, + "mean_token_accuracy": 0.6756107211112976, + "num_tokens": 44522471.0, + "step": 1741 + }, + { + "epoch": 0.19130243795299803, + "grad_norm": 2.2509605884552, + "learning_rate": 9.55543358946213e-07, + "loss": 0.9794, + "mean_token_accuracy": 0.7017203569412231, + "num_tokens": 44548237.0, + "step": 1742 + }, + { + "epoch": 0.19141225565561168, + "grad_norm": 2.033938407897949, + "learning_rate": 9.5609220636663e-07, + "loss": 1.0364, + "mean_token_accuracy": 0.6845709681510925, + "num_tokens": 44580272.0, + "step": 1743 + }, + { + "epoch": 0.19152207335822535, + "grad_norm": 2.347269058227539, + "learning_rate": 9.566410537870472e-07, + "loss": 1.0923, + "mean_token_accuracy": 0.6715070605278015, + "num_tokens": 44604261.0, + "step": 1744 + }, + { + "epoch": 0.191631891060839, + "grad_norm": 2.08962082862854, + "learning_rate": 9.571899012074642e-07, + "loss": 1.0853, + "mean_token_accuracy": 0.67327880859375, + "num_tokens": 44635348.0, + "step": 1745 + }, + { + "epoch": 0.19174170876345267, + "grad_norm": 2.2921555042266846, + "learning_rate": 9.577387486278815e-07, + "loss": 1.012, + "mean_token_accuracy": 0.6940737962722778, + "num_tokens": 44659612.0, + "step": 1746 + }, + { + "epoch": 0.19185152646606632, + "grad_norm": 2.38653564453125, + "learning_rate": 9.582875960482986e-07, + "loss": 0.911, + "mean_token_accuracy": 0.7179057002067566, + "num_tokens": 44682222.0, + "step": 1747 + }, + { + "epoch": 0.19196134416868, + "grad_norm": 2.1998183727264404, + "learning_rate": 9.588364434687156e-07, + "loss": 1.1027, + "mean_token_accuracy": 0.6801414489746094, + "num_tokens": 44709611.0, + "step": 1748 + }, + { + "epoch": 0.19207116187129367, + "grad_norm": 2.60508131980896, + "learning_rate": 9.593852908891327e-07, + "loss": 0.9099, + "mean_token_accuracy": 0.7159061431884766, + "num_tokens": 44729848.0, + "step": 1749 + }, + { + "epoch": 0.1921809795739073, + "grad_norm": 2.2855873107910156, + "learning_rate": 9.5993413830955e-07, + "loss": 0.9873, + "mean_token_accuracy": 0.6985549330711365, + "num_tokens": 44753756.0, + "step": 1750 + }, + { + "epoch": 0.19229079727652099, + "grad_norm": 2.4272899627685547, + "learning_rate": 9.60482985729967e-07, + "loss": 1.0614, + "mean_token_accuracy": 0.6840373873710632, + "num_tokens": 44776067.0, + "step": 1751 + }, + { + "epoch": 0.19240061497913463, + "grad_norm": 2.276298999786377, + "learning_rate": 9.61031833150384e-07, + "loss": 1.0125, + "mean_token_accuracy": 0.6934646368026733, + "num_tokens": 44801906.0, + "step": 1752 + }, + { + "epoch": 0.1925104326817483, + "grad_norm": 2.5340588092803955, + "learning_rate": 9.615806805708014e-07, + "loss": 1.0278, + "mean_token_accuracy": 0.6944001317024231, + "num_tokens": 44824700.0, + "step": 1753 + }, + { + "epoch": 0.19262025038436195, + "grad_norm": 2.0707621574401855, + "learning_rate": 9.621295279912184e-07, + "loss": 1.0837, + "mean_token_accuracy": 0.6767796277999878, + "num_tokens": 44853738.0, + "step": 1754 + }, + { + "epoch": 0.19273006808697563, + "grad_norm": 2.3103625774383545, + "learning_rate": 9.626783754116355e-07, + "loss": 0.9979, + "mean_token_accuracy": 0.7045307159423828, + "num_tokens": 44883636.0, + "step": 1755 + }, + { + "epoch": 0.19283988578958927, + "grad_norm": 2.6786720752716064, + "learning_rate": 9.632272228320525e-07, + "loss": 0.9892, + "mean_token_accuracy": 0.6996361017227173, + "num_tokens": 44902671.0, + "step": 1756 + }, + { + "epoch": 0.19294970349220295, + "grad_norm": 2.0977184772491455, + "learning_rate": 9.637760702524698e-07, + "loss": 1.0621, + "mean_token_accuracy": 0.6820281744003296, + "num_tokens": 44931758.0, + "step": 1757 + }, + { + "epoch": 0.1930595211948166, + "grad_norm": 2.3845345973968506, + "learning_rate": 9.643249176728869e-07, + "loss": 1.0339, + "mean_token_accuracy": 0.6882013082504272, + "num_tokens": 44957167.0, + "step": 1758 + }, + { + "epoch": 0.19316933889743026, + "grad_norm": 2.2363061904907227, + "learning_rate": 9.64873765093304e-07, + "loss": 1.0422, + "mean_token_accuracy": 0.6869920492172241, + "num_tokens": 44984181.0, + "step": 1759 + }, + { + "epoch": 0.19327915660004394, + "grad_norm": 2.1896040439605713, + "learning_rate": 9.654226125137212e-07, + "loss": 1.04, + "mean_token_accuracy": 0.6884065866470337, + "num_tokens": 45011617.0, + "step": 1760 + }, + { + "epoch": 0.19338897430265758, + "grad_norm": 2.228404998779297, + "learning_rate": 9.659714599341383e-07, + "loss": 1.1132, + "mean_token_accuracy": 0.6795565485954285, + "num_tokens": 45040774.0, + "step": 1761 + }, + { + "epoch": 0.19349879200527126, + "grad_norm": 2.0032713413238525, + "learning_rate": 9.665203073545553e-07, + "loss": 1.106, + "mean_token_accuracy": 0.6761767864227295, + "num_tokens": 45072997.0, + "step": 1762 + }, + { + "epoch": 0.1936086097078849, + "grad_norm": 2.459158420562744, + "learning_rate": 9.670691547749726e-07, + "loss": 1.062, + "mean_token_accuracy": 0.6777925491333008, + "num_tokens": 45098463.0, + "step": 1763 + }, + { + "epoch": 0.19371842741049858, + "grad_norm": 2.2796053886413574, + "learning_rate": 9.676180021953897e-07, + "loss": 1.0337, + "mean_token_accuracy": 0.6897953152656555, + "num_tokens": 45125505.0, + "step": 1764 + }, + { + "epoch": 0.19382824511311222, + "grad_norm": 2.315352439880371, + "learning_rate": 9.681668496158067e-07, + "loss": 0.9546, + "mean_token_accuracy": 0.7136217355728149, + "num_tokens": 45149477.0, + "step": 1765 + }, + { + "epoch": 0.1939380628157259, + "grad_norm": 2.6508278846740723, + "learning_rate": 9.68715697036224e-07, + "loss": 1.0214, + "mean_token_accuracy": 0.6920447945594788, + "num_tokens": 45169230.0, + "step": 1766 + }, + { + "epoch": 0.19404788051833954, + "grad_norm": 2.867685079574585, + "learning_rate": 9.692645444566409e-07, + "loss": 1.0665, + "mean_token_accuracy": 0.6888900995254517, + "num_tokens": 45187452.0, + "step": 1767 + }, + { + "epoch": 0.19415769822095322, + "grad_norm": 2.3683645725250244, + "learning_rate": 9.698133918770581e-07, + "loss": 1.0288, + "mean_token_accuracy": 0.6938867568969727, + "num_tokens": 45210778.0, + "step": 1768 + }, + { + "epoch": 0.1942675159235669, + "grad_norm": 2.354473114013672, + "learning_rate": 9.703622392974752e-07, + "loss": 1.0467, + "mean_token_accuracy": 0.6896626949310303, + "num_tokens": 45233812.0, + "step": 1769 + }, + { + "epoch": 0.19437733362618054, + "grad_norm": 2.296292304992676, + "learning_rate": 9.709110867178923e-07, + "loss": 0.9666, + "mean_token_accuracy": 0.7065830826759338, + "num_tokens": 45257077.0, + "step": 1770 + }, + { + "epoch": 0.1944871513287942, + "grad_norm": 2.3976798057556152, + "learning_rate": 9.714599341383095e-07, + "loss": 1.0891, + "mean_token_accuracy": 0.6791232228279114, + "num_tokens": 45281076.0, + "step": 1771 + }, + { + "epoch": 0.19459696903140786, + "grad_norm": 2.0526392459869385, + "learning_rate": 9.720087815587266e-07, + "loss": 0.9994, + "mean_token_accuracy": 0.7013273239135742, + "num_tokens": 45310671.0, + "step": 1772 + }, + { + "epoch": 0.19470678673402153, + "grad_norm": 2.6619107723236084, + "learning_rate": 9.725576289791437e-07, + "loss": 1.0598, + "mean_token_accuracy": 0.6790699362754822, + "num_tokens": 45331540.0, + "step": 1773 + }, + { + "epoch": 0.19481660443663518, + "grad_norm": 2.2393174171447754, + "learning_rate": 9.73106476399561e-07, + "loss": 1.0105, + "mean_token_accuracy": 0.6909286379814148, + "num_tokens": 45358407.0, + "step": 1774 + }, + { + "epoch": 0.19492642213924885, + "grad_norm": 2.3353896141052246, + "learning_rate": 9.73655323819978e-07, + "loss": 1.0128, + "mean_token_accuracy": 0.6922290325164795, + "num_tokens": 45384745.0, + "step": 1775 + }, + { + "epoch": 0.1950362398418625, + "grad_norm": 2.5028274059295654, + "learning_rate": 9.74204171240395e-07, + "loss": 0.9826, + "mean_token_accuracy": 0.7029174566268921, + "num_tokens": 45406556.0, + "step": 1776 + }, + { + "epoch": 0.19514605754447617, + "grad_norm": 1.950307011604309, + "learning_rate": 9.747530186608123e-07, + "loss": 1.1146, + "mean_token_accuracy": 0.6663811206817627, + "num_tokens": 45442643.0, + "step": 1777 + }, + { + "epoch": 0.19525587524708984, + "grad_norm": 2.1312661170959473, + "learning_rate": 9.753018660812294e-07, + "loss": 1.1067, + "mean_token_accuracy": 0.668168306350708, + "num_tokens": 45471623.0, + "step": 1778 + }, + { + "epoch": 0.1953656929497035, + "grad_norm": 2.311141014099121, + "learning_rate": 9.758507135016465e-07, + "loss": 1.0166, + "mean_token_accuracy": 0.6904484033584595, + "num_tokens": 45496271.0, + "step": 1779 + }, + { + "epoch": 0.19547551065231716, + "grad_norm": 2.2412073612213135, + "learning_rate": 9.763995609220637e-07, + "loss": 1.0577, + "mean_token_accuracy": 0.6815783977508545, + "num_tokens": 45522308.0, + "step": 1780 + }, + { + "epoch": 0.1955853283549308, + "grad_norm": 2.0928311347961426, + "learning_rate": 9.769484083424808e-07, + "loss": 1.0441, + "mean_token_accuracy": 0.6892786622047424, + "num_tokens": 45552197.0, + "step": 1781 + }, + { + "epoch": 0.19569514605754448, + "grad_norm": 2.353288412094116, + "learning_rate": 9.774972557628979e-07, + "loss": 1.1452, + "mean_token_accuracy": 0.6621942520141602, + "num_tokens": 45576789.0, + "step": 1782 + }, + { + "epoch": 0.19580496376015813, + "grad_norm": 2.179934024810791, + "learning_rate": 9.780461031833151e-07, + "loss": 1.0156, + "mean_token_accuracy": 0.6939922571182251, + "num_tokens": 45604129.0, + "step": 1783 + }, + { + "epoch": 0.1959147814627718, + "grad_norm": 2.657238483428955, + "learning_rate": 9.785949506037322e-07, + "loss": 0.9874, + "mean_token_accuracy": 0.6968331336975098, + "num_tokens": 45624439.0, + "step": 1784 + }, + { + "epoch": 0.19602459916538545, + "grad_norm": 2.0697028636932373, + "learning_rate": 9.791437980241493e-07, + "loss": 0.9896, + "mean_token_accuracy": 0.7047231793403625, + "num_tokens": 45653262.0, + "step": 1785 + }, + { + "epoch": 0.19613441686799912, + "grad_norm": 2.517725706100464, + "learning_rate": 9.796926454445663e-07, + "loss": 1.0265, + "mean_token_accuracy": 0.6885583996772766, + "num_tokens": 45675362.0, + "step": 1786 + }, + { + "epoch": 0.1962442345706128, + "grad_norm": 2.546428680419922, + "learning_rate": 9.802414928649834e-07, + "loss": 1.0288, + "mean_token_accuracy": 0.6887452006340027, + "num_tokens": 45697774.0, + "step": 1787 + }, + { + "epoch": 0.19635405227322644, + "grad_norm": 2.219104051589966, + "learning_rate": 9.807903402854007e-07, + "loss": 0.9502, + "mean_token_accuracy": 0.7194811701774597, + "num_tokens": 45724207.0, + "step": 1788 + }, + { + "epoch": 0.19646386997584012, + "grad_norm": 2.054917573928833, + "learning_rate": 9.813391877058177e-07, + "loss": 1.0572, + "mean_token_accuracy": 0.6855285167694092, + "num_tokens": 45756125.0, + "step": 1789 + }, + { + "epoch": 0.19657368767845376, + "grad_norm": 2.435804843902588, + "learning_rate": 9.818880351262348e-07, + "loss": 0.9423, + "mean_token_accuracy": 0.7112891674041748, + "num_tokens": 45777753.0, + "step": 1790 + }, + { + "epoch": 0.19668350538106744, + "grad_norm": 2.3757476806640625, + "learning_rate": 9.82436882546652e-07, + "loss": 1.0724, + "mean_token_accuracy": 0.6780211329460144, + "num_tokens": 45803009.0, + "step": 1791 + }, + { + "epoch": 0.19679332308368108, + "grad_norm": 2.288301944732666, + "learning_rate": 9.829857299670691e-07, + "loss": 1.0416, + "mean_token_accuracy": 0.6946715116500854, + "num_tokens": 45829511.0, + "step": 1792 + }, + { + "epoch": 0.19690314078629476, + "grad_norm": 2.3035292625427246, + "learning_rate": 9.835345773874862e-07, + "loss": 0.9734, + "mean_token_accuracy": 0.7042684555053711, + "num_tokens": 45853874.0, + "step": 1793 + }, + { + "epoch": 0.1970129584889084, + "grad_norm": 2.160720109939575, + "learning_rate": 9.840834248079035e-07, + "loss": 0.9908, + "mean_token_accuracy": 0.7104499340057373, + "num_tokens": 45884233.0, + "step": 1794 + }, + { + "epoch": 0.19712277619152208, + "grad_norm": 2.5617969036102295, + "learning_rate": 9.846322722283205e-07, + "loss": 0.9314, + "mean_token_accuracy": 0.7180473208427429, + "num_tokens": 45905364.0, + "step": 1795 + }, + { + "epoch": 0.19723259389413572, + "grad_norm": 2.3294565677642822, + "learning_rate": 9.851811196487376e-07, + "loss": 1.056, + "mean_token_accuracy": 0.6835500597953796, + "num_tokens": 45929436.0, + "step": 1796 + }, + { + "epoch": 0.1973424115967494, + "grad_norm": 2.543447256088257, + "learning_rate": 9.857299670691546e-07, + "loss": 1.0374, + "mean_token_accuracy": 0.6847593784332275, + "num_tokens": 45950068.0, + "step": 1797 + }, + { + "epoch": 0.19745222929936307, + "grad_norm": 2.1944010257720947, + "learning_rate": 9.86278814489572e-07, + "loss": 0.9786, + "mean_token_accuracy": 0.7051383256912231, + "num_tokens": 45975878.0, + "step": 1798 + }, + { + "epoch": 0.19756204700197671, + "grad_norm": 2.030322313308716, + "learning_rate": 9.86827661909989e-07, + "loss": 1.0815, + "mean_token_accuracy": 0.6783843636512756, + "num_tokens": 46006942.0, + "step": 1799 + }, + { + "epoch": 0.1976718647045904, + "grad_norm": 2.2529408931732178, + "learning_rate": 9.87376509330406e-07, + "loss": 1.0034, + "mean_token_accuracy": 0.6952861547470093, + "num_tokens": 46034450.0, + "step": 1800 + }, + { + "epoch": 0.19778168240720403, + "grad_norm": 2.4252712726593018, + "learning_rate": 9.879253567508233e-07, + "loss": 1.0122, + "mean_token_accuracy": 0.6958602666854858, + "num_tokens": 46058510.0, + "step": 1801 + }, + { + "epoch": 0.1978915001098177, + "grad_norm": 2.49257493019104, + "learning_rate": 9.884742041712404e-07, + "loss": 0.9691, + "mean_token_accuracy": 0.7080553770065308, + "num_tokens": 46078399.0, + "step": 1802 + }, + { + "epoch": 0.19800131781243135, + "grad_norm": 2.30922269821167, + "learning_rate": 9.890230515916574e-07, + "loss": 1.0125, + "mean_token_accuracy": 0.6956737041473389, + "num_tokens": 46103968.0, + "step": 1803 + }, + { + "epoch": 0.19811113551504503, + "grad_norm": 2.4281978607177734, + "learning_rate": 9.895718990120747e-07, + "loss": 0.9739, + "mean_token_accuracy": 0.7051549553871155, + "num_tokens": 46130039.0, + "step": 1804 + }, + { + "epoch": 0.19822095321765867, + "grad_norm": 2.2759780883789062, + "learning_rate": 9.901207464324918e-07, + "loss": 0.979, + "mean_token_accuracy": 0.7082558274269104, + "num_tokens": 46154785.0, + "step": 1805 + }, + { + "epoch": 0.19833077092027235, + "grad_norm": 2.3043220043182373, + "learning_rate": 9.906695938529088e-07, + "loss": 0.9581, + "mean_token_accuracy": 0.7158856391906738, + "num_tokens": 46179648.0, + "step": 1806 + }, + { + "epoch": 0.19844058862288602, + "grad_norm": 2.1171107292175293, + "learning_rate": 9.91218441273326e-07, + "loss": 1.0088, + "mean_token_accuracy": 0.6943789720535278, + "num_tokens": 46208711.0, + "step": 1807 + }, + { + "epoch": 0.19855040632549967, + "grad_norm": 2.3184430599212646, + "learning_rate": 9.91767288693743e-07, + "loss": 1.1048, + "mean_token_accuracy": 0.691165030002594, + "num_tokens": 46235401.0, + "step": 1808 + }, + { + "epoch": 0.19866022402811334, + "grad_norm": 2.5602731704711914, + "learning_rate": 9.923161361141602e-07, + "loss": 1.0389, + "mean_token_accuracy": 0.6874892115592957, + "num_tokens": 46257186.0, + "step": 1809 + }, + { + "epoch": 0.198770041730727, + "grad_norm": 2.3102145195007324, + "learning_rate": 9.928649835345773e-07, + "loss": 1.0312, + "mean_token_accuracy": 0.6916648149490356, + "num_tokens": 46283987.0, + "step": 1810 + }, + { + "epoch": 0.19887985943334066, + "grad_norm": 2.369692802429199, + "learning_rate": 9.934138309549944e-07, + "loss": 1.0524, + "mean_token_accuracy": 0.6788625717163086, + "num_tokens": 46309399.0, + "step": 1811 + }, + { + "epoch": 0.1989896771359543, + "grad_norm": 2.4970200061798096, + "learning_rate": 9.939626783754116e-07, + "loss": 0.9645, + "mean_token_accuracy": 0.7093005180358887, + "num_tokens": 46331178.0, + "step": 1812 + }, + { + "epoch": 0.19909949483856798, + "grad_norm": 2.271450996398926, + "learning_rate": 9.945115257958287e-07, + "loss": 1.0703, + "mean_token_accuracy": 0.6842641234397888, + "num_tokens": 46356849.0, + "step": 1813 + }, + { + "epoch": 0.19920931254118163, + "grad_norm": 2.3806521892547607, + "learning_rate": 9.950603732162458e-07, + "loss": 0.9894, + "mean_token_accuracy": 0.7077219486236572, + "num_tokens": 46383564.0, + "step": 1814 + }, + { + "epoch": 0.1993191302437953, + "grad_norm": 2.234938859939575, + "learning_rate": 9.95609220636663e-07, + "loss": 1.1536, + "mean_token_accuracy": 0.6609184145927429, + "num_tokens": 46410229.0, + "step": 1815 + }, + { + "epoch": 0.19942894794640897, + "grad_norm": 2.7573697566986084, + "learning_rate": 9.9615806805708e-07, + "loss": 1.0358, + "mean_token_accuracy": 0.6912277936935425, + "num_tokens": 46434342.0, + "step": 1816 + }, + { + "epoch": 0.19953876564902262, + "grad_norm": 2.90201473236084, + "learning_rate": 9.967069154774972e-07, + "loss": 0.9971, + "mean_token_accuracy": 0.7041137218475342, + "num_tokens": 46451007.0, + "step": 1817 + }, + { + "epoch": 0.1996485833516363, + "grad_norm": 2.2506470680236816, + "learning_rate": 9.972557628979144e-07, + "loss": 1.0934, + "mean_token_accuracy": 0.6783000230789185, + "num_tokens": 46483285.0, + "step": 1818 + }, + { + "epoch": 0.19975840105424994, + "grad_norm": 2.608396053314209, + "learning_rate": 9.978046103183315e-07, + "loss": 0.9263, + "mean_token_accuracy": 0.7126508951187134, + "num_tokens": 46502209.0, + "step": 1819 + }, + { + "epoch": 0.1998682187568636, + "grad_norm": 2.168879508972168, + "learning_rate": 9.983534577387486e-07, + "loss": 1.0398, + "mean_token_accuracy": 0.686578094959259, + "num_tokens": 46531601.0, + "step": 1820 + }, + { + "epoch": 0.19997803645947726, + "grad_norm": 2.275564193725586, + "learning_rate": 9.989023051591658e-07, + "loss": 1.0392, + "mean_token_accuracy": 0.6912411451339722, + "num_tokens": 46556449.0, + "step": 1821 + }, + { + "epoch": 0.20008785416209093, + "grad_norm": 1.9713053703308105, + "learning_rate": 9.994511525795829e-07, + "loss": 1.0929, + "mean_token_accuracy": 0.6706962585449219, + "num_tokens": 46588543.0, + "step": 1822 + }, + { + "epoch": 0.20019767186470458, + "grad_norm": 2.3027193546295166, + "learning_rate": 1e-06, + "loss": 1.1363, + "mean_token_accuracy": 0.6642377972602844, + "num_tokens": 46613757.0, + "step": 1823 + }, + { + "epoch": 0.20030748956731825, + "grad_norm": 2.513244152069092, + "learning_rate": 1e-06, + "loss": 1.0347, + "mean_token_accuracy": 0.7000427842140198, + "num_tokens": 46635470.0, + "step": 1824 + }, + { + "epoch": 0.2004173072699319, + "grad_norm": 2.6020865440368652, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7173088192939758, + "num_tokens": 46654305.0, + "step": 1825 + }, + { + "epoch": 0.20052712497254557, + "grad_norm": 2.3332204818725586, + "learning_rate": 1e-06, + "loss": 1.0848, + "mean_token_accuracy": 0.6791680455207825, + "num_tokens": 46681530.0, + "step": 1826 + }, + { + "epoch": 0.20063694267515925, + "grad_norm": 2.1316964626312256, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6967364549636841, + "num_tokens": 46710119.0, + "step": 1827 + }, + { + "epoch": 0.2007467603777729, + "grad_norm": 2.338561534881592, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6942013502120972, + "num_tokens": 46734243.0, + "step": 1828 + }, + { + "epoch": 0.20085657808038657, + "grad_norm": 2.1365091800689697, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7114748954772949, + "num_tokens": 46760888.0, + "step": 1829 + }, + { + "epoch": 0.2009663957830002, + "grad_norm": 2.5491883754730225, + "learning_rate": 1e-06, + "loss": 1.0941, + "mean_token_accuracy": 0.6750363111495972, + "num_tokens": 46783977.0, + "step": 1830 + }, + { + "epoch": 0.20107621348561389, + "grad_norm": 1.9590176343917847, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7069864273071289, + "num_tokens": 46814592.0, + "step": 1831 + }, + { + "epoch": 0.20118603118822753, + "grad_norm": 2.48539662361145, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7181174755096436, + "num_tokens": 46835429.0, + "step": 1832 + }, + { + "epoch": 0.2012958488908412, + "grad_norm": 2.28584885597229, + "learning_rate": 1e-06, + "loss": 1.0522, + "mean_token_accuracy": 0.6870952844619751, + "num_tokens": 46861851.0, + "step": 1833 + }, + { + "epoch": 0.20140566659345485, + "grad_norm": 2.69814133644104, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6990174651145935, + "num_tokens": 46880772.0, + "step": 1834 + }, + { + "epoch": 0.20151548429606853, + "grad_norm": 2.2452034950256348, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.7004464268684387, + "num_tokens": 46909737.0, + "step": 1835 + }, + { + "epoch": 0.2016253019986822, + "grad_norm": 2.6328747272491455, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.7003425359725952, + "num_tokens": 46929688.0, + "step": 1836 + }, + { + "epoch": 0.20173511970129585, + "grad_norm": 2.115058422088623, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.7021489143371582, + "num_tokens": 46957924.0, + "step": 1837 + }, + { + "epoch": 0.20184493740390952, + "grad_norm": 2.234545946121216, + "learning_rate": 1e-06, + "loss": 1.0901, + "mean_token_accuracy": 0.6942021250724792, + "num_tokens": 46985146.0, + "step": 1838 + }, + { + "epoch": 0.20195475510652316, + "grad_norm": 2.475194215774536, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7302334308624268, + "num_tokens": 47005421.0, + "step": 1839 + }, + { + "epoch": 0.20206457280913684, + "grad_norm": 2.654693126678467, + "learning_rate": 1e-06, + "loss": 1.054, + "mean_token_accuracy": 0.6867517232894897, + "num_tokens": 47026641.0, + "step": 1840 + }, + { + "epoch": 0.20217439051175048, + "grad_norm": 2.4638917446136475, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.6933250427246094, + "num_tokens": 47048081.0, + "step": 1841 + }, + { + "epoch": 0.20228420821436416, + "grad_norm": 2.333482503890991, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7082674503326416, + "num_tokens": 47071630.0, + "step": 1842 + }, + { + "epoch": 0.2023940259169778, + "grad_norm": 2.365332841873169, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.7065058946609497, + "num_tokens": 47095370.0, + "step": 1843 + }, + { + "epoch": 0.20250384361959148, + "grad_norm": 2.3013522624969482, + "learning_rate": 1e-06, + "loss": 1.0991, + "mean_token_accuracy": 0.675040602684021, + "num_tokens": 47121262.0, + "step": 1844 + }, + { + "epoch": 0.20261366132220515, + "grad_norm": 2.4589104652404785, + "learning_rate": 1e-06, + "loss": 1.0414, + "mean_token_accuracy": 0.6862367987632751, + "num_tokens": 47142933.0, + "step": 1845 + }, + { + "epoch": 0.2027234790248188, + "grad_norm": 2.055002450942993, + "learning_rate": 1e-06, + "loss": 1.0752, + "mean_token_accuracy": 0.6727116703987122, + "num_tokens": 47172525.0, + "step": 1846 + }, + { + "epoch": 0.20283329672743247, + "grad_norm": 2.0830326080322266, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7146694660186768, + "num_tokens": 47198480.0, + "step": 1847 + }, + { + "epoch": 0.20294311443004612, + "grad_norm": 2.529892683029175, + "learning_rate": 1e-06, + "loss": 1.0452, + "mean_token_accuracy": 0.6899440288543701, + "num_tokens": 47220124.0, + "step": 1848 + }, + { + "epoch": 0.2030529321326598, + "grad_norm": 2.4242823123931885, + "learning_rate": 1e-06, + "loss": 1.1107, + "mean_token_accuracy": 0.6705341339111328, + "num_tokens": 47244560.0, + "step": 1849 + }, + { + "epoch": 0.20316274983527344, + "grad_norm": 2.498946189880371, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6926785707473755, + "num_tokens": 47265742.0, + "step": 1850 + }, + { + "epoch": 0.2032725675378871, + "grad_norm": 2.49015474319458, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7184073328971863, + "num_tokens": 47286717.0, + "step": 1851 + }, + { + "epoch": 0.20338238524050076, + "grad_norm": 2.1045775413513184, + "learning_rate": 1e-06, + "loss": 1.0648, + "mean_token_accuracy": 0.6813512444496155, + "num_tokens": 47313453.0, + "step": 1852 + }, + { + "epoch": 0.20349220294311443, + "grad_norm": 2.1678411960601807, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6971445083618164, + "num_tokens": 47341931.0, + "step": 1853 + }, + { + "epoch": 0.2036020206457281, + "grad_norm": 2.269519329071045, + "learning_rate": 1e-06, + "loss": 1.068, + "mean_token_accuracy": 0.6795332431793213, + "num_tokens": 47369460.0, + "step": 1854 + }, + { + "epoch": 0.20371183834834175, + "grad_norm": 2.218919515609741, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.696904182434082, + "num_tokens": 47396860.0, + "step": 1855 + }, + { + "epoch": 0.20382165605095542, + "grad_norm": 2.0909125804901123, + "learning_rate": 1e-06, + "loss": 1.0968, + "mean_token_accuracy": 0.673143744468689, + "num_tokens": 47426033.0, + "step": 1856 + }, + { + "epoch": 0.20393147375356907, + "grad_norm": 2.0578949451446533, + "learning_rate": 1e-06, + "loss": 1.0921, + "mean_token_accuracy": 0.6847690343856812, + "num_tokens": 47458161.0, + "step": 1857 + }, + { + "epoch": 0.20404129145618274, + "grad_norm": 2.402831792831421, + "learning_rate": 1e-06, + "loss": 1.0502, + "mean_token_accuracy": 0.691557765007019, + "num_tokens": 47481334.0, + "step": 1858 + }, + { + "epoch": 0.2041511091587964, + "grad_norm": 2.351309299468994, + "learning_rate": 1e-06, + "loss": 1.0735, + "mean_token_accuracy": 0.6786553859710693, + "num_tokens": 47507958.0, + "step": 1859 + }, + { + "epoch": 0.20426092686141006, + "grad_norm": 2.4238619804382324, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7147888541221619, + "num_tokens": 47530783.0, + "step": 1860 + }, + { + "epoch": 0.2043707445640237, + "grad_norm": 2.4559919834136963, + "learning_rate": 1e-06, + "loss": 1.0515, + "mean_token_accuracy": 0.6888359785079956, + "num_tokens": 47553301.0, + "step": 1861 + }, + { + "epoch": 0.20448056226663738, + "grad_norm": 2.441408157348633, + "learning_rate": 1e-06, + "loss": 1.0787, + "mean_token_accuracy": 0.6783185005187988, + "num_tokens": 47577407.0, + "step": 1862 + }, + { + "epoch": 0.20459037996925103, + "grad_norm": 2.3348593711853027, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.7074417471885681, + "num_tokens": 47601285.0, + "step": 1863 + }, + { + "epoch": 0.2047001976718647, + "grad_norm": 2.1157150268554688, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6984650492668152, + "num_tokens": 47631835.0, + "step": 1864 + }, + { + "epoch": 0.20481001537447838, + "grad_norm": 2.050396680831909, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7109096050262451, + "num_tokens": 47659919.0, + "step": 1865 + }, + { + "epoch": 0.20491983307709202, + "grad_norm": 2.378657341003418, + "learning_rate": 1e-06, + "loss": 1.0878, + "mean_token_accuracy": 0.6743635535240173, + "num_tokens": 47683710.0, + "step": 1866 + }, + { + "epoch": 0.2050296507797057, + "grad_norm": 2.7365872859954834, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.7016863822937012, + "num_tokens": 47701494.0, + "step": 1867 + }, + { + "epoch": 0.20513946848231934, + "grad_norm": 2.2659192085266113, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6981157660484314, + "num_tokens": 47728177.0, + "step": 1868 + }, + { + "epoch": 0.20524928618493302, + "grad_norm": 2.2258639335632324, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7009007930755615, + "num_tokens": 47752155.0, + "step": 1869 + }, + { + "epoch": 0.20535910388754666, + "grad_norm": 2.1655547618865967, + "learning_rate": 1e-06, + "loss": 1.0515, + "mean_token_accuracy": 0.6868809461593628, + "num_tokens": 47782615.0, + "step": 1870 + }, + { + "epoch": 0.20546892159016034, + "grad_norm": 2.3538269996643066, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.696394681930542, + "num_tokens": 47808979.0, + "step": 1871 + }, + { + "epoch": 0.20557873929277398, + "grad_norm": 2.4103589057922363, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6860994696617126, + "num_tokens": 47832870.0, + "step": 1872 + }, + { + "epoch": 0.20568855699538766, + "grad_norm": 2.0956788063049316, + "learning_rate": 1e-06, + "loss": 1.0739, + "mean_token_accuracy": 0.6770166754722595, + "num_tokens": 47865188.0, + "step": 1873 + }, + { + "epoch": 0.20579837469800133, + "grad_norm": 2.0057523250579834, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6967742443084717, + "num_tokens": 47897744.0, + "step": 1874 + }, + { + "epoch": 0.20590819240061498, + "grad_norm": 2.1659958362579346, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.71153324842453, + "num_tokens": 47925645.0, + "step": 1875 + }, + { + "epoch": 0.20601801010322865, + "grad_norm": 2.177724838256836, + "learning_rate": 1e-06, + "loss": 1.0414, + "mean_token_accuracy": 0.6846160292625427, + "num_tokens": 47954104.0, + "step": 1876 + }, + { + "epoch": 0.2061278278058423, + "grad_norm": 2.532130718231201, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7161047458648682, + "num_tokens": 47974280.0, + "step": 1877 + }, + { + "epoch": 0.20623764550845597, + "grad_norm": 2.0336461067199707, + "learning_rate": 1e-06, + "loss": 1.1065, + "mean_token_accuracy": 0.6696026921272278, + "num_tokens": 48005287.0, + "step": 1878 + }, + { + "epoch": 0.20634746321106961, + "grad_norm": 2.6437220573425293, + "learning_rate": 1e-06, + "loss": 1.0395, + "mean_token_accuracy": 0.6841713190078735, + "num_tokens": 48028248.0, + "step": 1879 + }, + { + "epoch": 0.2064572809136833, + "grad_norm": 2.246614456176758, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.7013244032859802, + "num_tokens": 48053122.0, + "step": 1880 + }, + { + "epoch": 0.20656709861629693, + "grad_norm": 2.1543467044830322, + "learning_rate": 1e-06, + "loss": 1.1006, + "mean_token_accuracy": 0.6708167791366577, + "num_tokens": 48081956.0, + "step": 1881 + }, + { + "epoch": 0.2066769163189106, + "grad_norm": 2.273804187774658, + "learning_rate": 1e-06, + "loss": 1.046, + "mean_token_accuracy": 0.6897244453430176, + "num_tokens": 48107970.0, + "step": 1882 + }, + { + "epoch": 0.20678673402152428, + "grad_norm": 2.2680275440216064, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.6972048282623291, + "num_tokens": 48135538.0, + "step": 1883 + }, + { + "epoch": 0.20689655172413793, + "grad_norm": 2.405132532119751, + "learning_rate": 1e-06, + "loss": 1.0771, + "mean_token_accuracy": 0.6807844042778015, + "num_tokens": 48158089.0, + "step": 1884 + }, + { + "epoch": 0.2070063694267516, + "grad_norm": 2.2039072513580322, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7226579189300537, + "num_tokens": 48183758.0, + "step": 1885 + }, + { + "epoch": 0.20711618712936525, + "grad_norm": 2.155282497406006, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6882498264312744, + "num_tokens": 48213289.0, + "step": 1886 + }, + { + "epoch": 0.20722600483197892, + "grad_norm": 2.4165122509002686, + "learning_rate": 1e-06, + "loss": 1.0647, + "mean_token_accuracy": 0.6822922825813293, + "num_tokens": 48234962.0, + "step": 1887 + }, + { + "epoch": 0.20733582253459257, + "grad_norm": 2.294889211654663, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6858533620834351, + "num_tokens": 48261151.0, + "step": 1888 + }, + { + "epoch": 0.20744564023720624, + "grad_norm": 2.682572841644287, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.6972176432609558, + "num_tokens": 48283133.0, + "step": 1889 + }, + { + "epoch": 0.2075554579398199, + "grad_norm": 2.200028896331787, + "learning_rate": 1e-06, + "loss": 1.0402, + "mean_token_accuracy": 0.6898897886276245, + "num_tokens": 48311018.0, + "step": 1890 + }, + { + "epoch": 0.20766527564243356, + "grad_norm": 2.097991466522217, + "learning_rate": 1e-06, + "loss": 1.1358, + "mean_token_accuracy": 0.6659257411956787, + "num_tokens": 48343756.0, + "step": 1891 + }, + { + "epoch": 0.20777509334504723, + "grad_norm": 2.4066522121429443, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7088838815689087, + "num_tokens": 48366495.0, + "step": 1892 + }, + { + "epoch": 0.20788491104766088, + "grad_norm": 2.297111749649048, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.6921029686927795, + "num_tokens": 48392866.0, + "step": 1893 + }, + { + "epoch": 0.20799472875027455, + "grad_norm": 2.147834300994873, + "learning_rate": 1e-06, + "loss": 1.0582, + "mean_token_accuracy": 0.6771566867828369, + "num_tokens": 48420029.0, + "step": 1894 + }, + { + "epoch": 0.2081045464528882, + "grad_norm": 2.2842671871185303, + "learning_rate": 1e-06, + "loss": 1.0808, + "mean_token_accuracy": 0.6776063442230225, + "num_tokens": 48445153.0, + "step": 1895 + }, + { + "epoch": 0.20821436415550187, + "grad_norm": 2.2868685722351074, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6837844252586365, + "num_tokens": 48470168.0, + "step": 1896 + }, + { + "epoch": 0.20832418185811552, + "grad_norm": 2.8496341705322266, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.714709997177124, + "num_tokens": 48486847.0, + "step": 1897 + }, + { + "epoch": 0.2084339995607292, + "grad_norm": 2.796449661254883, + "learning_rate": 1e-06, + "loss": 1.0595, + "mean_token_accuracy": 0.6849372386932373, + "num_tokens": 48505926.0, + "step": 1898 + }, + { + "epoch": 0.20854381726334284, + "grad_norm": 2.44385027885437, + "learning_rate": 1e-06, + "loss": 1.0783, + "mean_token_accuracy": 0.6822844743728638, + "num_tokens": 48529926.0, + "step": 1899 + }, + { + "epoch": 0.2086536349659565, + "grad_norm": 2.553133964538574, + "learning_rate": 1e-06, + "loss": 1.1006, + "mean_token_accuracy": 0.6838258504867554, + "num_tokens": 48552307.0, + "step": 1900 + }, + { + "epoch": 0.20876345266857016, + "grad_norm": 1.980873942375183, + "learning_rate": 1e-06, + "loss": 1.054, + "mean_token_accuracy": 0.6860529184341431, + "num_tokens": 48583065.0, + "step": 1901 + }, + { + "epoch": 0.20887327037118383, + "grad_norm": 2.027776002883911, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7100578546524048, + "num_tokens": 48613182.0, + "step": 1902 + }, + { + "epoch": 0.2089830880737975, + "grad_norm": 2.4904236793518066, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6934614181518555, + "num_tokens": 48633452.0, + "step": 1903 + }, + { + "epoch": 0.20909290577641115, + "grad_norm": 2.0998640060424805, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6950647830963135, + "num_tokens": 48660322.0, + "step": 1904 + }, + { + "epoch": 0.20920272347902483, + "grad_norm": 2.2172658443450928, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7153409123420715, + "num_tokens": 48685662.0, + "step": 1905 + }, + { + "epoch": 0.20931254118163847, + "grad_norm": 2.364196538925171, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6999065279960632, + "num_tokens": 48707873.0, + "step": 1906 + }, + { + "epoch": 0.20942235888425215, + "grad_norm": 2.4063422679901123, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6954221129417419, + "num_tokens": 48729951.0, + "step": 1907 + }, + { + "epoch": 0.2095321765868658, + "grad_norm": 2.4764678478240967, + "learning_rate": 1e-06, + "loss": 1.0878, + "mean_token_accuracy": 0.6820201873779297, + "num_tokens": 48751700.0, + "step": 1908 + }, + { + "epoch": 0.20964199428947947, + "grad_norm": 2.261475086212158, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6897470355033875, + "num_tokens": 48776222.0, + "step": 1909 + }, + { + "epoch": 0.2097518119920931, + "grad_norm": 2.3947463035583496, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6896913647651672, + "num_tokens": 48801287.0, + "step": 1910 + }, + { + "epoch": 0.20986162969470679, + "grad_norm": 2.0228850841522217, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7187700271606445, + "num_tokens": 48829803.0, + "step": 1911 + }, + { + "epoch": 0.20997144739732046, + "grad_norm": 2.564877986907959, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7277208566665649, + "num_tokens": 48847722.0, + "step": 1912 + }, + { + "epoch": 0.2100812650999341, + "grad_norm": 2.588674545288086, + "learning_rate": 1e-06, + "loss": 1.0636, + "mean_token_accuracy": 0.6833575963973999, + "num_tokens": 48870153.0, + "step": 1913 + }, + { + "epoch": 0.21019108280254778, + "grad_norm": 2.1458942890167236, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6985551118850708, + "num_tokens": 48899040.0, + "step": 1914 + }, + { + "epoch": 0.21030090050516143, + "grad_norm": 2.3880996704101562, + "learning_rate": 1e-06, + "loss": 1.0527, + "mean_token_accuracy": 0.6860530972480774, + "num_tokens": 48922587.0, + "step": 1915 + }, + { + "epoch": 0.2104107182077751, + "grad_norm": 2.545480251312256, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.728163480758667, + "num_tokens": 48942408.0, + "step": 1916 + }, + { + "epoch": 0.21052053591038875, + "grad_norm": 2.275566816329956, + "learning_rate": 1e-06, + "loss": 1.0785, + "mean_token_accuracy": 0.6761937737464905, + "num_tokens": 48966986.0, + "step": 1917 + }, + { + "epoch": 0.21063035361300242, + "grad_norm": 2.520226001739502, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6968988180160522, + "num_tokens": 48988525.0, + "step": 1918 + }, + { + "epoch": 0.21074017131561606, + "grad_norm": 2.4388318061828613, + "learning_rate": 1e-06, + "loss": 1.0428, + "mean_token_accuracy": 0.6897082328796387, + "num_tokens": 49010971.0, + "step": 1919 + }, + { + "epoch": 0.21084998901822974, + "grad_norm": 2.096583843231201, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6895023584365845, + "num_tokens": 49042136.0, + "step": 1920 + }, + { + "epoch": 0.2109598067208434, + "grad_norm": 2.0488409996032715, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6914236545562744, + "num_tokens": 49069766.0, + "step": 1921 + }, + { + "epoch": 0.21106962442345706, + "grad_norm": 2.32594633102417, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6929502487182617, + "num_tokens": 49096277.0, + "step": 1922 + }, + { + "epoch": 0.21117944212607073, + "grad_norm": 2.105558395385742, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7167443037033081, + "num_tokens": 49123580.0, + "step": 1923 + }, + { + "epoch": 0.21128925982868438, + "grad_norm": 2.3163247108459473, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6918730735778809, + "num_tokens": 49147144.0, + "step": 1924 + }, + { + "epoch": 0.21139907753129805, + "grad_norm": 2.2932684421539307, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.706766664981842, + "num_tokens": 49174193.0, + "step": 1925 + }, + { + "epoch": 0.2115088952339117, + "grad_norm": 2.460555076599121, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7069861888885498, + "num_tokens": 49196312.0, + "step": 1926 + }, + { + "epoch": 0.21161871293652537, + "grad_norm": 2.054407835006714, + "learning_rate": 1e-06, + "loss": 1.0644, + "mean_token_accuracy": 0.6811555027961731, + "num_tokens": 49227283.0, + "step": 1927 + }, + { + "epoch": 0.21172853063913902, + "grad_norm": 2.251877784729004, + "learning_rate": 1e-06, + "loss": 1.0369, + "mean_token_accuracy": 0.6884908676147461, + "num_tokens": 49252931.0, + "step": 1928 + }, + { + "epoch": 0.2118383483417527, + "grad_norm": 2.3864150047302246, + "learning_rate": 1e-06, + "loss": 1.068, + "mean_token_accuracy": 0.6808903217315674, + "num_tokens": 49277021.0, + "step": 1929 + }, + { + "epoch": 0.21194816604436637, + "grad_norm": 2.2805016040802, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7085047960281372, + "num_tokens": 49301587.0, + "step": 1930 + }, + { + "epoch": 0.21205798374698, + "grad_norm": 2.4051942825317383, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.733830988407135, + "num_tokens": 49322318.0, + "step": 1931 + }, + { + "epoch": 0.21216780144959368, + "grad_norm": 2.373690128326416, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6968914270401001, + "num_tokens": 49347649.0, + "step": 1932 + }, + { + "epoch": 0.21227761915220733, + "grad_norm": 2.3411026000976562, + "learning_rate": 1e-06, + "loss": 1.0965, + "mean_token_accuracy": 0.6761943101882935, + "num_tokens": 49372143.0, + "step": 1933 + }, + { + "epoch": 0.212387436854821, + "grad_norm": 2.035910129547119, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6901968717575073, + "num_tokens": 49403822.0, + "step": 1934 + }, + { + "epoch": 0.21249725455743465, + "grad_norm": 2.468761920928955, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7324004173278809, + "num_tokens": 49423069.0, + "step": 1935 + }, + { + "epoch": 0.21260707226004832, + "grad_norm": 2.432884693145752, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6984744668006897, + "num_tokens": 49444716.0, + "step": 1936 + }, + { + "epoch": 0.21271688996266197, + "grad_norm": 2.04754638671875, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.6958156824111938, + "num_tokens": 49476179.0, + "step": 1937 + }, + { + "epoch": 0.21282670766527564, + "grad_norm": 2.205831289291382, + "learning_rate": 1e-06, + "loss": 1.1218, + "mean_token_accuracy": 0.6692982912063599, + "num_tokens": 49505411.0, + "step": 1938 + }, + { + "epoch": 0.2129365253678893, + "grad_norm": 1.891098141670227, + "learning_rate": 1e-06, + "loss": 1.0499, + "mean_token_accuracy": 0.6833625435829163, + "num_tokens": 49541834.0, + "step": 1939 + }, + { + "epoch": 0.21304634307050296, + "grad_norm": 2.515676259994507, + "learning_rate": 1e-06, + "loss": 1.0624, + "mean_token_accuracy": 0.6840407848358154, + "num_tokens": 49562829.0, + "step": 1940 + }, + { + "epoch": 0.21315616077311664, + "grad_norm": 2.11606502532959, + "learning_rate": 1e-06, + "loss": 1.0426, + "mean_token_accuracy": 0.6932578086853027, + "num_tokens": 49592651.0, + "step": 1941 + }, + { + "epoch": 0.21326597847573028, + "grad_norm": 2.0335264205932617, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6859308481216431, + "num_tokens": 49625634.0, + "step": 1942 + }, + { + "epoch": 0.21337579617834396, + "grad_norm": 2.330066680908203, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6977394819259644, + "num_tokens": 49649908.0, + "step": 1943 + }, + { + "epoch": 0.2134856138809576, + "grad_norm": 2.0895907878875732, + "learning_rate": 1e-06, + "loss": 1.1826, + "mean_token_accuracy": 0.6559913158416748, + "num_tokens": 49680097.0, + "step": 1944 + }, + { + "epoch": 0.21359543158357128, + "grad_norm": 2.1964101791381836, + "learning_rate": 1e-06, + "loss": 1.0839, + "mean_token_accuracy": 0.6791330575942993, + "num_tokens": 49710218.0, + "step": 1945 + }, + { + "epoch": 0.21370524928618492, + "grad_norm": 2.568178176879883, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.7078662514686584, + "num_tokens": 49732067.0, + "step": 1946 + }, + { + "epoch": 0.2138150669887986, + "grad_norm": 2.1208603382110596, + "learning_rate": 1e-06, + "loss": 1.1394, + "mean_token_accuracy": 0.6604553461074829, + "num_tokens": 49764254.0, + "step": 1947 + }, + { + "epoch": 0.21392488469141224, + "grad_norm": 2.009845495223999, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7146695852279663, + "num_tokens": 49793435.0, + "step": 1948 + }, + { + "epoch": 0.21403470239402592, + "grad_norm": 2.3640403747558594, + "learning_rate": 1e-06, + "loss": 1.1192, + "mean_token_accuracy": 0.665379524230957, + "num_tokens": 49816468.0, + "step": 1949 + }, + { + "epoch": 0.2141445200966396, + "grad_norm": 2.100396156311035, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.7022665143013, + "num_tokens": 49843227.0, + "step": 1950 + }, + { + "epoch": 0.21425433779925324, + "grad_norm": 2.2217931747436523, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6888375282287598, + "num_tokens": 49869761.0, + "step": 1951 + }, + { + "epoch": 0.2143641555018669, + "grad_norm": 2.4060471057891846, + "learning_rate": 1e-06, + "loss": 1.0812, + "mean_token_accuracy": 0.6753180623054504, + "num_tokens": 49894006.0, + "step": 1952 + }, + { + "epoch": 0.21447397320448056, + "grad_norm": 2.1809685230255127, + "learning_rate": 1e-06, + "loss": 1.062, + "mean_token_accuracy": 0.6922162771224976, + "num_tokens": 49918965.0, + "step": 1953 + }, + { + "epoch": 0.21458379090709423, + "grad_norm": 2.612128496170044, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.6997504234313965, + "num_tokens": 49937634.0, + "step": 1954 + }, + { + "epoch": 0.21469360860970788, + "grad_norm": 2.6696388721466064, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.7130120992660522, + "num_tokens": 49957172.0, + "step": 1955 + }, + { + "epoch": 0.21480342631232155, + "grad_norm": 2.283064842224121, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7207237482070923, + "num_tokens": 49982507.0, + "step": 1956 + }, + { + "epoch": 0.2149132440149352, + "grad_norm": 2.1420326232910156, + "learning_rate": 1e-06, + "loss": 1.0744, + "mean_token_accuracy": 0.6774305105209351, + "num_tokens": 50011256.0, + "step": 1957 + }, + { + "epoch": 0.21502306171754887, + "grad_norm": 2.34824538230896, + "learning_rate": 1e-06, + "loss": 1.1207, + "mean_token_accuracy": 0.6650610566139221, + "num_tokens": 50035369.0, + "step": 1958 + }, + { + "epoch": 0.21513287942016254, + "grad_norm": 2.4491055011749268, + "learning_rate": 1e-06, + "loss": 1.0224, + "mean_token_accuracy": 0.7026232481002808, + "num_tokens": 50057966.0, + "step": 1959 + }, + { + "epoch": 0.2152426971227762, + "grad_norm": 2.0786261558532715, + "learning_rate": 1e-06, + "loss": 1.0245, + "mean_token_accuracy": 0.6877356767654419, + "num_tokens": 50086636.0, + "step": 1960 + }, + { + "epoch": 0.21535251482538986, + "grad_norm": 2.2127020359039307, + "learning_rate": 1e-06, + "loss": 1.0711, + "mean_token_accuracy": 0.6749294996261597, + "num_tokens": 50115818.0, + "step": 1961 + }, + { + "epoch": 0.2154623325280035, + "grad_norm": 2.113180160522461, + "learning_rate": 1e-06, + "loss": 1.0756, + "mean_token_accuracy": 0.6785433292388916, + "num_tokens": 50144680.0, + "step": 1962 + }, + { + "epoch": 0.21557215023061718, + "grad_norm": 2.5324811935424805, + "learning_rate": 1e-06, + "loss": 1.0595, + "mean_token_accuracy": 0.6828330755233765, + "num_tokens": 50167673.0, + "step": 1963 + }, + { + "epoch": 0.21568196793323083, + "grad_norm": 2.016183376312256, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7080689072608948, + "num_tokens": 50195321.0, + "step": 1964 + }, + { + "epoch": 0.2157917856358445, + "grad_norm": 2.1114563941955566, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.6936190128326416, + "num_tokens": 50223635.0, + "step": 1965 + }, + { + "epoch": 0.21590160333845815, + "grad_norm": 2.1020009517669678, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.6924461722373962, + "num_tokens": 50251021.0, + "step": 1966 + }, + { + "epoch": 0.21601142104107182, + "grad_norm": 2.3164734840393066, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7028310298919678, + "num_tokens": 50274287.0, + "step": 1967 + }, + { + "epoch": 0.2161212387436855, + "grad_norm": 2.476219415664673, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.704311728477478, + "num_tokens": 50295530.0, + "step": 1968 + }, + { + "epoch": 0.21623105644629914, + "grad_norm": 2.227315902709961, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6974782347679138, + "num_tokens": 50321981.0, + "step": 1969 + }, + { + "epoch": 0.21634087414891282, + "grad_norm": 2.3460853099823, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.6849560141563416, + "num_tokens": 50345638.0, + "step": 1970 + }, + { + "epoch": 0.21645069185152646, + "grad_norm": 2.3959906101226807, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.695178210735321, + "num_tokens": 50368192.0, + "step": 1971 + }, + { + "epoch": 0.21656050955414013, + "grad_norm": 2.193660259246826, + "learning_rate": 1e-06, + "loss": 1.0347, + "mean_token_accuracy": 0.6902568936347961, + "num_tokens": 50393911.0, + "step": 1972 + }, + { + "epoch": 0.21667032725675378, + "grad_norm": 2.509861946105957, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6998263001441956, + "num_tokens": 50414914.0, + "step": 1973 + }, + { + "epoch": 0.21678014495936745, + "grad_norm": 2.187065601348877, + "learning_rate": 1e-06, + "loss": 1.069, + "mean_token_accuracy": 0.6759079694747925, + "num_tokens": 50441648.0, + "step": 1974 + }, + { + "epoch": 0.2168899626619811, + "grad_norm": 2.216291904449463, + "learning_rate": 1e-06, + "loss": 1.0531, + "mean_token_accuracy": 0.6884482502937317, + "num_tokens": 50467045.0, + "step": 1975 + }, + { + "epoch": 0.21699978036459477, + "grad_norm": 2.089512348175049, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7147188186645508, + "num_tokens": 50492571.0, + "step": 1976 + }, + { + "epoch": 0.21710959806720842, + "grad_norm": 1.9892034530639648, + "learning_rate": 1e-06, + "loss": 1.0726, + "mean_token_accuracy": 0.6831976175308228, + "num_tokens": 50527019.0, + "step": 1977 + }, + { + "epoch": 0.2172194157698221, + "grad_norm": 2.099569082260132, + "learning_rate": 1e-06, + "loss": 1.0357, + "mean_token_accuracy": 0.688362717628479, + "num_tokens": 50555012.0, + "step": 1978 + }, + { + "epoch": 0.21732923347243577, + "grad_norm": 2.15997314453125, + "learning_rate": 1e-06, + "loss": 1.0705, + "mean_token_accuracy": 0.6786351203918457, + "num_tokens": 50584889.0, + "step": 1979 + }, + { + "epoch": 0.2174390511750494, + "grad_norm": 2.3508245944976807, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7001764178276062, + "num_tokens": 50609100.0, + "step": 1980 + }, + { + "epoch": 0.2175488688776631, + "grad_norm": 2.2376084327697754, + "learning_rate": 1e-06, + "loss": 1.0556, + "mean_token_accuracy": 0.688019871711731, + "num_tokens": 50636536.0, + "step": 1981 + }, + { + "epoch": 0.21765868658027673, + "grad_norm": 2.276437759399414, + "learning_rate": 1e-06, + "loss": 1.0821, + "mean_token_accuracy": 0.6738598346710205, + "num_tokens": 50662465.0, + "step": 1982 + }, + { + "epoch": 0.2177685042828904, + "grad_norm": 2.1189680099487305, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7119998931884766, + "num_tokens": 50694038.0, + "step": 1983 + }, + { + "epoch": 0.21787832198550405, + "grad_norm": 2.497349977493286, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6952686309814453, + "num_tokens": 50716443.0, + "step": 1984 + }, + { + "epoch": 0.21798813968811773, + "grad_norm": 1.9393726587295532, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6907398104667664, + "num_tokens": 50749715.0, + "step": 1985 + }, + { + "epoch": 0.21809795739073137, + "grad_norm": 2.720735788345337, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7130879163742065, + "num_tokens": 50766928.0, + "step": 1986 + }, + { + "epoch": 0.21820777509334505, + "grad_norm": 2.3185179233551025, + "learning_rate": 1e-06, + "loss": 1.0532, + "mean_token_accuracy": 0.6795613765716553, + "num_tokens": 50791742.0, + "step": 1987 + }, + { + "epoch": 0.21831759279595872, + "grad_norm": 2.5189380645751953, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6930254697799683, + "num_tokens": 50815113.0, + "step": 1988 + }, + { + "epoch": 0.21842741049857237, + "grad_norm": 2.342998504638672, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.6927565932273865, + "num_tokens": 50841560.0, + "step": 1989 + }, + { + "epoch": 0.21853722820118604, + "grad_norm": 2.7602930068969727, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7049797177314758, + "num_tokens": 50860772.0, + "step": 1990 + }, + { + "epoch": 0.21864704590379969, + "grad_norm": 2.057929277420044, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6939098834991455, + "num_tokens": 50892096.0, + "step": 1991 + }, + { + "epoch": 0.21875686360641336, + "grad_norm": 2.486650228500366, + "learning_rate": 1e-06, + "loss": 1.0634, + "mean_token_accuracy": 0.6853376626968384, + "num_tokens": 50915607.0, + "step": 1992 + }, + { + "epoch": 0.218866681309027, + "grad_norm": 2.791534423828125, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7145019173622131, + "num_tokens": 50933161.0, + "step": 1993 + }, + { + "epoch": 0.21897649901164068, + "grad_norm": 2.2255330085754395, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7077815532684326, + "num_tokens": 50958303.0, + "step": 1994 + }, + { + "epoch": 0.21908631671425433, + "grad_norm": 2.002261161804199, + "learning_rate": 1e-06, + "loss": 1.0565, + "mean_token_accuracy": 0.6998137831687927, + "num_tokens": 50989588.0, + "step": 1995 + }, + { + "epoch": 0.219196134416868, + "grad_norm": 2.2512378692626953, + "learning_rate": 1e-06, + "loss": 1.0517, + "mean_token_accuracy": 0.6812781095504761, + "num_tokens": 51016401.0, + "step": 1996 + }, + { + "epoch": 0.21930595211948167, + "grad_norm": 2.256937026977539, + "learning_rate": 1e-06, + "loss": 1.0418, + "mean_token_accuracy": 0.6902430653572083, + "num_tokens": 51042620.0, + "step": 1997 + }, + { + "epoch": 0.21941576982209532, + "grad_norm": 2.2625133991241455, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7171866297721863, + "num_tokens": 51068446.0, + "step": 1998 + }, + { + "epoch": 0.219525587524709, + "grad_norm": 2.5280513763427734, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.703180193901062, + "num_tokens": 51090367.0, + "step": 1999 + }, + { + "epoch": 0.21963540522732264, + "grad_norm": 1.9335026741027832, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7274835109710693, + "num_tokens": 51121259.0, + "step": 2000 + }, + { + "epoch": 0.2197452229299363, + "grad_norm": 2.4954845905303955, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.6876058578491211, + "num_tokens": 51143680.0, + "step": 2001 + }, + { + "epoch": 0.21985504063254996, + "grad_norm": 2.4906692504882812, + "learning_rate": 1e-06, + "loss": 1.1094, + "mean_token_accuracy": 0.6779308319091797, + "num_tokens": 51168332.0, + "step": 2002 + }, + { + "epoch": 0.21996485833516363, + "grad_norm": 2.1995887756347656, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.713604211807251, + "num_tokens": 51194712.0, + "step": 2003 + }, + { + "epoch": 0.22007467603777728, + "grad_norm": 2.3390817642211914, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.709116518497467, + "num_tokens": 51217878.0, + "step": 2004 + }, + { + "epoch": 0.22018449374039095, + "grad_norm": 2.087034225463867, + "learning_rate": 1e-06, + "loss": 1.1313, + "mean_token_accuracy": 0.6706610321998596, + "num_tokens": 51250811.0, + "step": 2005 + }, + { + "epoch": 0.22029431144300463, + "grad_norm": 2.6284067630767822, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7167506814002991, + "num_tokens": 51271845.0, + "step": 2006 + }, + { + "epoch": 0.22040412914561827, + "grad_norm": 2.58792781829834, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7048782110214233, + "num_tokens": 51291845.0, + "step": 2007 + }, + { + "epoch": 0.22051394684823195, + "grad_norm": 2.31709361076355, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7023517489433289, + "num_tokens": 51314762.0, + "step": 2008 + }, + { + "epoch": 0.2206237645508456, + "grad_norm": 2.3482882976531982, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7144532203674316, + "num_tokens": 51338483.0, + "step": 2009 + }, + { + "epoch": 0.22073358225345927, + "grad_norm": 2.4961555004119873, + "learning_rate": 1e-06, + "loss": 1.0691, + "mean_token_accuracy": 0.6863775253295898, + "num_tokens": 51360182.0, + "step": 2010 + }, + { + "epoch": 0.2208433999560729, + "grad_norm": 2.393362522125244, + "learning_rate": 1e-06, + "loss": 1.0749, + "mean_token_accuracy": 0.6751556396484375, + "num_tokens": 51385173.0, + "step": 2011 + }, + { + "epoch": 0.22095321765868658, + "grad_norm": 2.23922061920166, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.709606409072876, + "num_tokens": 51411957.0, + "step": 2012 + }, + { + "epoch": 0.22106303536130023, + "grad_norm": 2.5107338428497314, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6979807019233704, + "num_tokens": 51434583.0, + "step": 2013 + }, + { + "epoch": 0.2211728530639139, + "grad_norm": 2.0573723316192627, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.702570915222168, + "num_tokens": 51464425.0, + "step": 2014 + }, + { + "epoch": 0.22128267076652755, + "grad_norm": 1.9856196641921997, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6932947635650635, + "num_tokens": 51498330.0, + "step": 2015 + }, + { + "epoch": 0.22139248846914122, + "grad_norm": 2.2076563835144043, + "learning_rate": 1e-06, + "loss": 1.0597, + "mean_token_accuracy": 0.6898460388183594, + "num_tokens": 51527517.0, + "step": 2016 + }, + { + "epoch": 0.2215023061717549, + "grad_norm": 2.271796941757202, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7161080241203308, + "num_tokens": 51552838.0, + "step": 2017 + }, + { + "epoch": 0.22161212387436854, + "grad_norm": 2.5084338188171387, + "learning_rate": 1e-06, + "loss": 1.0209, + "mean_token_accuracy": 0.6884559392929077, + "num_tokens": 51575709.0, + "step": 2018 + }, + { + "epoch": 0.22172194157698222, + "grad_norm": 2.2782294750213623, + "learning_rate": 1e-06, + "loss": 1.0239, + "mean_token_accuracy": 0.7079895734786987, + "num_tokens": 51601735.0, + "step": 2019 + }, + { + "epoch": 0.22183175927959586, + "grad_norm": 2.2163941860198975, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6958099007606506, + "num_tokens": 51626040.0, + "step": 2020 + }, + { + "epoch": 0.22194157698220954, + "grad_norm": 2.3437182903289795, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.6896656155586243, + "num_tokens": 51650222.0, + "step": 2021 + }, + { + "epoch": 0.22205139468482318, + "grad_norm": 2.477796792984009, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7003207206726074, + "num_tokens": 51672885.0, + "step": 2022 + }, + { + "epoch": 0.22216121238743686, + "grad_norm": 2.3444409370422363, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7167688012123108, + "num_tokens": 51697891.0, + "step": 2023 + }, + { + "epoch": 0.2222710300900505, + "grad_norm": 2.5115766525268555, + "learning_rate": 1e-06, + "loss": 1.0725, + "mean_token_accuracy": 0.6836473941802979, + "num_tokens": 51719948.0, + "step": 2024 + }, + { + "epoch": 0.22238084779266418, + "grad_norm": 2.464266538619995, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6913582682609558, + "num_tokens": 51744700.0, + "step": 2025 + }, + { + "epoch": 0.22249066549527785, + "grad_norm": 2.2292776107788086, + "learning_rate": 1e-06, + "loss": 1.0902, + "mean_token_accuracy": 0.6843171119689941, + "num_tokens": 51771083.0, + "step": 2026 + }, + { + "epoch": 0.2226004831978915, + "grad_norm": 2.4344253540039062, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.7025764584541321, + "num_tokens": 51794110.0, + "step": 2027 + }, + { + "epoch": 0.22271030090050517, + "grad_norm": 2.41605806350708, + "learning_rate": 1e-06, + "loss": 1.0353, + "mean_token_accuracy": 0.68967604637146, + "num_tokens": 51819052.0, + "step": 2028 + }, + { + "epoch": 0.22282011860311882, + "grad_norm": 2.37288761138916, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7081816792488098, + "num_tokens": 51841222.0, + "step": 2029 + }, + { + "epoch": 0.2229299363057325, + "grad_norm": 2.5457053184509277, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7214468121528625, + "num_tokens": 51861368.0, + "step": 2030 + }, + { + "epoch": 0.22303975400834614, + "grad_norm": 2.186593770980835, + "learning_rate": 1e-06, + "loss": 1.1818, + "mean_token_accuracy": 0.6585807800292969, + "num_tokens": 51891917.0, + "step": 2031 + }, + { + "epoch": 0.2231495717109598, + "grad_norm": 2.215005397796631, + "learning_rate": 1e-06, + "loss": 1.0958, + "mean_token_accuracy": 0.6687296032905579, + "num_tokens": 51920935.0, + "step": 2032 + }, + { + "epoch": 0.22325938941357346, + "grad_norm": 2.2410387992858887, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6928939819335938, + "num_tokens": 51946989.0, + "step": 2033 + }, + { + "epoch": 0.22336920711618713, + "grad_norm": 2.1870028972625732, + "learning_rate": 1e-06, + "loss": 1.0321, + "mean_token_accuracy": 0.6975562572479248, + "num_tokens": 51975471.0, + "step": 2034 + }, + { + "epoch": 0.2234790248188008, + "grad_norm": 2.061955213546753, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7073518633842468, + "num_tokens": 52006676.0, + "step": 2035 + }, + { + "epoch": 0.22358884252141445, + "grad_norm": 2.090054512023926, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.7022300362586975, + "num_tokens": 52035085.0, + "step": 2036 + }, + { + "epoch": 0.22369866022402812, + "grad_norm": 2.1362814903259277, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7046578526496887, + "num_tokens": 52062354.0, + "step": 2037 + }, + { + "epoch": 0.22380847792664177, + "grad_norm": 2.126786708831787, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.6946349740028381, + "num_tokens": 52090532.0, + "step": 2038 + }, + { + "epoch": 0.22391829562925544, + "grad_norm": 2.0369701385498047, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7142785787582397, + "num_tokens": 52118610.0, + "step": 2039 + }, + { + "epoch": 0.2240281133318691, + "grad_norm": 2.284008264541626, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7210185527801514, + "num_tokens": 52143180.0, + "step": 2040 + }, + { + "epoch": 0.22413793103448276, + "grad_norm": 2.2956809997558594, + "learning_rate": 1e-06, + "loss": 1.0726, + "mean_token_accuracy": 0.6775778532028198, + "num_tokens": 52166897.0, + "step": 2041 + }, + { + "epoch": 0.2242477487370964, + "grad_norm": 2.0963711738586426, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6989285945892334, + "num_tokens": 52194960.0, + "step": 2042 + }, + { + "epoch": 0.22435756643971008, + "grad_norm": 2.547194004058838, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7277809381484985, + "num_tokens": 52216993.0, + "step": 2043 + }, + { + "epoch": 0.22446738414232376, + "grad_norm": 2.51204776763916, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.7093788385391235, + "num_tokens": 52239636.0, + "step": 2044 + }, + { + "epoch": 0.2245772018449374, + "grad_norm": 2.1503658294677734, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7048693895339966, + "num_tokens": 52266583.0, + "step": 2045 + }, + { + "epoch": 0.22468701954755108, + "grad_norm": 2.534644842147827, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6840623617172241, + "num_tokens": 52288097.0, + "step": 2046 + }, + { + "epoch": 0.22479683725016472, + "grad_norm": 2.222400188446045, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6986130475997925, + "num_tokens": 52312787.0, + "step": 2047 + }, + { + "epoch": 0.2249066549527784, + "grad_norm": 2.377748727798462, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.7068277597427368, + "num_tokens": 52335776.0, + "step": 2048 + }, + { + "epoch": 0.22501647265539204, + "grad_norm": 2.762730836868286, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7101894617080688, + "num_tokens": 52354216.0, + "step": 2049 + }, + { + "epoch": 0.22512629035800572, + "grad_norm": 2.0626490116119385, + "learning_rate": 1e-06, + "loss": 1.0895, + "mean_token_accuracy": 0.6775778532028198, + "num_tokens": 52385568.0, + "step": 2050 + }, + { + "epoch": 0.22523610806061936, + "grad_norm": 2.306175470352173, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6912713646888733, + "num_tokens": 52408780.0, + "step": 2051 + }, + { + "epoch": 0.22534592576323303, + "grad_norm": 2.0755207538604736, + "learning_rate": 1e-06, + "loss": 1.0631, + "mean_token_accuracy": 0.6861204504966736, + "num_tokens": 52436905.0, + "step": 2052 + }, + { + "epoch": 0.22545574346584668, + "grad_norm": 2.5676894187927246, + "learning_rate": 1e-06, + "loss": 1.045, + "mean_token_accuracy": 0.6834136247634888, + "num_tokens": 52457574.0, + "step": 2053 + }, + { + "epoch": 0.22556556116846035, + "grad_norm": 2.6228253841400146, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.7015937566757202, + "num_tokens": 52477862.0, + "step": 2054 + }, + { + "epoch": 0.22567537887107403, + "grad_norm": 1.979398250579834, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7244019508361816, + "num_tokens": 52507018.0, + "step": 2055 + }, + { + "epoch": 0.22578519657368767, + "grad_norm": 2.3138229846954346, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.6774870753288269, + "num_tokens": 52532978.0, + "step": 2056 + }, + { + "epoch": 0.22589501427630135, + "grad_norm": 2.3999767303466797, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6993035078048706, + "num_tokens": 52554380.0, + "step": 2057 + }, + { + "epoch": 0.226004831978915, + "grad_norm": 2.545773983001709, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.6967002153396606, + "num_tokens": 52576442.0, + "step": 2058 + }, + { + "epoch": 0.22611464968152867, + "grad_norm": 2.0978386402130127, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6936999559402466, + "num_tokens": 52607411.0, + "step": 2059 + }, + { + "epoch": 0.2262244673841423, + "grad_norm": 2.7044360637664795, + "learning_rate": 1e-06, + "loss": 1.0459, + "mean_token_accuracy": 0.688835084438324, + "num_tokens": 52628569.0, + "step": 2060 + }, + { + "epoch": 0.226334285086756, + "grad_norm": 2.534059524536133, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.7111531496047974, + "num_tokens": 52649432.0, + "step": 2061 + }, + { + "epoch": 0.22644410278936963, + "grad_norm": 2.1406185626983643, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6900594234466553, + "num_tokens": 52679537.0, + "step": 2062 + }, + { + "epoch": 0.2265539204919833, + "grad_norm": 2.3030807971954346, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6937736868858337, + "num_tokens": 52704530.0, + "step": 2063 + }, + { + "epoch": 0.22666373819459698, + "grad_norm": 2.134268045425415, + "learning_rate": 1e-06, + "loss": 1.0303, + "mean_token_accuracy": 0.6884203553199768, + "num_tokens": 52732715.0, + "step": 2064 + }, + { + "epoch": 0.22677355589721063, + "grad_norm": 2.3445467948913574, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6890325546264648, + "num_tokens": 52757788.0, + "step": 2065 + }, + { + "epoch": 0.2268833735998243, + "grad_norm": 2.1894140243530273, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7103773355484009, + "num_tokens": 52785080.0, + "step": 2066 + }, + { + "epoch": 0.22699319130243795, + "grad_norm": 2.5825374126434326, + "learning_rate": 1e-06, + "loss": 1.0705, + "mean_token_accuracy": 0.6810725331306458, + "num_tokens": 52808388.0, + "step": 2067 + }, + { + "epoch": 0.22710300900505162, + "grad_norm": 2.499274969100952, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6971656084060669, + "num_tokens": 52832364.0, + "step": 2068 + }, + { + "epoch": 0.22721282670766527, + "grad_norm": 2.123314380645752, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6882812976837158, + "num_tokens": 52861182.0, + "step": 2069 + }, + { + "epoch": 0.22732264441027894, + "grad_norm": 2.288208484649658, + "learning_rate": 1e-06, + "loss": 1.0334, + "mean_token_accuracy": 0.6909385323524475, + "num_tokens": 52887106.0, + "step": 2070 + }, + { + "epoch": 0.22743246211289259, + "grad_norm": 2.4979052543640137, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.6981197595596313, + "num_tokens": 52906021.0, + "step": 2071 + }, + { + "epoch": 0.22754227981550626, + "grad_norm": 2.2926506996154785, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.6998646259307861, + "num_tokens": 52930228.0, + "step": 2072 + }, + { + "epoch": 0.22765209751811993, + "grad_norm": 2.307002305984497, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7205971479415894, + "num_tokens": 52952935.0, + "step": 2073 + }, + { + "epoch": 0.22776191522073358, + "grad_norm": 2.183159828186035, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6951915621757507, + "num_tokens": 52979744.0, + "step": 2074 + }, + { + "epoch": 0.22787173292334725, + "grad_norm": 2.146047353744507, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7128599286079407, + "num_tokens": 53010075.0, + "step": 2075 + }, + { + "epoch": 0.2279815506259609, + "grad_norm": 2.0765769481658936, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.6941786408424377, + "num_tokens": 53040487.0, + "step": 2076 + }, + { + "epoch": 0.22809136832857457, + "grad_norm": 2.2252655029296875, + "learning_rate": 1e-06, + "loss": 1.0575, + "mean_token_accuracy": 0.6836350560188293, + "num_tokens": 53069110.0, + "step": 2077 + }, + { + "epoch": 0.22820118603118822, + "grad_norm": 2.940404176712036, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7193260192871094, + "num_tokens": 53086481.0, + "step": 2078 + }, + { + "epoch": 0.2283110037338019, + "grad_norm": 2.5543246269226074, + "learning_rate": 1e-06, + "loss": 1.0235, + "mean_token_accuracy": 0.687081515789032, + "num_tokens": 53107681.0, + "step": 2079 + }, + { + "epoch": 0.22842082143641554, + "grad_norm": 2.4428696632385254, + "learning_rate": 1e-06, + "loss": 1.0388, + "mean_token_accuracy": 0.685219407081604, + "num_tokens": 53130222.0, + "step": 2080 + }, + { + "epoch": 0.2285306391390292, + "grad_norm": 2.3362491130828857, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.698672354221344, + "num_tokens": 53153636.0, + "step": 2081 + }, + { + "epoch": 0.2286404568416429, + "grad_norm": 1.9940959215164185, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6983926296234131, + "num_tokens": 53185884.0, + "step": 2082 + }, + { + "epoch": 0.22875027454425653, + "grad_norm": 2.376664400100708, + "learning_rate": 1e-06, + "loss": 1.0307, + "mean_token_accuracy": 0.698960542678833, + "num_tokens": 53208352.0, + "step": 2083 + }, + { + "epoch": 0.2288600922468702, + "grad_norm": 2.3914432525634766, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.6961572766304016, + "num_tokens": 53230153.0, + "step": 2084 + }, + { + "epoch": 0.22896990994948385, + "grad_norm": 2.429004669189453, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7234215140342712, + "num_tokens": 53252569.0, + "step": 2085 + }, + { + "epoch": 0.22907972765209753, + "grad_norm": 2.4058916568756104, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7051661014556885, + "num_tokens": 53273330.0, + "step": 2086 + }, + { + "epoch": 0.22918954535471117, + "grad_norm": 2.473168134689331, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7129473090171814, + "num_tokens": 53295183.0, + "step": 2087 + }, + { + "epoch": 0.22929936305732485, + "grad_norm": 2.3988823890686035, + "learning_rate": 1e-06, + "loss": 1.0868, + "mean_token_accuracy": 0.6865665912628174, + "num_tokens": 53319125.0, + "step": 2088 + }, + { + "epoch": 0.2294091807599385, + "grad_norm": 2.4822981357574463, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.7038345336914062, + "num_tokens": 53341952.0, + "step": 2089 + }, + { + "epoch": 0.22951899846255217, + "grad_norm": 1.9126379489898682, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6922447681427002, + "num_tokens": 53376467.0, + "step": 2090 + }, + { + "epoch": 0.2296288161651658, + "grad_norm": 2.434422731399536, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.7046487331390381, + "num_tokens": 53400079.0, + "step": 2091 + }, + { + "epoch": 0.22973863386777948, + "grad_norm": 2.2132558822631836, + "learning_rate": 1e-06, + "loss": 1.0486, + "mean_token_accuracy": 0.6785005331039429, + "num_tokens": 53428702.0, + "step": 2092 + }, + { + "epoch": 0.22984845157039316, + "grad_norm": 1.9757230281829834, + "learning_rate": 1e-06, + "loss": 1.0983, + "mean_token_accuracy": 0.6742502450942993, + "num_tokens": 53459838.0, + "step": 2093 + }, + { + "epoch": 0.2299582692730068, + "grad_norm": 2.415780782699585, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7261691093444824, + "num_tokens": 53481126.0, + "step": 2094 + }, + { + "epoch": 0.23006808697562048, + "grad_norm": 2.268782377243042, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.7034447193145752, + "num_tokens": 53507707.0, + "step": 2095 + }, + { + "epoch": 0.23017790467823412, + "grad_norm": 2.7336504459381104, + "learning_rate": 1e-06, + "loss": 1.0293, + "mean_token_accuracy": 0.6928160190582275, + "num_tokens": 53526887.0, + "step": 2096 + }, + { + "epoch": 0.2302877223808478, + "grad_norm": 1.9932918548583984, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7352814674377441, + "num_tokens": 53554968.0, + "step": 2097 + }, + { + "epoch": 0.23039754008346144, + "grad_norm": 2.7576839923858643, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7153180837631226, + "num_tokens": 53572880.0, + "step": 2098 + }, + { + "epoch": 0.23050735778607512, + "grad_norm": 2.025472402572632, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7024489641189575, + "num_tokens": 53601574.0, + "step": 2099 + }, + { + "epoch": 0.23061717548868876, + "grad_norm": 2.144376516342163, + "learning_rate": 1e-06, + "loss": 1.0579, + "mean_token_accuracy": 0.6937100887298584, + "num_tokens": 53631998.0, + "step": 2100 + }, + { + "epoch": 0.23072699319130244, + "grad_norm": 2.4197027683258057, + "learning_rate": 1e-06, + "loss": 1.0239, + "mean_token_accuracy": 0.6889132857322693, + "num_tokens": 53656875.0, + "step": 2101 + }, + { + "epoch": 0.2308368108939161, + "grad_norm": 2.1196072101593018, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6955048441886902, + "num_tokens": 53686024.0, + "step": 2102 + }, + { + "epoch": 0.23094662859652976, + "grad_norm": 2.6857197284698486, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.6913772821426392, + "num_tokens": 53706666.0, + "step": 2103 + }, + { + "epoch": 0.23105644629914343, + "grad_norm": 2.203479766845703, + "learning_rate": 1e-06, + "loss": 1.0596, + "mean_token_accuracy": 0.6881370544433594, + "num_tokens": 53732992.0, + "step": 2104 + }, + { + "epoch": 0.23116626400175708, + "grad_norm": 2.483558177947998, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6979705095291138, + "num_tokens": 53753845.0, + "step": 2105 + }, + { + "epoch": 0.23127608170437075, + "grad_norm": 2.0895023345947266, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.6941996812820435, + "num_tokens": 53780935.0, + "step": 2106 + }, + { + "epoch": 0.2313858994069844, + "grad_norm": 2.2555758953094482, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.6914482116699219, + "num_tokens": 53808592.0, + "step": 2107 + }, + { + "epoch": 0.23149571710959807, + "grad_norm": 2.144091844558716, + "learning_rate": 1e-06, + "loss": 1.0672, + "mean_token_accuracy": 0.6783584952354431, + "num_tokens": 53839648.0, + "step": 2108 + }, + { + "epoch": 0.23160553481221172, + "grad_norm": 2.503445863723755, + "learning_rate": 1e-06, + "loss": 1.043, + "mean_token_accuracy": 0.6885764002799988, + "num_tokens": 53859982.0, + "step": 2109 + }, + { + "epoch": 0.2317153525148254, + "grad_norm": 2.263679027557373, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7107335925102234, + "num_tokens": 53883053.0, + "step": 2110 + }, + { + "epoch": 0.23182517021743906, + "grad_norm": 2.5295820236206055, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.7017840147018433, + "num_tokens": 53904026.0, + "step": 2111 + }, + { + "epoch": 0.2319349879200527, + "grad_norm": 2.4582631587982178, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.714657187461853, + "num_tokens": 53927818.0, + "step": 2112 + }, + { + "epoch": 0.23204480562266638, + "grad_norm": 2.427264451980591, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7027292251586914, + "num_tokens": 53950520.0, + "step": 2113 + }, + { + "epoch": 0.23215462332528003, + "grad_norm": 2.5219109058380127, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.6934407949447632, + "num_tokens": 53972310.0, + "step": 2114 + }, + { + "epoch": 0.2322644410278937, + "grad_norm": 2.335491895675659, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.699783205986023, + "num_tokens": 53997622.0, + "step": 2115 + }, + { + "epoch": 0.23237425873050735, + "grad_norm": 2.226397752761841, + "learning_rate": 1e-06, + "loss": 1.0717, + "mean_token_accuracy": 0.6855508089065552, + "num_tokens": 54024991.0, + "step": 2116 + }, + { + "epoch": 0.23248407643312102, + "grad_norm": 2.121243953704834, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.709330141544342, + "num_tokens": 54053482.0, + "step": 2117 + }, + { + "epoch": 0.23259389413573467, + "grad_norm": 2.451625347137451, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.691875696182251, + "num_tokens": 54078638.0, + "step": 2118 + }, + { + "epoch": 0.23270371183834834, + "grad_norm": 2.325317859649658, + "learning_rate": 1e-06, + "loss": 1.0683, + "mean_token_accuracy": 0.6823628544807434, + "num_tokens": 54105231.0, + "step": 2119 + }, + { + "epoch": 0.23281352954096202, + "grad_norm": 2.484441041946411, + "learning_rate": 1e-06, + "loss": 1.0719, + "mean_token_accuracy": 0.6785069704055786, + "num_tokens": 54127800.0, + "step": 2120 + }, + { + "epoch": 0.23292334724357566, + "grad_norm": 2.4723613262176514, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7291486263275146, + "num_tokens": 54149267.0, + "step": 2121 + }, + { + "epoch": 0.23303316494618934, + "grad_norm": 2.3440120220184326, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.7001615762710571, + "num_tokens": 54173454.0, + "step": 2122 + }, + { + "epoch": 0.23314298264880298, + "grad_norm": 2.4708411693573, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.695961594581604, + "num_tokens": 54197638.0, + "step": 2123 + }, + { + "epoch": 0.23325280035141666, + "grad_norm": 2.5549163818359375, + "learning_rate": 1e-06, + "loss": 1.0423, + "mean_token_accuracy": 0.6919109225273132, + "num_tokens": 54217332.0, + "step": 2124 + }, + { + "epoch": 0.2333626180540303, + "grad_norm": 2.3821542263031006, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6948298215866089, + "num_tokens": 54242067.0, + "step": 2125 + }, + { + "epoch": 0.23347243575664398, + "grad_norm": 2.2068705558776855, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.6998459100723267, + "num_tokens": 54268651.0, + "step": 2126 + }, + { + "epoch": 0.23358225345925762, + "grad_norm": 2.4893269538879395, + "learning_rate": 1e-06, + "loss": 1.0196, + "mean_token_accuracy": 0.6925676465034485, + "num_tokens": 54290036.0, + "step": 2127 + }, + { + "epoch": 0.2336920711618713, + "grad_norm": 2.119856357574463, + "learning_rate": 1e-06, + "loss": 1.0355, + "mean_token_accuracy": 0.6902309656143188, + "num_tokens": 54320118.0, + "step": 2128 + }, + { + "epoch": 0.23380188886448494, + "grad_norm": 2.0047078132629395, + "learning_rate": 1e-06, + "loss": 1.134, + "mean_token_accuracy": 0.6618880033493042, + "num_tokens": 54354377.0, + "step": 2129 + }, + { + "epoch": 0.23391170656709862, + "grad_norm": 2.0438642501831055, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7000687122344971, + "num_tokens": 54386638.0, + "step": 2130 + }, + { + "epoch": 0.2340215242697123, + "grad_norm": 2.332770824432373, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7084639072418213, + "num_tokens": 54409312.0, + "step": 2131 + }, + { + "epoch": 0.23413134197232593, + "grad_norm": 2.2977144718170166, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6958604454994202, + "num_tokens": 54434429.0, + "step": 2132 + }, + { + "epoch": 0.2342411596749396, + "grad_norm": 2.2457993030548096, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.7041929960250854, + "num_tokens": 54460175.0, + "step": 2133 + }, + { + "epoch": 0.23435097737755325, + "grad_norm": 2.2928295135498047, + "learning_rate": 1e-06, + "loss": 1.0306, + "mean_token_accuracy": 0.6876436471939087, + "num_tokens": 54487512.0, + "step": 2134 + }, + { + "epoch": 0.23446079508016693, + "grad_norm": 2.6361122131347656, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.7003072500228882, + "num_tokens": 54510114.0, + "step": 2135 + }, + { + "epoch": 0.23457061278278057, + "grad_norm": 2.819324016571045, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.706379234790802, + "num_tokens": 54526467.0, + "step": 2136 + }, + { + "epoch": 0.23468043048539425, + "grad_norm": 2.2012877464294434, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.6872155070304871, + "num_tokens": 54552705.0, + "step": 2137 + }, + { + "epoch": 0.2347902481880079, + "grad_norm": 2.479492664337158, + "learning_rate": 1e-06, + "loss": 1.075, + "mean_token_accuracy": 0.6785778999328613, + "num_tokens": 54576659.0, + "step": 2138 + }, + { + "epoch": 0.23490006589062157, + "grad_norm": 2.2930893898010254, + "learning_rate": 1e-06, + "loss": 1.0922, + "mean_token_accuracy": 0.6729604005813599, + "num_tokens": 54603737.0, + "step": 2139 + }, + { + "epoch": 0.23500988359323524, + "grad_norm": 2.412149429321289, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7226758003234863, + "num_tokens": 54624442.0, + "step": 2140 + }, + { + "epoch": 0.2351197012958489, + "grad_norm": 2.0718231201171875, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6994544267654419, + "num_tokens": 54654754.0, + "step": 2141 + }, + { + "epoch": 0.23522951899846256, + "grad_norm": 2.246873378753662, + "learning_rate": 1e-06, + "loss": 1.0701, + "mean_token_accuracy": 0.6808125972747803, + "num_tokens": 54681455.0, + "step": 2142 + }, + { + "epoch": 0.2353393367010762, + "grad_norm": 1.9419769048690796, + "learning_rate": 1e-06, + "loss": 1.1, + "mean_token_accuracy": 0.6782257556915283, + "num_tokens": 54715894.0, + "step": 2143 + }, + { + "epoch": 0.23544915440368988, + "grad_norm": 2.027878761291504, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7201208472251892, + "num_tokens": 54745627.0, + "step": 2144 + }, + { + "epoch": 0.23555897210630353, + "grad_norm": 2.245532274246216, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6900110244750977, + "num_tokens": 54771315.0, + "step": 2145 + }, + { + "epoch": 0.2356687898089172, + "grad_norm": 2.332223653793335, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.7026910781860352, + "num_tokens": 54796481.0, + "step": 2146 + }, + { + "epoch": 0.23577860751153085, + "grad_norm": 2.2816543579101562, + "learning_rate": 1e-06, + "loss": 1.1378, + "mean_token_accuracy": 0.6657535433769226, + "num_tokens": 54822941.0, + "step": 2147 + }, + { + "epoch": 0.23588842521414452, + "grad_norm": 1.9333686828613281, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7103592157363892, + "num_tokens": 54854853.0, + "step": 2148 + }, + { + "epoch": 0.2359982429167582, + "grad_norm": 2.513094902038574, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7210102081298828, + "num_tokens": 54874689.0, + "step": 2149 + }, + { + "epoch": 0.23610806061937184, + "grad_norm": 2.3025331497192383, + "learning_rate": 1e-06, + "loss": 1.0644, + "mean_token_accuracy": 0.6935861706733704, + "num_tokens": 54898900.0, + "step": 2150 + }, + { + "epoch": 0.2362178783219855, + "grad_norm": 2.262421131134033, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7002440690994263, + "num_tokens": 54924673.0, + "step": 2151 + }, + { + "epoch": 0.23632769602459916, + "grad_norm": 2.2019643783569336, + "learning_rate": 1e-06, + "loss": 1.0973, + "mean_token_accuracy": 0.681674599647522, + "num_tokens": 54954453.0, + "step": 2152 + }, + { + "epoch": 0.23643751372721283, + "grad_norm": 2.283379077911377, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7180832028388977, + "num_tokens": 54980778.0, + "step": 2153 + }, + { + "epoch": 0.23654733142982648, + "grad_norm": 2.295869827270508, + "learning_rate": 1e-06, + "loss": 1.101, + "mean_token_accuracy": 0.6743607521057129, + "num_tokens": 55006536.0, + "step": 2154 + }, + { + "epoch": 0.23665714913244015, + "grad_norm": 2.0049378871917725, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7217074632644653, + "num_tokens": 55037000.0, + "step": 2155 + }, + { + "epoch": 0.2367669668350538, + "grad_norm": 2.37386417388916, + "learning_rate": 1e-06, + "loss": 1.0791, + "mean_token_accuracy": 0.6816424131393433, + "num_tokens": 55062442.0, + "step": 2156 + }, + { + "epoch": 0.23687678453766747, + "grad_norm": 2.661403179168701, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7127342224121094, + "num_tokens": 55082166.0, + "step": 2157 + }, + { + "epoch": 0.23698660224028115, + "grad_norm": 2.2499947547912598, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.6988745927810669, + "num_tokens": 55107444.0, + "step": 2158 + }, + { + "epoch": 0.2370964199428948, + "grad_norm": 2.3257617950439453, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6999521851539612, + "num_tokens": 55131280.0, + "step": 2159 + }, + { + "epoch": 0.23720623764550847, + "grad_norm": 2.449629068374634, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7036058902740479, + "num_tokens": 55155076.0, + "step": 2160 + }, + { + "epoch": 0.2373160553481221, + "grad_norm": 2.463277578353882, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7085884213447571, + "num_tokens": 55176624.0, + "step": 2161 + }, + { + "epoch": 0.2374258730507358, + "grad_norm": 2.1881871223449707, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.7068885564804077, + "num_tokens": 55203481.0, + "step": 2162 + }, + { + "epoch": 0.23753569075334943, + "grad_norm": 2.4970016479492188, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7047770619392395, + "num_tokens": 55225509.0, + "step": 2163 + }, + { + "epoch": 0.2376455084559631, + "grad_norm": 2.4091756343841553, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6951253414154053, + "num_tokens": 55248517.0, + "step": 2164 + }, + { + "epoch": 0.23775532615857675, + "grad_norm": 2.1343865394592285, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7070342898368835, + "num_tokens": 55275595.0, + "step": 2165 + }, + { + "epoch": 0.23786514386119043, + "grad_norm": 2.2258944511413574, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6954175233840942, + "num_tokens": 55301355.0, + "step": 2166 + }, + { + "epoch": 0.23797496156380407, + "grad_norm": 2.2071526050567627, + "learning_rate": 1e-06, + "loss": 1.048, + "mean_token_accuracy": 0.6867226362228394, + "num_tokens": 55329195.0, + "step": 2167 + }, + { + "epoch": 0.23808477926641775, + "grad_norm": 2.2820026874542236, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7090831398963928, + "num_tokens": 55353890.0, + "step": 2168 + }, + { + "epoch": 0.23819459696903142, + "grad_norm": 2.2142839431762695, + "learning_rate": 1e-06, + "loss": 1.0463, + "mean_token_accuracy": 0.6862945556640625, + "num_tokens": 55379649.0, + "step": 2169 + }, + { + "epoch": 0.23830441467164507, + "grad_norm": 2.3492634296417236, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7087602615356445, + "num_tokens": 55404854.0, + "step": 2170 + }, + { + "epoch": 0.23841423237425874, + "grad_norm": 2.3649096488952637, + "learning_rate": 1e-06, + "loss": 1.0353, + "mean_token_accuracy": 0.6930969953536987, + "num_tokens": 55430663.0, + "step": 2171 + }, + { + "epoch": 0.23852405007687238, + "grad_norm": 2.406378984451294, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7115584015846252, + "num_tokens": 55452698.0, + "step": 2172 + }, + { + "epoch": 0.23863386777948606, + "grad_norm": 2.1459085941314697, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6920902729034424, + "num_tokens": 55480422.0, + "step": 2173 + }, + { + "epoch": 0.2387436854820997, + "grad_norm": 2.0008983612060547, + "learning_rate": 1e-06, + "loss": 1.0787, + "mean_token_accuracy": 0.6794865131378174, + "num_tokens": 55514144.0, + "step": 2174 + }, + { + "epoch": 0.23885350318471338, + "grad_norm": 2.2108688354492188, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.7005900740623474, + "num_tokens": 55541192.0, + "step": 2175 + }, + { + "epoch": 0.23896332088732702, + "grad_norm": 2.043182849884033, + "learning_rate": 1e-06, + "loss": 1.0509, + "mean_token_accuracy": 0.6801480054855347, + "num_tokens": 55574250.0, + "step": 2176 + }, + { + "epoch": 0.2390731385899407, + "grad_norm": 2.9024147987365723, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7016929984092712, + "num_tokens": 55589970.0, + "step": 2177 + }, + { + "epoch": 0.23918295629255437, + "grad_norm": 2.491602897644043, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6983867287635803, + "num_tokens": 55611661.0, + "step": 2178 + }, + { + "epoch": 0.23929277399516802, + "grad_norm": 2.01562237739563, + "learning_rate": 1e-06, + "loss": 1.0588, + "mean_token_accuracy": 0.6790683269500732, + "num_tokens": 55642688.0, + "step": 2179 + }, + { + "epoch": 0.2394025916977817, + "grad_norm": 2.315993309020996, + "learning_rate": 1e-06, + "loss": 1.0329, + "mean_token_accuracy": 0.6919920444488525, + "num_tokens": 55668164.0, + "step": 2180 + }, + { + "epoch": 0.23951240940039534, + "grad_norm": 2.3519301414489746, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6988693475723267, + "num_tokens": 55692450.0, + "step": 2181 + }, + { + "epoch": 0.239622227103009, + "grad_norm": 2.302129030227661, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6996734142303467, + "num_tokens": 55717050.0, + "step": 2182 + }, + { + "epoch": 0.23973204480562266, + "grad_norm": 2.2902798652648926, + "learning_rate": 1e-06, + "loss": 1.0474, + "mean_token_accuracy": 0.6843060255050659, + "num_tokens": 55743357.0, + "step": 2183 + }, + { + "epoch": 0.23984186250823633, + "grad_norm": 2.1723101139068604, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.6981917023658752, + "num_tokens": 55768564.0, + "step": 2184 + }, + { + "epoch": 0.23995168021084998, + "grad_norm": 2.1981146335601807, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.7042827606201172, + "num_tokens": 55794072.0, + "step": 2185 + }, + { + "epoch": 0.24006149791346365, + "grad_norm": 2.396660566329956, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7116763591766357, + "num_tokens": 55817396.0, + "step": 2186 + }, + { + "epoch": 0.24017131561607732, + "grad_norm": 2.4428160190582275, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6923628449440002, + "num_tokens": 55839441.0, + "step": 2187 + }, + { + "epoch": 0.24028113331869097, + "grad_norm": 1.893621802330017, + "learning_rate": 1e-06, + "loss": 1.0209, + "mean_token_accuracy": 0.6900839805603027, + "num_tokens": 55874537.0, + "step": 2188 + }, + { + "epoch": 0.24039095102130464, + "grad_norm": 2.326892614364624, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.7152804732322693, + "num_tokens": 55900084.0, + "step": 2189 + }, + { + "epoch": 0.2405007687239183, + "grad_norm": 2.3135104179382324, + "learning_rate": 1e-06, + "loss": 1.0521, + "mean_token_accuracy": 0.6900067329406738, + "num_tokens": 55925501.0, + "step": 2190 + }, + { + "epoch": 0.24061058642653196, + "grad_norm": 2.1828784942626953, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7049898505210876, + "num_tokens": 55953029.0, + "step": 2191 + }, + { + "epoch": 0.2407204041291456, + "grad_norm": 2.463747262954712, + "learning_rate": 1e-06, + "loss": 1.0801, + "mean_token_accuracy": 0.6770645380020142, + "num_tokens": 55975662.0, + "step": 2192 + }, + { + "epoch": 0.24083022183175928, + "grad_norm": 2.2731311321258545, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7119332551956177, + "num_tokens": 55999918.0, + "step": 2193 + }, + { + "epoch": 0.24094003953437293, + "grad_norm": 2.4241864681243896, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.7078741192817688, + "num_tokens": 56024817.0, + "step": 2194 + }, + { + "epoch": 0.2410498572369866, + "grad_norm": 2.4143788814544678, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.7060084342956543, + "num_tokens": 56047805.0, + "step": 2195 + }, + { + "epoch": 0.24115967493960028, + "grad_norm": 2.2040112018585205, + "learning_rate": 1e-06, + "loss": 1.0769, + "mean_token_accuracy": 0.6791459321975708, + "num_tokens": 56076575.0, + "step": 2196 + }, + { + "epoch": 0.24126949264221392, + "grad_norm": 2.0077624320983887, + "learning_rate": 1e-06, + "loss": 1.0774, + "mean_token_accuracy": 0.6918196082115173, + "num_tokens": 56108038.0, + "step": 2197 + }, + { + "epoch": 0.2413793103448276, + "grad_norm": 2.19718861579895, + "learning_rate": 1e-06, + "loss": 1.0358, + "mean_token_accuracy": 0.6893712282180786, + "num_tokens": 56135124.0, + "step": 2198 + }, + { + "epoch": 0.24148912804744124, + "grad_norm": 2.2391483783721924, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.7015910148620605, + "num_tokens": 56160323.0, + "step": 2199 + }, + { + "epoch": 0.24159894575005492, + "grad_norm": 2.60295033454895, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6904940009117126, + "num_tokens": 56182766.0, + "step": 2200 + }, + { + "epoch": 0.24170876345266856, + "grad_norm": 2.207622528076172, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7277724146842957, + "num_tokens": 56207472.0, + "step": 2201 + }, + { + "epoch": 0.24181858115528224, + "grad_norm": 2.053313732147217, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6840699911117554, + "num_tokens": 56236624.0, + "step": 2202 + }, + { + "epoch": 0.24192839885789588, + "grad_norm": 2.0830655097961426, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.6851476430892944, + "num_tokens": 56265075.0, + "step": 2203 + }, + { + "epoch": 0.24203821656050956, + "grad_norm": 2.109342575073242, + "learning_rate": 1e-06, + "loss": 1.0612, + "mean_token_accuracy": 0.6771094799041748, + "num_tokens": 56294432.0, + "step": 2204 + }, + { + "epoch": 0.2421480342631232, + "grad_norm": 2.2960662841796875, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6946235299110413, + "num_tokens": 56320016.0, + "step": 2205 + }, + { + "epoch": 0.24225785196573688, + "grad_norm": 2.758915662765503, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6931107044219971, + "num_tokens": 56341272.0, + "step": 2206 + }, + { + "epoch": 0.24236766966835055, + "grad_norm": 2.3277580738067627, + "learning_rate": 1e-06, + "loss": 1.0876, + "mean_token_accuracy": 0.6719439029693604, + "num_tokens": 56366389.0, + "step": 2207 + }, + { + "epoch": 0.2424774873709642, + "grad_norm": 2.1031763553619385, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7154404520988464, + "num_tokens": 56395115.0, + "step": 2208 + }, + { + "epoch": 0.24258730507357787, + "grad_norm": 2.051419973373413, + "learning_rate": 1e-06, + "loss": 1.0868, + "mean_token_accuracy": 0.6799392104148865, + "num_tokens": 56427663.0, + "step": 2209 + }, + { + "epoch": 0.24269712277619152, + "grad_norm": 2.5331878662109375, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.6936854124069214, + "num_tokens": 56449665.0, + "step": 2210 + }, + { + "epoch": 0.2428069404788052, + "grad_norm": 2.206663131713867, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.7060869932174683, + "num_tokens": 56475883.0, + "step": 2211 + }, + { + "epoch": 0.24291675818141883, + "grad_norm": 2.2117161750793457, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7071520090103149, + "num_tokens": 56501743.0, + "step": 2212 + }, + { + "epoch": 0.2430265758840325, + "grad_norm": 2.283670663833618, + "learning_rate": 1e-06, + "loss": 1.0477, + "mean_token_accuracy": 0.688404381275177, + "num_tokens": 56526560.0, + "step": 2213 + }, + { + "epoch": 0.24313639358664615, + "grad_norm": 2.0868499279022217, + "learning_rate": 1e-06, + "loss": 1.1057, + "mean_token_accuracy": 0.6706717610359192, + "num_tokens": 56556583.0, + "step": 2214 + }, + { + "epoch": 0.24324621128925983, + "grad_norm": 2.1040494441986084, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7033225297927856, + "num_tokens": 56585009.0, + "step": 2215 + }, + { + "epoch": 0.2433560289918735, + "grad_norm": 2.13908052444458, + "learning_rate": 1e-06, + "loss": 1.1113, + "mean_token_accuracy": 0.6796170473098755, + "num_tokens": 56614687.0, + "step": 2216 + }, + { + "epoch": 0.24346584669448715, + "grad_norm": 2.1964545249938965, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7074702382087708, + "num_tokens": 56641631.0, + "step": 2217 + }, + { + "epoch": 0.24357566439710082, + "grad_norm": 2.3952887058258057, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6924262642860413, + "num_tokens": 56665854.0, + "step": 2218 + }, + { + "epoch": 0.24368548209971447, + "grad_norm": 2.5716392993927, + "learning_rate": 1e-06, + "loss": 1.0697, + "mean_token_accuracy": 0.6751466989517212, + "num_tokens": 56687935.0, + "step": 2219 + }, + { + "epoch": 0.24379529980232814, + "grad_norm": 2.1218888759613037, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6902963519096375, + "num_tokens": 56718433.0, + "step": 2220 + }, + { + "epoch": 0.2439051175049418, + "grad_norm": 2.409475088119507, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7138561010360718, + "num_tokens": 56740775.0, + "step": 2221 + }, + { + "epoch": 0.24401493520755546, + "grad_norm": 2.0298168659210205, + "learning_rate": 1e-06, + "loss": 1.0397, + "mean_token_accuracy": 0.6862417459487915, + "num_tokens": 56770766.0, + "step": 2222 + }, + { + "epoch": 0.2441247529101691, + "grad_norm": 2.30653715133667, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7107431888580322, + "num_tokens": 56793794.0, + "step": 2223 + }, + { + "epoch": 0.24423457061278278, + "grad_norm": 2.016773223876953, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6989566683769226, + "num_tokens": 56823026.0, + "step": 2224 + }, + { + "epoch": 0.24434438831539645, + "grad_norm": 2.0468966960906982, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.7009063363075256, + "num_tokens": 56852409.0, + "step": 2225 + }, + { + "epoch": 0.2444542060180101, + "grad_norm": 2.3282599449157715, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.7078384160995483, + "num_tokens": 56877330.0, + "step": 2226 + }, + { + "epoch": 0.24456402372062377, + "grad_norm": 2.3554201126098633, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7020220756530762, + "num_tokens": 56900030.0, + "step": 2227 + }, + { + "epoch": 0.24467384142323742, + "grad_norm": 2.3875181674957275, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7174580693244934, + "num_tokens": 56920940.0, + "step": 2228 + }, + { + "epoch": 0.2447836591258511, + "grad_norm": 2.1983749866485596, + "learning_rate": 1e-06, + "loss": 1.0507, + "mean_token_accuracy": 0.6843060255050659, + "num_tokens": 56949889.0, + "step": 2229 + }, + { + "epoch": 0.24489347682846474, + "grad_norm": 2.3006930351257324, + "learning_rate": 1e-06, + "loss": 1.0249, + "mean_token_accuracy": 0.6986798644065857, + "num_tokens": 56973623.0, + "step": 2230 + }, + { + "epoch": 0.2450032945310784, + "grad_norm": 2.0623090267181396, + "learning_rate": 1e-06, + "loss": 1.0869, + "mean_token_accuracy": 0.6828909516334534, + "num_tokens": 57004719.0, + "step": 2231 + }, + { + "epoch": 0.24511311223369206, + "grad_norm": 2.4338207244873047, + "learning_rate": 1e-06, + "loss": 1.0539, + "mean_token_accuracy": 0.6832891702651978, + "num_tokens": 57027800.0, + "step": 2232 + }, + { + "epoch": 0.24522292993630573, + "grad_norm": 2.3144686222076416, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6926366090774536, + "num_tokens": 57053395.0, + "step": 2233 + }, + { + "epoch": 0.2453327476389194, + "grad_norm": 2.5042388439178467, + "learning_rate": 1e-06, + "loss": 1.0412, + "mean_token_accuracy": 0.6870416402816772, + "num_tokens": 57076408.0, + "step": 2234 + }, + { + "epoch": 0.24544256534153305, + "grad_norm": 2.629608392715454, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7087273001670837, + "num_tokens": 57095343.0, + "step": 2235 + }, + { + "epoch": 0.24555238304414673, + "grad_norm": 2.2881200313568115, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7131026983261108, + "num_tokens": 57119013.0, + "step": 2236 + }, + { + "epoch": 0.24566220074676037, + "grad_norm": 2.1758925914764404, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6949794292449951, + "num_tokens": 57146665.0, + "step": 2237 + }, + { + "epoch": 0.24577201844937405, + "grad_norm": 2.2830331325531006, + "learning_rate": 1e-06, + "loss": 1.0337, + "mean_token_accuracy": 0.6836706399917603, + "num_tokens": 57171444.0, + "step": 2238 + }, + { + "epoch": 0.2458818361519877, + "grad_norm": 2.1968541145324707, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.6984691619873047, + "num_tokens": 57199675.0, + "step": 2239 + }, + { + "epoch": 0.24599165385460137, + "grad_norm": 2.194840908050537, + "learning_rate": 1e-06, + "loss": 1.0488, + "mean_token_accuracy": 0.6885656118392944, + "num_tokens": 57228958.0, + "step": 2240 + }, + { + "epoch": 0.246101471557215, + "grad_norm": 2.3914456367492676, + "learning_rate": 1e-06, + "loss": 1.0414, + "mean_token_accuracy": 0.6870994567871094, + "num_tokens": 57252216.0, + "step": 2241 + }, + { + "epoch": 0.2462112892598287, + "grad_norm": 2.496333599090576, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7166610956192017, + "num_tokens": 57271319.0, + "step": 2242 + }, + { + "epoch": 0.24632110696244233, + "grad_norm": 2.2255330085754395, + "learning_rate": 1e-06, + "loss": 1.0895, + "mean_token_accuracy": 0.6737532615661621, + "num_tokens": 57297712.0, + "step": 2243 + }, + { + "epoch": 0.246430924665056, + "grad_norm": 2.2353832721710205, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.7068089246749878, + "num_tokens": 57325843.0, + "step": 2244 + }, + { + "epoch": 0.24654074236766968, + "grad_norm": 2.3782291412353516, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.7122018337249756, + "num_tokens": 57349356.0, + "step": 2245 + }, + { + "epoch": 0.24665056007028333, + "grad_norm": 2.3087990283966064, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7160118222236633, + "num_tokens": 57372809.0, + "step": 2246 + }, + { + "epoch": 0.246760377772897, + "grad_norm": 1.9653304815292358, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7196561098098755, + "num_tokens": 57405197.0, + "step": 2247 + }, + { + "epoch": 0.24687019547551065, + "grad_norm": 2.5349619388580322, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7190405130386353, + "num_tokens": 57425112.0, + "step": 2248 + }, + { + "epoch": 0.24698001317812432, + "grad_norm": 2.1429123878479004, + "learning_rate": 1e-06, + "loss": 1.0392, + "mean_token_accuracy": 0.6865688562393188, + "num_tokens": 57452187.0, + "step": 2249 + }, + { + "epoch": 0.24708983088073797, + "grad_norm": 2.0269930362701416, + "learning_rate": 1e-06, + "loss": 1.0662, + "mean_token_accuracy": 0.6758673191070557, + "num_tokens": 57482552.0, + "step": 2250 + }, + { + "epoch": 0.24719964858335164, + "grad_norm": 2.2115731239318848, + "learning_rate": 1e-06, + "loss": 1.0995, + "mean_token_accuracy": 0.6776158213615417, + "num_tokens": 57512273.0, + "step": 2251 + }, + { + "epoch": 0.24730946628596528, + "grad_norm": 2.1849145889282227, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.7116802930831909, + "num_tokens": 57539498.0, + "step": 2252 + }, + { + "epoch": 0.24741928398857896, + "grad_norm": 2.328030824661255, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7181078195571899, + "num_tokens": 57563892.0, + "step": 2253 + }, + { + "epoch": 0.24752910169119263, + "grad_norm": 2.266627550125122, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7090314626693726, + "num_tokens": 57587661.0, + "step": 2254 + }, + { + "epoch": 0.24763891939380628, + "grad_norm": 2.1981005668640137, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.7025703191757202, + "num_tokens": 57615130.0, + "step": 2255 + }, + { + "epoch": 0.24774873709641995, + "grad_norm": 2.2094178199768066, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.6963474750518799, + "num_tokens": 57641326.0, + "step": 2256 + }, + { + "epoch": 0.2478585547990336, + "grad_norm": 1.9281599521636963, + "learning_rate": 1e-06, + "loss": 1.0567, + "mean_token_accuracy": 0.6833175420761108, + "num_tokens": 57677313.0, + "step": 2257 + }, + { + "epoch": 0.24796837250164727, + "grad_norm": 2.140899181365967, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6969690322875977, + "num_tokens": 57703263.0, + "step": 2258 + }, + { + "epoch": 0.24807819020426092, + "grad_norm": 2.2509043216705322, + "learning_rate": 1e-06, + "loss": 1.0716, + "mean_token_accuracy": 0.6734979152679443, + "num_tokens": 57729481.0, + "step": 2259 + }, + { + "epoch": 0.2481880079068746, + "grad_norm": 2.149949550628662, + "learning_rate": 1e-06, + "loss": 1.0534, + "mean_token_accuracy": 0.6944884061813354, + "num_tokens": 57756889.0, + "step": 2260 + }, + { + "epoch": 0.24829782560948824, + "grad_norm": 2.3055639266967773, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7055152654647827, + "num_tokens": 57780485.0, + "step": 2261 + }, + { + "epoch": 0.2484076433121019, + "grad_norm": 2.276512861251831, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6979683041572571, + "num_tokens": 57805384.0, + "step": 2262 + }, + { + "epoch": 0.24851746101471558, + "grad_norm": 2.3146281242370605, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7107795476913452, + "num_tokens": 57828309.0, + "step": 2263 + }, + { + "epoch": 0.24862727871732923, + "grad_norm": 2.0935373306274414, + "learning_rate": 1e-06, + "loss": 1.0578, + "mean_token_accuracy": 0.6820464134216309, + "num_tokens": 57860122.0, + "step": 2264 + }, + { + "epoch": 0.2487370964199429, + "grad_norm": 2.411402702331543, + "learning_rate": 1e-06, + "loss": 1.0749, + "mean_token_accuracy": 0.6883832812309265, + "num_tokens": 57884701.0, + "step": 2265 + }, + { + "epoch": 0.24884691412255655, + "grad_norm": 2.5555734634399414, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.6970138549804688, + "num_tokens": 57904964.0, + "step": 2266 + }, + { + "epoch": 0.24895673182517022, + "grad_norm": 2.4860293865203857, + "learning_rate": 1e-06, + "loss": 1.028, + "mean_token_accuracy": 0.6956223249435425, + "num_tokens": 57926776.0, + "step": 2267 + }, + { + "epoch": 0.24906654952778387, + "grad_norm": 2.2234652042388916, + "learning_rate": 1e-06, + "loss": 1.1089, + "mean_token_accuracy": 0.6765828132629395, + "num_tokens": 57954431.0, + "step": 2268 + }, + { + "epoch": 0.24917636723039754, + "grad_norm": 2.217101812362671, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.6996710300445557, + "num_tokens": 57979066.0, + "step": 2269 + }, + { + "epoch": 0.2492861849330112, + "grad_norm": 2.406777858734131, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7074276208877563, + "num_tokens": 58000206.0, + "step": 2270 + }, + { + "epoch": 0.24939600263562486, + "grad_norm": 2.0931365489959717, + "learning_rate": 1e-06, + "loss": 1.1064, + "mean_token_accuracy": 0.6734522581100464, + "num_tokens": 58031451.0, + "step": 2271 + }, + { + "epoch": 0.24950582033823854, + "grad_norm": 2.148503303527832, + "learning_rate": 1e-06, + "loss": 1.0577, + "mean_token_accuracy": 0.6921744346618652, + "num_tokens": 58058960.0, + "step": 2272 + }, + { + "epoch": 0.24961563804085218, + "grad_norm": 2.218677043914795, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.716098964214325, + "num_tokens": 58085391.0, + "step": 2273 + }, + { + "epoch": 0.24972545574346586, + "grad_norm": 2.259662389755249, + "learning_rate": 1e-06, + "loss": 1.1182, + "mean_token_accuracy": 0.6848876476287842, + "num_tokens": 58111264.0, + "step": 2274 + }, + { + "epoch": 0.2498352734460795, + "grad_norm": 2.3965847492218018, + "learning_rate": 1e-06, + "loss": 1.0677, + "mean_token_accuracy": 0.67768394947052, + "num_tokens": 58134763.0, + "step": 2275 + }, + { + "epoch": 0.24994509114869318, + "grad_norm": 2.378568649291992, + "learning_rate": 1e-06, + "loss": 1.0812, + "mean_token_accuracy": 0.67568039894104, + "num_tokens": 58158931.0, + "step": 2276 + }, + { + "epoch": 0.2500549088513068, + "grad_norm": 2.2279539108276367, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.6935895681381226, + "num_tokens": 58184909.0, + "step": 2277 + }, + { + "epoch": 0.25016472655392047, + "grad_norm": 2.1199138164520264, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7249516248703003, + "num_tokens": 58210969.0, + "step": 2278 + }, + { + "epoch": 0.25027454425653417, + "grad_norm": 2.3432576656341553, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7136523723602295, + "num_tokens": 58232750.0, + "step": 2279 + }, + { + "epoch": 0.2503843619591478, + "grad_norm": 2.252241611480713, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6924220323562622, + "num_tokens": 58258544.0, + "step": 2280 + }, + { + "epoch": 0.25049417966176146, + "grad_norm": 2.3001675605773926, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.6981280446052551, + "num_tokens": 58283582.0, + "step": 2281 + }, + { + "epoch": 0.25060399736437516, + "grad_norm": 2.395524024963379, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7093260288238525, + "num_tokens": 58306462.0, + "step": 2282 + }, + { + "epoch": 0.2507138150669888, + "grad_norm": 2.438753366470337, + "learning_rate": 1e-06, + "loss": 1.1127, + "mean_token_accuracy": 0.6678549647331238, + "num_tokens": 58329973.0, + "step": 2283 + }, + { + "epoch": 0.25082363276960246, + "grad_norm": 2.1781673431396484, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7047196626663208, + "num_tokens": 58356718.0, + "step": 2284 + }, + { + "epoch": 0.2509334504722161, + "grad_norm": 2.1858017444610596, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7182191610336304, + "num_tokens": 58383086.0, + "step": 2285 + }, + { + "epoch": 0.2510432681748298, + "grad_norm": 2.284466505050659, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7028231620788574, + "num_tokens": 58406086.0, + "step": 2286 + }, + { + "epoch": 0.25115308587744345, + "grad_norm": 2.1299259662628174, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.7078073620796204, + "num_tokens": 58431639.0, + "step": 2287 + }, + { + "epoch": 0.2512629035800571, + "grad_norm": 2.199028491973877, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6851081848144531, + "num_tokens": 58459630.0, + "step": 2288 + }, + { + "epoch": 0.25137272128267074, + "grad_norm": 2.0090084075927734, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6943703889846802, + "num_tokens": 58491372.0, + "step": 2289 + }, + { + "epoch": 0.25148253898528444, + "grad_norm": 2.229212760925293, + "learning_rate": 1e-06, + "loss": 1.0985, + "mean_token_accuracy": 0.6752866506576538, + "num_tokens": 58518407.0, + "step": 2290 + }, + { + "epoch": 0.2515923566878981, + "grad_norm": 2.236816644668579, + "learning_rate": 1e-06, + "loss": 1.079, + "mean_token_accuracy": 0.6829952597618103, + "num_tokens": 58544643.0, + "step": 2291 + }, + { + "epoch": 0.25170217439051173, + "grad_norm": 2.330409049987793, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7050004005432129, + "num_tokens": 58567382.0, + "step": 2292 + }, + { + "epoch": 0.25181199209312544, + "grad_norm": 2.4906630516052246, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7149267196655273, + "num_tokens": 58588170.0, + "step": 2293 + }, + { + "epoch": 0.2519218097957391, + "grad_norm": 2.490373134613037, + "learning_rate": 1e-06, + "loss": 1.025, + "mean_token_accuracy": 0.6942718029022217, + "num_tokens": 58610301.0, + "step": 2294 + }, + { + "epoch": 0.25203162749835273, + "grad_norm": 2.5109267234802246, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7033807039260864, + "num_tokens": 58632488.0, + "step": 2295 + }, + { + "epoch": 0.2521414452009664, + "grad_norm": 2.213366985321045, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6983538269996643, + "num_tokens": 58659403.0, + "step": 2296 + }, + { + "epoch": 0.2522512629035801, + "grad_norm": 2.346011161804199, + "learning_rate": 1e-06, + "loss": 1.115, + "mean_token_accuracy": 0.6693907976150513, + "num_tokens": 58688038.0, + "step": 2297 + }, + { + "epoch": 0.2523610806061937, + "grad_norm": 2.0605366230010986, + "learning_rate": 1e-06, + "loss": 1.1264, + "mean_token_accuracy": 0.6618427038192749, + "num_tokens": 58719377.0, + "step": 2298 + }, + { + "epoch": 0.25247089830880737, + "grad_norm": 2.1415696144104004, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6868687272071838, + "num_tokens": 58747604.0, + "step": 2299 + }, + { + "epoch": 0.252580716011421, + "grad_norm": 2.2390530109405518, + "learning_rate": 1e-06, + "loss": 1.0409, + "mean_token_accuracy": 0.6904201507568359, + "num_tokens": 58776741.0, + "step": 2300 + }, + { + "epoch": 0.2526905337140347, + "grad_norm": 2.1717917919158936, + "learning_rate": 1e-06, + "loss": 1.0591, + "mean_token_accuracy": 0.6887940764427185, + "num_tokens": 58803805.0, + "step": 2301 + }, + { + "epoch": 0.25280035141664836, + "grad_norm": 2.0785367488861084, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7092531323432922, + "num_tokens": 58831254.0, + "step": 2302 + }, + { + "epoch": 0.252910169119262, + "grad_norm": 2.2458198070526123, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7317039966583252, + "num_tokens": 58855871.0, + "step": 2303 + }, + { + "epoch": 0.2530199868218757, + "grad_norm": 1.9587674140930176, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7024233341217041, + "num_tokens": 58886285.0, + "step": 2304 + }, + { + "epoch": 0.25312980452448935, + "grad_norm": 2.1716630458831787, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.7014132738113403, + "num_tokens": 58911515.0, + "step": 2305 + }, + { + "epoch": 0.253239622227103, + "grad_norm": 2.2328054904937744, + "learning_rate": 1e-06, + "loss": 1.0641, + "mean_token_accuracy": 0.6775952577590942, + "num_tokens": 58939350.0, + "step": 2306 + }, + { + "epoch": 0.25334943992971665, + "grad_norm": 2.43611478805542, + "learning_rate": 1e-06, + "loss": 1.0845, + "mean_token_accuracy": 0.6845956444740295, + "num_tokens": 58961493.0, + "step": 2307 + }, + { + "epoch": 0.25345925763233035, + "grad_norm": 2.2175137996673584, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6965942978858948, + "num_tokens": 58989442.0, + "step": 2308 + }, + { + "epoch": 0.253569075334944, + "grad_norm": 2.1605348587036133, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6978649497032166, + "num_tokens": 59017037.0, + "step": 2309 + }, + { + "epoch": 0.25367889303755764, + "grad_norm": 2.005462646484375, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7088892459869385, + "num_tokens": 59046245.0, + "step": 2310 + }, + { + "epoch": 0.25378871074017134, + "grad_norm": 2.266108274459839, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6993244290351868, + "num_tokens": 59072448.0, + "step": 2311 + }, + { + "epoch": 0.253898528442785, + "grad_norm": 2.113067865371704, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7102128863334656, + "num_tokens": 59098565.0, + "step": 2312 + }, + { + "epoch": 0.25400834614539863, + "grad_norm": 2.00521183013916, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6960042715072632, + "num_tokens": 59129323.0, + "step": 2313 + }, + { + "epoch": 0.2541181638480123, + "grad_norm": 2.4006834030151367, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.714449942111969, + "num_tokens": 59151765.0, + "step": 2314 + }, + { + "epoch": 0.254227981550626, + "grad_norm": 2.2092981338500977, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6903270483016968, + "num_tokens": 59177722.0, + "step": 2315 + }, + { + "epoch": 0.2543377992532396, + "grad_norm": 2.332632541656494, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.701090931892395, + "num_tokens": 59202233.0, + "step": 2316 + }, + { + "epoch": 0.2544476169558533, + "grad_norm": 2.4635236263275146, + "learning_rate": 1e-06, + "loss": 1.0221, + "mean_token_accuracy": 0.6886163353919983, + "num_tokens": 59223971.0, + "step": 2317 + }, + { + "epoch": 0.2545574346584669, + "grad_norm": 2.2709696292877197, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.6983920335769653, + "num_tokens": 59248388.0, + "step": 2318 + }, + { + "epoch": 0.2546672523610806, + "grad_norm": 2.4559075832366943, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7210369110107422, + "num_tokens": 59269659.0, + "step": 2319 + }, + { + "epoch": 0.25477707006369427, + "grad_norm": 2.2452051639556885, + "learning_rate": 1e-06, + "loss": 1.038, + "mean_token_accuracy": 0.6969588398933411, + "num_tokens": 59294665.0, + "step": 2320 + }, + { + "epoch": 0.2548868877663079, + "grad_norm": 2.1085779666900635, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6891506314277649, + "num_tokens": 59322599.0, + "step": 2321 + }, + { + "epoch": 0.2549967054689216, + "grad_norm": 2.5056440830230713, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6992851495742798, + "num_tokens": 59343446.0, + "step": 2322 + }, + { + "epoch": 0.25510652317153526, + "grad_norm": 2.3128273487091064, + "learning_rate": 1e-06, + "loss": 1.082, + "mean_token_accuracy": 0.6715510487556458, + "num_tokens": 59366674.0, + "step": 2323 + }, + { + "epoch": 0.2552163408741489, + "grad_norm": 2.2418267726898193, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7031278610229492, + "num_tokens": 59391153.0, + "step": 2324 + }, + { + "epoch": 0.25532615857676255, + "grad_norm": 2.1087191104888916, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.6932128071784973, + "num_tokens": 59418973.0, + "step": 2325 + }, + { + "epoch": 0.25543597627937625, + "grad_norm": 2.4023640155792236, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7222988605499268, + "num_tokens": 59439675.0, + "step": 2326 + }, + { + "epoch": 0.2555457939819899, + "grad_norm": 2.0914511680603027, + "learning_rate": 1e-06, + "loss": 1.1306, + "mean_token_accuracy": 0.6655921339988708, + "num_tokens": 59468908.0, + "step": 2327 + }, + { + "epoch": 0.25565561168460355, + "grad_norm": 2.1434998512268066, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6957365274429321, + "num_tokens": 59495735.0, + "step": 2328 + }, + { + "epoch": 0.25576542938721725, + "grad_norm": 2.1928558349609375, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7266668081283569, + "num_tokens": 59519996.0, + "step": 2329 + }, + { + "epoch": 0.2558752470898309, + "grad_norm": 2.2996366024017334, + "learning_rate": 1e-06, + "loss": 1.0843, + "mean_token_accuracy": 0.6734849810600281, + "num_tokens": 59546009.0, + "step": 2330 + }, + { + "epoch": 0.25598506479244454, + "grad_norm": 2.31377911567688, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.7087854146957397, + "num_tokens": 59570374.0, + "step": 2331 + }, + { + "epoch": 0.2560948824950582, + "grad_norm": 2.040231227874756, + "learning_rate": 1e-06, + "loss": 1.0548, + "mean_token_accuracy": 0.6825549602508545, + "num_tokens": 59600847.0, + "step": 2332 + }, + { + "epoch": 0.2562047001976719, + "grad_norm": 2.551116943359375, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7154449224472046, + "num_tokens": 59621537.0, + "step": 2333 + }, + { + "epoch": 0.25631451790028553, + "grad_norm": 2.3834004402160645, + "learning_rate": 1e-06, + "loss": 1.0571, + "mean_token_accuracy": 0.6863020658493042, + "num_tokens": 59644931.0, + "step": 2334 + }, + { + "epoch": 0.2564243356028992, + "grad_norm": 2.3586716651916504, + "learning_rate": 1e-06, + "loss": 1.0632, + "mean_token_accuracy": 0.679784893989563, + "num_tokens": 59669667.0, + "step": 2335 + }, + { + "epoch": 0.2565341533055128, + "grad_norm": 2.188328981399536, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.704577624797821, + "num_tokens": 59699150.0, + "step": 2336 + }, + { + "epoch": 0.2566439710081265, + "grad_norm": 2.4371888637542725, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.703970193862915, + "num_tokens": 59721049.0, + "step": 2337 + }, + { + "epoch": 0.25675378871074017, + "grad_norm": 2.1521859169006348, + "learning_rate": 1e-06, + "loss": 1.0824, + "mean_token_accuracy": 0.6801319122314453, + "num_tokens": 59750950.0, + "step": 2338 + }, + { + "epoch": 0.2568636064133538, + "grad_norm": 2.2192347049713135, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7083762884140015, + "num_tokens": 59776563.0, + "step": 2339 + }, + { + "epoch": 0.2569734241159675, + "grad_norm": 2.363516330718994, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6889452338218689, + "num_tokens": 59800182.0, + "step": 2340 + }, + { + "epoch": 0.25708324181858117, + "grad_norm": 2.10534405708313, + "learning_rate": 1e-06, + "loss": 1.095, + "mean_token_accuracy": 0.6714403033256531, + "num_tokens": 59828537.0, + "step": 2341 + }, + { + "epoch": 0.2571930595211948, + "grad_norm": 2.305783271789551, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.689285159111023, + "num_tokens": 59852676.0, + "step": 2342 + }, + { + "epoch": 0.25730287722380846, + "grad_norm": 2.21266508102417, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7130411267280579, + "num_tokens": 59876900.0, + "step": 2343 + }, + { + "epoch": 0.25741269492642216, + "grad_norm": 2.3236987590789795, + "learning_rate": 1e-06, + "loss": 1.048, + "mean_token_accuracy": 0.6823994517326355, + "num_tokens": 59901091.0, + "step": 2344 + }, + { + "epoch": 0.2575225126290358, + "grad_norm": 2.1819887161254883, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7102507948875427, + "num_tokens": 59927073.0, + "step": 2345 + }, + { + "epoch": 0.25763233033164945, + "grad_norm": 2.37796688079834, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6843104362487793, + "num_tokens": 59949661.0, + "step": 2346 + }, + { + "epoch": 0.2577421480342631, + "grad_norm": 2.587048053741455, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7116816639900208, + "num_tokens": 59968840.0, + "step": 2347 + }, + { + "epoch": 0.2578519657368768, + "grad_norm": 2.343675374984741, + "learning_rate": 1e-06, + "loss": 1.0404, + "mean_token_accuracy": 0.684493899345398, + "num_tokens": 59993073.0, + "step": 2348 + }, + { + "epoch": 0.25796178343949044, + "grad_norm": 2.1167023181915283, + "learning_rate": 1e-06, + "loss": 1.0947, + "mean_token_accuracy": 0.6726024746894836, + "num_tokens": 60022868.0, + "step": 2349 + }, + { + "epoch": 0.2580716011421041, + "grad_norm": 2.317328929901123, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.698825478553772, + "num_tokens": 60046397.0, + "step": 2350 + }, + { + "epoch": 0.2581814188447178, + "grad_norm": 2.2365047931671143, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6919347047805786, + "num_tokens": 60072107.0, + "step": 2351 + }, + { + "epoch": 0.25829123654733144, + "grad_norm": 2.2558934688568115, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7032157182693481, + "num_tokens": 60099179.0, + "step": 2352 + }, + { + "epoch": 0.2584010542499451, + "grad_norm": 2.1419482231140137, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.705333411693573, + "num_tokens": 60129177.0, + "step": 2353 + }, + { + "epoch": 0.25851087195255873, + "grad_norm": 2.5885403156280518, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7342994213104248, + "num_tokens": 60147648.0, + "step": 2354 + }, + { + "epoch": 0.25862068965517243, + "grad_norm": 2.2084500789642334, + "learning_rate": 1e-06, + "loss": 1.0108, + "mean_token_accuracy": 0.6998224258422852, + "num_tokens": 60174209.0, + "step": 2355 + }, + { + "epoch": 0.2587305073577861, + "grad_norm": 2.381218910217285, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6954797506332397, + "num_tokens": 60195245.0, + "step": 2356 + }, + { + "epoch": 0.2588403250603997, + "grad_norm": 2.584840774536133, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7080757021903992, + "num_tokens": 60215489.0, + "step": 2357 + }, + { + "epoch": 0.2589501427630134, + "grad_norm": 2.1444618701934814, + "learning_rate": 1e-06, + "loss": 1.0374, + "mean_token_accuracy": 0.6828088164329529, + "num_tokens": 60242332.0, + "step": 2358 + }, + { + "epoch": 0.25905996046562707, + "grad_norm": 2.211076021194458, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.696682333946228, + "num_tokens": 60268872.0, + "step": 2359 + }, + { + "epoch": 0.2591697781682407, + "grad_norm": 1.983537197113037, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.6891945600509644, + "num_tokens": 60298568.0, + "step": 2360 + }, + { + "epoch": 0.25927959587085436, + "grad_norm": 2.594778299331665, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7152144312858582, + "num_tokens": 60316823.0, + "step": 2361 + }, + { + "epoch": 0.25938941357346806, + "grad_norm": 2.1513373851776123, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7131138443946838, + "num_tokens": 60341540.0, + "step": 2362 + }, + { + "epoch": 0.2594992312760817, + "grad_norm": 2.358773708343506, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6996962428092957, + "num_tokens": 60364541.0, + "step": 2363 + }, + { + "epoch": 0.25960904897869536, + "grad_norm": 2.2483367919921875, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.6944872140884399, + "num_tokens": 60390281.0, + "step": 2364 + }, + { + "epoch": 0.259718866681309, + "grad_norm": 2.2654738426208496, + "learning_rate": 1e-06, + "loss": 1.082, + "mean_token_accuracy": 0.6739651560783386, + "num_tokens": 60419177.0, + "step": 2365 + }, + { + "epoch": 0.2598286843839227, + "grad_norm": 2.1216020584106445, + "learning_rate": 1e-06, + "loss": 1.0617, + "mean_token_accuracy": 0.6840465664863586, + "num_tokens": 60447637.0, + "step": 2366 + }, + { + "epoch": 0.25993850208653635, + "grad_norm": 2.3164234161376953, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.7011555433273315, + "num_tokens": 60471545.0, + "step": 2367 + }, + { + "epoch": 0.26004831978915, + "grad_norm": 2.31754207611084, + "learning_rate": 1e-06, + "loss": 1.0531, + "mean_token_accuracy": 0.6820158958435059, + "num_tokens": 60496586.0, + "step": 2368 + }, + { + "epoch": 0.2601581374917637, + "grad_norm": 2.194291830062866, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6945375204086304, + "num_tokens": 60521940.0, + "step": 2369 + }, + { + "epoch": 0.26026795519437734, + "grad_norm": 2.470409870147705, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6913496255874634, + "num_tokens": 60544756.0, + "step": 2370 + }, + { + "epoch": 0.260377772896991, + "grad_norm": 2.6220602989196777, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7033429741859436, + "num_tokens": 60564672.0, + "step": 2371 + }, + { + "epoch": 0.26048759059960463, + "grad_norm": 1.9942330121994019, + "learning_rate": 1e-06, + "loss": 1.1087, + "mean_token_accuracy": 0.6759186387062073, + "num_tokens": 60599475.0, + "step": 2372 + }, + { + "epoch": 0.26059740830221834, + "grad_norm": 2.1625521183013916, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7210648059844971, + "num_tokens": 60622730.0, + "step": 2373 + }, + { + "epoch": 0.260707226004832, + "grad_norm": 2.32586932182312, + "learning_rate": 1e-06, + "loss": 1.0434, + "mean_token_accuracy": 0.6887695789337158, + "num_tokens": 60646867.0, + "step": 2374 + }, + { + "epoch": 0.26081704370744563, + "grad_norm": 2.3281192779541016, + "learning_rate": 1e-06, + "loss": 1.1132, + "mean_token_accuracy": 0.6687790155410767, + "num_tokens": 60670751.0, + "step": 2375 + }, + { + "epoch": 0.2609268614100593, + "grad_norm": 2.4473342895507812, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7141475677490234, + "num_tokens": 60691588.0, + "step": 2376 + }, + { + "epoch": 0.261036679112673, + "grad_norm": 2.2648041248321533, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.6996826529502869, + "num_tokens": 60717645.0, + "step": 2377 + }, + { + "epoch": 0.2611464968152866, + "grad_norm": 2.528130292892456, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7004886865615845, + "num_tokens": 60738204.0, + "step": 2378 + }, + { + "epoch": 0.26125631451790027, + "grad_norm": 2.246645927429199, + "learning_rate": 1e-06, + "loss": 1.0667, + "mean_token_accuracy": 0.6818182468414307, + "num_tokens": 60764667.0, + "step": 2379 + }, + { + "epoch": 0.26136613222051397, + "grad_norm": 2.238455057144165, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6869950890541077, + "num_tokens": 60792522.0, + "step": 2380 + }, + { + "epoch": 0.2614759499231276, + "grad_norm": 2.5566859245300293, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7069205045700073, + "num_tokens": 60813260.0, + "step": 2381 + }, + { + "epoch": 0.26158576762574126, + "grad_norm": 2.276294231414795, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.7036796808242798, + "num_tokens": 60840660.0, + "step": 2382 + }, + { + "epoch": 0.2616955853283549, + "grad_norm": 2.5649592876434326, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.695645272731781, + "num_tokens": 60861313.0, + "step": 2383 + }, + { + "epoch": 0.2618054030309686, + "grad_norm": 2.0128190517425537, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.7006368041038513, + "num_tokens": 60894479.0, + "step": 2384 + }, + { + "epoch": 0.26191522073358225, + "grad_norm": 2.2099196910858154, + "learning_rate": 1e-06, + "loss": 1.0465, + "mean_token_accuracy": 0.6868930459022522, + "num_tokens": 60920149.0, + "step": 2385 + }, + { + "epoch": 0.2620250384361959, + "grad_norm": 2.264435052871704, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7095597982406616, + "num_tokens": 60946208.0, + "step": 2386 + }, + { + "epoch": 0.2621348561388096, + "grad_norm": 2.445378541946411, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7020633220672607, + "num_tokens": 60968392.0, + "step": 2387 + }, + { + "epoch": 0.26224467384142325, + "grad_norm": 2.2748286724090576, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.6870609521865845, + "num_tokens": 60993358.0, + "step": 2388 + }, + { + "epoch": 0.2623544915440369, + "grad_norm": 2.035156488418579, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6904735565185547, + "num_tokens": 61024940.0, + "step": 2389 + }, + { + "epoch": 0.26246430924665054, + "grad_norm": 2.5290441513061523, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6940624117851257, + "num_tokens": 61047341.0, + "step": 2390 + }, + { + "epoch": 0.26257412694926424, + "grad_norm": 2.2767789363861084, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6879050135612488, + "num_tokens": 61071639.0, + "step": 2391 + }, + { + "epoch": 0.2626839446518779, + "grad_norm": 2.1833605766296387, + "learning_rate": 1e-06, + "loss": 1.0364, + "mean_token_accuracy": 0.6852440237998962, + "num_tokens": 61100620.0, + "step": 2392 + }, + { + "epoch": 0.26279376235449153, + "grad_norm": 2.318392276763916, + "learning_rate": 1e-06, + "loss": 1.0347, + "mean_token_accuracy": 0.6870545744895935, + "num_tokens": 61124686.0, + "step": 2393 + }, + { + "epoch": 0.2629035800571052, + "grad_norm": 2.2248504161834717, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.7000077962875366, + "num_tokens": 61151513.0, + "step": 2394 + }, + { + "epoch": 0.2630133977597189, + "grad_norm": 2.2856898307800293, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7121918797492981, + "num_tokens": 61174681.0, + "step": 2395 + }, + { + "epoch": 0.2631232154623325, + "grad_norm": 1.912369728088379, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.6924715042114258, + "num_tokens": 61207577.0, + "step": 2396 + }, + { + "epoch": 0.2632330331649462, + "grad_norm": 2.1696109771728516, + "learning_rate": 1e-06, + "loss": 1.0277, + "mean_token_accuracy": 0.6879640221595764, + "num_tokens": 61235796.0, + "step": 2397 + }, + { + "epoch": 0.2633428508675599, + "grad_norm": 2.1746582984924316, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.7024743556976318, + "num_tokens": 61262307.0, + "step": 2398 + }, + { + "epoch": 0.2634526685701735, + "grad_norm": 2.2088840007781982, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6923546195030212, + "num_tokens": 61290118.0, + "step": 2399 + }, + { + "epoch": 0.26356248627278717, + "grad_norm": 2.054227828979492, + "learning_rate": 1e-06, + "loss": 1.0911, + "mean_token_accuracy": 0.6728808879852295, + "num_tokens": 61322120.0, + "step": 2400 + }, + { + "epoch": 0.2636723039754008, + "grad_norm": 2.0294766426086426, + "learning_rate": 1e-06, + "loss": 1.0682, + "mean_token_accuracy": 0.6769708395004272, + "num_tokens": 61352429.0, + "step": 2401 + }, + { + "epoch": 0.2637821216780145, + "grad_norm": 2.120346784591675, + "learning_rate": 1e-06, + "loss": 1.0399, + "mean_token_accuracy": 0.6930686831474304, + "num_tokens": 61379863.0, + "step": 2402 + }, + { + "epoch": 0.26389193938062816, + "grad_norm": 2.1116714477539062, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7158775925636292, + "num_tokens": 61407470.0, + "step": 2403 + }, + { + "epoch": 0.2640017570832418, + "grad_norm": 2.391069173812866, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7060251235961914, + "num_tokens": 61429687.0, + "step": 2404 + }, + { + "epoch": 0.2641115747858555, + "grad_norm": 2.3521130084991455, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7067916393280029, + "num_tokens": 61453654.0, + "step": 2405 + }, + { + "epoch": 0.26422139248846915, + "grad_norm": 2.3257405757904053, + "learning_rate": 1e-06, + "loss": 1.0587, + "mean_token_accuracy": 0.6901538372039795, + "num_tokens": 61478439.0, + "step": 2406 + }, + { + "epoch": 0.2643312101910828, + "grad_norm": 2.618422031402588, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.7003120183944702, + "num_tokens": 61501040.0, + "step": 2407 + }, + { + "epoch": 0.26444102789369645, + "grad_norm": 2.2487752437591553, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7228276133537292, + "num_tokens": 61525554.0, + "step": 2408 + }, + { + "epoch": 0.26455084559631015, + "grad_norm": 2.137620687484741, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6891855597496033, + "num_tokens": 61552935.0, + "step": 2409 + }, + { + "epoch": 0.2646606632989238, + "grad_norm": 2.0543248653411865, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7107011675834656, + "num_tokens": 61581753.0, + "step": 2410 + }, + { + "epoch": 0.26477048100153744, + "grad_norm": 2.3290016651153564, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.69301438331604, + "num_tokens": 61606829.0, + "step": 2411 + }, + { + "epoch": 0.2648802987041511, + "grad_norm": 2.6844351291656494, + "learning_rate": 1e-06, + "loss": 1.0736, + "mean_token_accuracy": 0.6786407828330994, + "num_tokens": 61627504.0, + "step": 2412 + }, + { + "epoch": 0.2649901164067648, + "grad_norm": 2.4543025493621826, + "learning_rate": 1e-06, + "loss": 1.1148, + "mean_token_accuracy": 0.6645876169204712, + "num_tokens": 61653347.0, + "step": 2413 + }, + { + "epoch": 0.26509993410937843, + "grad_norm": 2.436485528945923, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6945318579673767, + "num_tokens": 61676609.0, + "step": 2414 + }, + { + "epoch": 0.2652097518119921, + "grad_norm": 2.6797704696655273, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.6990455985069275, + "num_tokens": 61695330.0, + "step": 2415 + }, + { + "epoch": 0.2653195695146058, + "grad_norm": 2.3979833126068115, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.7067865133285522, + "num_tokens": 61719506.0, + "step": 2416 + }, + { + "epoch": 0.2654293872172194, + "grad_norm": 2.107109546661377, + "learning_rate": 1e-06, + "loss": 1.0674, + "mean_token_accuracy": 0.6761358976364136, + "num_tokens": 61748886.0, + "step": 2417 + }, + { + "epoch": 0.26553920491983307, + "grad_norm": 1.9455339908599854, + "learning_rate": 1e-06, + "loss": 1.0431, + "mean_token_accuracy": 0.6930098533630371, + "num_tokens": 61783778.0, + "step": 2418 + }, + { + "epoch": 0.2656490226224467, + "grad_norm": 2.191951274871826, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.6921590566635132, + "num_tokens": 61811306.0, + "step": 2419 + }, + { + "epoch": 0.2657588403250604, + "grad_norm": 2.165997266769409, + "learning_rate": 1e-06, + "loss": 1.1649, + "mean_token_accuracy": 0.6658210158348083, + "num_tokens": 61840159.0, + "step": 2420 + }, + { + "epoch": 0.26586865802767407, + "grad_norm": 2.134336471557617, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6974807381629944, + "num_tokens": 61866390.0, + "step": 2421 + }, + { + "epoch": 0.2659784757302877, + "grad_norm": 2.135638475418091, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.693790078163147, + "num_tokens": 61896557.0, + "step": 2422 + }, + { + "epoch": 0.26608829343290136, + "grad_norm": 2.7200541496276855, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7246291637420654, + "num_tokens": 61912391.0, + "step": 2423 + }, + { + "epoch": 0.26619811113551506, + "grad_norm": 2.2828924655914307, + "learning_rate": 1e-06, + "loss": 1.0573, + "mean_token_accuracy": 0.6829543113708496, + "num_tokens": 61937766.0, + "step": 2424 + }, + { + "epoch": 0.2663079288381287, + "grad_norm": 2.561356544494629, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7124321460723877, + "num_tokens": 61959799.0, + "step": 2425 + }, + { + "epoch": 0.26641774654074235, + "grad_norm": 2.1015822887420654, + "learning_rate": 1e-06, + "loss": 1.0771, + "mean_token_accuracy": 0.6751280426979065, + "num_tokens": 61991831.0, + "step": 2426 + }, + { + "epoch": 0.26652756424335605, + "grad_norm": 2.2796030044555664, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.7070865631103516, + "num_tokens": 62015792.0, + "step": 2427 + }, + { + "epoch": 0.2666373819459697, + "grad_norm": 1.936172366142273, + "learning_rate": 1e-06, + "loss": 1.1131, + "mean_token_accuracy": 0.6736799478530884, + "num_tokens": 62051219.0, + "step": 2428 + }, + { + "epoch": 0.26674719964858334, + "grad_norm": 2.3300490379333496, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.7011280655860901, + "num_tokens": 62074450.0, + "step": 2429 + }, + { + "epoch": 0.266857017351197, + "grad_norm": 2.187790870666504, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6922122836112976, + "num_tokens": 62099219.0, + "step": 2430 + }, + { + "epoch": 0.2669668350538107, + "grad_norm": 2.337642192840576, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6952238082885742, + "num_tokens": 62123969.0, + "step": 2431 + }, + { + "epoch": 0.26707665275642434, + "grad_norm": 2.1276402473449707, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7062562704086304, + "num_tokens": 62149278.0, + "step": 2432 + }, + { + "epoch": 0.267186470459038, + "grad_norm": 2.3430421352386475, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7316575050354004, + "num_tokens": 62169716.0, + "step": 2433 + }, + { + "epoch": 0.2672962881616517, + "grad_norm": 2.0192477703094482, + "learning_rate": 1e-06, + "loss": 1.0559, + "mean_token_accuracy": 0.6853946447372437, + "num_tokens": 62203279.0, + "step": 2434 + }, + { + "epoch": 0.26740610586426533, + "grad_norm": 2.2472681999206543, + "learning_rate": 1e-06, + "loss": 1.1135, + "mean_token_accuracy": 0.6692081689834595, + "num_tokens": 62231670.0, + "step": 2435 + }, + { + "epoch": 0.267515923566879, + "grad_norm": 2.474335193634033, + "learning_rate": 1e-06, + "loss": 1.0665, + "mean_token_accuracy": 0.6830393075942993, + "num_tokens": 62256251.0, + "step": 2436 + }, + { + "epoch": 0.2676257412694926, + "grad_norm": 2.3097987174987793, + "learning_rate": 1e-06, + "loss": 1.0795, + "mean_token_accuracy": 0.6770362854003906, + "num_tokens": 62280681.0, + "step": 2437 + }, + { + "epoch": 0.2677355589721063, + "grad_norm": 2.0771424770355225, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6853530406951904, + "num_tokens": 62309679.0, + "step": 2438 + }, + { + "epoch": 0.26784537667471997, + "grad_norm": 2.3272132873535156, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7161433696746826, + "num_tokens": 62333484.0, + "step": 2439 + }, + { + "epoch": 0.2679551943773336, + "grad_norm": 2.489342212677002, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.7064510583877563, + "num_tokens": 62355010.0, + "step": 2440 + }, + { + "epoch": 0.26806501207994726, + "grad_norm": 2.270608901977539, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6885063648223877, + "num_tokens": 62380414.0, + "step": 2441 + }, + { + "epoch": 0.26817482978256096, + "grad_norm": 2.0833771228790283, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7112612724304199, + "num_tokens": 62408115.0, + "step": 2442 + }, + { + "epoch": 0.2682846474851746, + "grad_norm": 2.1769559383392334, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.6995769739151001, + "num_tokens": 62434101.0, + "step": 2443 + }, + { + "epoch": 0.26839446518778826, + "grad_norm": 2.2236087322235107, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7133684158325195, + "num_tokens": 62460659.0, + "step": 2444 + }, + { + "epoch": 0.26850428289040196, + "grad_norm": 2.1735262870788574, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6857339143753052, + "num_tokens": 62488424.0, + "step": 2445 + }, + { + "epoch": 0.2686141005930156, + "grad_norm": 2.6244914531707764, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7255263328552246, + "num_tokens": 62507796.0, + "step": 2446 + }, + { + "epoch": 0.26872391829562925, + "grad_norm": 2.0290298461914062, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6936029195785522, + "num_tokens": 62537262.0, + "step": 2447 + }, + { + "epoch": 0.2688337359982429, + "grad_norm": 2.1171646118164062, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.6957529187202454, + "num_tokens": 62564009.0, + "step": 2448 + }, + { + "epoch": 0.2689435537008566, + "grad_norm": 2.4059455394744873, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7090873718261719, + "num_tokens": 62586426.0, + "step": 2449 + }, + { + "epoch": 0.26905337140347024, + "grad_norm": 2.301894426345825, + "learning_rate": 1e-06, + "loss": 1.0777, + "mean_token_accuracy": 0.6801207661628723, + "num_tokens": 62612245.0, + "step": 2450 + }, + { + "epoch": 0.2691631891060839, + "grad_norm": 2.2876858711242676, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.7044512033462524, + "num_tokens": 62638175.0, + "step": 2451 + }, + { + "epoch": 0.26927300680869753, + "grad_norm": 2.166238307952881, + "learning_rate": 1e-06, + "loss": 1.0362, + "mean_token_accuracy": 0.7006576061248779, + "num_tokens": 62665529.0, + "step": 2452 + }, + { + "epoch": 0.26938282451131124, + "grad_norm": 2.229275703430176, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.7010617256164551, + "num_tokens": 62692370.0, + "step": 2453 + }, + { + "epoch": 0.2694926422139249, + "grad_norm": 2.184448480606079, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7072464227676392, + "num_tokens": 62717731.0, + "step": 2454 + }, + { + "epoch": 0.26960245991653853, + "grad_norm": 2.5393362045288086, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.708441436290741, + "num_tokens": 62738234.0, + "step": 2455 + }, + { + "epoch": 0.26971227761915223, + "grad_norm": 2.0218589305877686, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.698572039604187, + "num_tokens": 62768498.0, + "step": 2456 + }, + { + "epoch": 0.2698220953217659, + "grad_norm": 2.3499596118927, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6989153623580933, + "num_tokens": 62791802.0, + "step": 2457 + }, + { + "epoch": 0.2699319130243795, + "grad_norm": 2.375483512878418, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7145278453826904, + "num_tokens": 62813417.0, + "step": 2458 + }, + { + "epoch": 0.27004173072699317, + "grad_norm": 2.506340742111206, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7017896175384521, + "num_tokens": 62832839.0, + "step": 2459 + }, + { + "epoch": 0.27015154842960687, + "grad_norm": 1.9043424129486084, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6931276321411133, + "num_tokens": 62863999.0, + "step": 2460 + }, + { + "epoch": 0.2702613661322205, + "grad_norm": 2.1811633110046387, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.6880080699920654, + "num_tokens": 62890965.0, + "step": 2461 + }, + { + "epoch": 0.27037118383483416, + "grad_norm": 2.2185540199279785, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7054194211959839, + "num_tokens": 62916213.0, + "step": 2462 + }, + { + "epoch": 0.27048100153744786, + "grad_norm": 2.1924142837524414, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.692497968673706, + "num_tokens": 62941541.0, + "step": 2463 + }, + { + "epoch": 0.2705908192400615, + "grad_norm": 2.2441000938415527, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6922152042388916, + "num_tokens": 62967788.0, + "step": 2464 + }, + { + "epoch": 0.27070063694267515, + "grad_norm": 2.433807849884033, + "learning_rate": 1e-06, + "loss": 1.0433, + "mean_token_accuracy": 0.6863075494766235, + "num_tokens": 62988792.0, + "step": 2465 + }, + { + "epoch": 0.2708104546452888, + "grad_norm": 2.5516648292541504, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6983711123466492, + "num_tokens": 63008842.0, + "step": 2466 + }, + { + "epoch": 0.2709202723479025, + "grad_norm": 2.361974000930786, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6903567314147949, + "num_tokens": 63032415.0, + "step": 2467 + }, + { + "epoch": 0.27103009005051615, + "grad_norm": 2.1846249103546143, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6959118843078613, + "num_tokens": 63060424.0, + "step": 2468 + }, + { + "epoch": 0.2711399077531298, + "grad_norm": 2.0834124088287354, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7122718095779419, + "num_tokens": 63087332.0, + "step": 2469 + }, + { + "epoch": 0.27124972545574344, + "grad_norm": 2.1321310997009277, + "learning_rate": 1e-06, + "loss": 1.0475, + "mean_token_accuracy": 0.690150260925293, + "num_tokens": 63116928.0, + "step": 2470 + }, + { + "epoch": 0.27135954315835714, + "grad_norm": 2.410614490509033, + "learning_rate": 1e-06, + "loss": 1.1005, + "mean_token_accuracy": 0.6641944646835327, + "num_tokens": 63140646.0, + "step": 2471 + }, + { + "epoch": 0.2714693608609708, + "grad_norm": 2.2206332683563232, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7049455046653748, + "num_tokens": 63166362.0, + "step": 2472 + }, + { + "epoch": 0.27157917856358443, + "grad_norm": 2.0235114097595215, + "learning_rate": 1e-06, + "loss": 1.1045, + "mean_token_accuracy": 0.6658315658569336, + "num_tokens": 63199490.0, + "step": 2473 + }, + { + "epoch": 0.27168899626619814, + "grad_norm": 2.429008960723877, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6951342821121216, + "num_tokens": 63223423.0, + "step": 2474 + }, + { + "epoch": 0.2717988139688118, + "grad_norm": 2.1294314861297607, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.6957656741142273, + "num_tokens": 63251296.0, + "step": 2475 + }, + { + "epoch": 0.2719086316714254, + "grad_norm": 2.4364476203918457, + "learning_rate": 1e-06, + "loss": 1.0707, + "mean_token_accuracy": 0.6841748356819153, + "num_tokens": 63274535.0, + "step": 2476 + }, + { + "epoch": 0.2720184493740391, + "grad_norm": 2.1276872158050537, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.7023266553878784, + "num_tokens": 63301892.0, + "step": 2477 + }, + { + "epoch": 0.2721282670766528, + "grad_norm": 2.2190825939178467, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.7043394446372986, + "num_tokens": 63325324.0, + "step": 2478 + }, + { + "epoch": 0.2722380847792664, + "grad_norm": 2.1671860218048096, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6950470805168152, + "num_tokens": 63353410.0, + "step": 2479 + }, + { + "epoch": 0.27234790248188007, + "grad_norm": 2.3859364986419678, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6985310316085815, + "num_tokens": 63377193.0, + "step": 2480 + }, + { + "epoch": 0.27245772018449377, + "grad_norm": 2.2094123363494873, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6988515257835388, + "num_tokens": 63405305.0, + "step": 2481 + }, + { + "epoch": 0.2725675378871074, + "grad_norm": 1.954272747039795, + "learning_rate": 1e-06, + "loss": 1.0353, + "mean_token_accuracy": 0.696609616279602, + "num_tokens": 63438448.0, + "step": 2482 + }, + { + "epoch": 0.27267735558972106, + "grad_norm": 2.488574743270874, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6948980093002319, + "num_tokens": 63460375.0, + "step": 2483 + }, + { + "epoch": 0.2727871732923347, + "grad_norm": 1.9943097829818726, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6976409554481506, + "num_tokens": 63490583.0, + "step": 2484 + }, + { + "epoch": 0.2728969909949484, + "grad_norm": 2.468158721923828, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7123095989227295, + "num_tokens": 63511216.0, + "step": 2485 + }, + { + "epoch": 0.27300680869756205, + "grad_norm": 2.273160934448242, + "learning_rate": 1e-06, + "loss": 1.0567, + "mean_token_accuracy": 0.6844362020492554, + "num_tokens": 63536257.0, + "step": 2486 + }, + { + "epoch": 0.2731166264001757, + "grad_norm": 1.9099143743515015, + "learning_rate": 1e-06, + "loss": 1.082, + "mean_token_accuracy": 0.6774210929870605, + "num_tokens": 63571033.0, + "step": 2487 + }, + { + "epoch": 0.27322644410278935, + "grad_norm": 2.306457042694092, + "learning_rate": 1e-06, + "loss": 1.0565, + "mean_token_accuracy": 0.6846132874488831, + "num_tokens": 63597793.0, + "step": 2488 + }, + { + "epoch": 0.27333626180540305, + "grad_norm": 2.0948286056518555, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.7037152051925659, + "num_tokens": 63626160.0, + "step": 2489 + }, + { + "epoch": 0.2734460795080167, + "grad_norm": 2.102078676223755, + "learning_rate": 1e-06, + "loss": 1.0844, + "mean_token_accuracy": 0.6815969944000244, + "num_tokens": 63654873.0, + "step": 2490 + }, + { + "epoch": 0.27355589721063034, + "grad_norm": 2.5918381214141846, + "learning_rate": 1e-06, + "loss": 1.0854, + "mean_token_accuracy": 0.6693123579025269, + "num_tokens": 63674906.0, + "step": 2491 + }, + { + "epoch": 0.27366571491324404, + "grad_norm": 2.35896372795105, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7319358587265015, + "num_tokens": 63697227.0, + "step": 2492 + }, + { + "epoch": 0.2737755326158577, + "grad_norm": 2.6438536643981934, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.6986962556838989, + "num_tokens": 63717929.0, + "step": 2493 + }, + { + "epoch": 0.27388535031847133, + "grad_norm": 2.252760887145996, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7208479642868042, + "num_tokens": 63742186.0, + "step": 2494 + }, + { + "epoch": 0.273995168021085, + "grad_norm": 2.634472608566284, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.71448814868927, + "num_tokens": 63761677.0, + "step": 2495 + }, + { + "epoch": 0.2741049857236987, + "grad_norm": 2.1914432048797607, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.686762809753418, + "num_tokens": 63791428.0, + "step": 2496 + }, + { + "epoch": 0.2742148034263123, + "grad_norm": 2.2704453468322754, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.7132384181022644, + "num_tokens": 63816471.0, + "step": 2497 + }, + { + "epoch": 0.27432462112892597, + "grad_norm": 2.0283474922180176, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.7019782066345215, + "num_tokens": 63847562.0, + "step": 2498 + }, + { + "epoch": 0.2744344388315396, + "grad_norm": 2.2310941219329834, + "learning_rate": 1e-06, + "loss": 1.0564, + "mean_token_accuracy": 0.6866379976272583, + "num_tokens": 63875758.0, + "step": 2499 + }, + { + "epoch": 0.2745442565341533, + "grad_norm": 2.288997173309326, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7051383256912231, + "num_tokens": 63900409.0, + "step": 2500 + }, + { + "epoch": 0.27465407423676697, + "grad_norm": 2.3381776809692383, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7053126096725464, + "num_tokens": 63926106.0, + "step": 2501 + }, + { + "epoch": 0.2747638919393806, + "grad_norm": 2.5833382606506348, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7113110423088074, + "num_tokens": 63945036.0, + "step": 2502 + }, + { + "epoch": 0.2748737096419943, + "grad_norm": 2.453460216522217, + "learning_rate": 1e-06, + "loss": 1.0738, + "mean_token_accuracy": 0.6828811168670654, + "num_tokens": 63968455.0, + "step": 2503 + }, + { + "epoch": 0.27498352734460796, + "grad_norm": 2.1824772357940674, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.692825198173523, + "num_tokens": 63995837.0, + "step": 2504 + }, + { + "epoch": 0.2750933450472216, + "grad_norm": 2.1776633262634277, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.702522873878479, + "num_tokens": 64021620.0, + "step": 2505 + }, + { + "epoch": 0.27520316274983525, + "grad_norm": 2.1524763107299805, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.7021176218986511, + "num_tokens": 64049784.0, + "step": 2506 + }, + { + "epoch": 0.27531298045244895, + "grad_norm": 2.2974672317504883, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.7034837007522583, + "num_tokens": 64074145.0, + "step": 2507 + }, + { + "epoch": 0.2754227981550626, + "grad_norm": 2.5274248123168945, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.6950281262397766, + "num_tokens": 64095196.0, + "step": 2508 + }, + { + "epoch": 0.27553261585767624, + "grad_norm": 2.300022840499878, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6975696682929993, + "num_tokens": 64119993.0, + "step": 2509 + }, + { + "epoch": 0.27564243356028995, + "grad_norm": 2.4357540607452393, + "learning_rate": 1e-06, + "loss": 1.0835, + "mean_token_accuracy": 0.673126220703125, + "num_tokens": 64141658.0, + "step": 2510 + }, + { + "epoch": 0.2757522512629036, + "grad_norm": 1.9498188495635986, + "learning_rate": 1e-06, + "loss": 1.0583, + "mean_token_accuracy": 0.6814817190170288, + "num_tokens": 64173344.0, + "step": 2511 + }, + { + "epoch": 0.27586206896551724, + "grad_norm": 2.300942897796631, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6979366540908813, + "num_tokens": 64197642.0, + "step": 2512 + }, + { + "epoch": 0.2759718866681309, + "grad_norm": 2.2744743824005127, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6973669528961182, + "num_tokens": 64223023.0, + "step": 2513 + }, + { + "epoch": 0.2760817043707446, + "grad_norm": 2.4002726078033447, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.696422815322876, + "num_tokens": 64248026.0, + "step": 2514 + }, + { + "epoch": 0.27619152207335823, + "grad_norm": 2.6481919288635254, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7161542773246765, + "num_tokens": 64268512.0, + "step": 2515 + }, + { + "epoch": 0.2763013397759719, + "grad_norm": 2.2329459190368652, + "learning_rate": 1e-06, + "loss": 1.0574, + "mean_token_accuracy": 0.6918315887451172, + "num_tokens": 64295454.0, + "step": 2516 + }, + { + "epoch": 0.2764111574785855, + "grad_norm": 2.525739908218384, + "learning_rate": 1e-06, + "loss": 1.0457, + "mean_token_accuracy": 0.6888241767883301, + "num_tokens": 64317464.0, + "step": 2517 + }, + { + "epoch": 0.2765209751811992, + "grad_norm": 1.985634684562683, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6989680528640747, + "num_tokens": 64351659.0, + "step": 2518 + }, + { + "epoch": 0.27663079288381287, + "grad_norm": 2.307882070541382, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7007918953895569, + "num_tokens": 64377045.0, + "step": 2519 + }, + { + "epoch": 0.2767406105864265, + "grad_norm": 2.2073135375976562, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6902941465377808, + "num_tokens": 64405275.0, + "step": 2520 + }, + { + "epoch": 0.2768504282890402, + "grad_norm": 2.349346399307251, + "learning_rate": 1e-06, + "loss": 1.0348, + "mean_token_accuracy": 0.6934027671813965, + "num_tokens": 64429427.0, + "step": 2521 + }, + { + "epoch": 0.27696024599165386, + "grad_norm": 2.321037530899048, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7154359817504883, + "num_tokens": 64452227.0, + "step": 2522 + }, + { + "epoch": 0.2770700636942675, + "grad_norm": 2.12846040725708, + "learning_rate": 1e-06, + "loss": 1.0234, + "mean_token_accuracy": 0.6956469416618347, + "num_tokens": 64480562.0, + "step": 2523 + }, + { + "epoch": 0.27717988139688116, + "grad_norm": 2.556363582611084, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.702580988407135, + "num_tokens": 64500726.0, + "step": 2524 + }, + { + "epoch": 0.27728969909949486, + "grad_norm": 2.29477858543396, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7085250020027161, + "num_tokens": 64524619.0, + "step": 2525 + }, + { + "epoch": 0.2773995168021085, + "grad_norm": 2.668210744857788, + "learning_rate": 1e-06, + "loss": 1.0474, + "mean_token_accuracy": 0.6823869943618774, + "num_tokens": 64542149.0, + "step": 2526 + }, + { + "epoch": 0.27750933450472215, + "grad_norm": 2.0409324169158936, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.6983039379119873, + "num_tokens": 64574049.0, + "step": 2527 + }, + { + "epoch": 0.2776191522073358, + "grad_norm": 2.064523458480835, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6965345740318298, + "num_tokens": 64604318.0, + "step": 2528 + }, + { + "epoch": 0.2777289699099495, + "grad_norm": 2.1702280044555664, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7048263549804688, + "num_tokens": 64632673.0, + "step": 2529 + }, + { + "epoch": 0.27783878761256314, + "grad_norm": 2.1355350017547607, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7121925354003906, + "num_tokens": 64658018.0, + "step": 2530 + }, + { + "epoch": 0.2779486053151768, + "grad_norm": 2.552351713180542, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.712482213973999, + "num_tokens": 64677050.0, + "step": 2531 + }, + { + "epoch": 0.2780584230177905, + "grad_norm": 2.1806752681732178, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7024074792861938, + "num_tokens": 64702699.0, + "step": 2532 + }, + { + "epoch": 0.27816824072040414, + "grad_norm": 2.381033182144165, + "learning_rate": 1e-06, + "loss": 1.1113, + "mean_token_accuracy": 0.6645489931106567, + "num_tokens": 64727740.0, + "step": 2533 + }, + { + "epoch": 0.2782780584230178, + "grad_norm": 2.6276752948760986, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7050963640213013, + "num_tokens": 64749883.0, + "step": 2534 + }, + { + "epoch": 0.27838787612563143, + "grad_norm": 2.0717034339904785, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7254841923713684, + "num_tokens": 64776936.0, + "step": 2535 + }, + { + "epoch": 0.27849769382824513, + "grad_norm": 2.192262887954712, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7208715081214905, + "num_tokens": 64801700.0, + "step": 2536 + }, + { + "epoch": 0.2786075115308588, + "grad_norm": 2.2673704624176025, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7078806757926941, + "num_tokens": 64825015.0, + "step": 2537 + }, + { + "epoch": 0.2787173292334724, + "grad_norm": 2.3164050579071045, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.7029478549957275, + "num_tokens": 64849540.0, + "step": 2538 + }, + { + "epoch": 0.2788271469360861, + "grad_norm": 1.9975354671478271, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6885606050491333, + "num_tokens": 64881462.0, + "step": 2539 + }, + { + "epoch": 0.27893696463869977, + "grad_norm": 2.4588241577148438, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6947513818740845, + "num_tokens": 64904567.0, + "step": 2540 + }, + { + "epoch": 0.2790467823413134, + "grad_norm": 2.060112714767456, + "learning_rate": 1e-06, + "loss": 1.0855, + "mean_token_accuracy": 0.6818886995315552, + "num_tokens": 64933992.0, + "step": 2541 + }, + { + "epoch": 0.27915660004392706, + "grad_norm": 2.345048427581787, + "learning_rate": 1e-06, + "loss": 1.0326, + "mean_token_accuracy": 0.6834138631820679, + "num_tokens": 64957890.0, + "step": 2542 + }, + { + "epoch": 0.27926641774654076, + "grad_norm": 2.0837349891662598, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7086438536643982, + "num_tokens": 64985285.0, + "step": 2543 + }, + { + "epoch": 0.2793762354491544, + "grad_norm": 2.403721570968628, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.6891399025917053, + "num_tokens": 65008581.0, + "step": 2544 + }, + { + "epoch": 0.27948605315176805, + "grad_norm": 2.0806708335876465, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.697907030582428, + "num_tokens": 65037334.0, + "step": 2545 + }, + { + "epoch": 0.2795958708543817, + "grad_norm": 2.4800350666046143, + "learning_rate": 1e-06, + "loss": 1.0704, + "mean_token_accuracy": 0.6792898178100586, + "num_tokens": 65058838.0, + "step": 2546 + }, + { + "epoch": 0.2797056885569954, + "grad_norm": 2.345032215118408, + "learning_rate": 1e-06, + "loss": 1.0427, + "mean_token_accuracy": 0.6863342523574829, + "num_tokens": 65082277.0, + "step": 2547 + }, + { + "epoch": 0.27981550625960905, + "grad_norm": 2.0633699893951416, + "learning_rate": 1e-06, + "loss": 1.1033, + "mean_token_accuracy": 0.66949862241745, + "num_tokens": 65114396.0, + "step": 2548 + }, + { + "epoch": 0.2799253239622227, + "grad_norm": 2.2727766036987305, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7081067562103271, + "num_tokens": 65140333.0, + "step": 2549 + }, + { + "epoch": 0.2800351416648364, + "grad_norm": 2.4072067737579346, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.7030788660049438, + "num_tokens": 65163529.0, + "step": 2550 + }, + { + "epoch": 0.28014495936745004, + "grad_norm": 2.615272283554077, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6860972046852112, + "num_tokens": 65183450.0, + "step": 2551 + }, + { + "epoch": 0.2802547770700637, + "grad_norm": 2.3823320865631104, + "learning_rate": 1e-06, + "loss": 1.0434, + "mean_token_accuracy": 0.6880142688751221, + "num_tokens": 65205790.0, + "step": 2552 + }, + { + "epoch": 0.28036459477267733, + "grad_norm": 2.1252641677856445, + "learning_rate": 1e-06, + "loss": 1.1041, + "mean_token_accuracy": 0.6725629568099976, + "num_tokens": 65234769.0, + "step": 2553 + }, + { + "epoch": 0.28047441247529104, + "grad_norm": 2.05196475982666, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.6867678761482239, + "num_tokens": 65265631.0, + "step": 2554 + }, + { + "epoch": 0.2805842301779047, + "grad_norm": 2.3548660278320312, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.7097405195236206, + "num_tokens": 65288628.0, + "step": 2555 + }, + { + "epoch": 0.2806940478805183, + "grad_norm": 2.4156529903411865, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.692599892616272, + "num_tokens": 65311137.0, + "step": 2556 + }, + { + "epoch": 0.28080386558313203, + "grad_norm": 2.1182234287261963, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.7015590667724609, + "num_tokens": 65339649.0, + "step": 2557 + }, + { + "epoch": 0.2809136832857457, + "grad_norm": 2.4744155406951904, + "learning_rate": 1e-06, + "loss": 1.0807, + "mean_token_accuracy": 0.6740902662277222, + "num_tokens": 65362568.0, + "step": 2558 + }, + { + "epoch": 0.2810235009883593, + "grad_norm": 2.1627800464630127, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7147535085678101, + "num_tokens": 65387034.0, + "step": 2559 + }, + { + "epoch": 0.28113331869097297, + "grad_norm": 2.41037917137146, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7080749273300171, + "num_tokens": 65409315.0, + "step": 2560 + }, + { + "epoch": 0.28124313639358667, + "grad_norm": 2.1667683124542236, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7053278684616089, + "num_tokens": 65434572.0, + "step": 2561 + }, + { + "epoch": 0.2813529540962003, + "grad_norm": 2.153390884399414, + "learning_rate": 1e-06, + "loss": 1.0667, + "mean_token_accuracy": 0.6819567680358887, + "num_tokens": 65461677.0, + "step": 2562 + }, + { + "epoch": 0.28146277179881396, + "grad_norm": 2.6817731857299805, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7112293839454651, + "num_tokens": 65480824.0, + "step": 2563 + }, + { + "epoch": 0.2815725895014276, + "grad_norm": 2.160196304321289, + "learning_rate": 1e-06, + "loss": 1.0697, + "mean_token_accuracy": 0.6778261065483093, + "num_tokens": 65509899.0, + "step": 2564 + }, + { + "epoch": 0.2816824072040413, + "grad_norm": 2.3096210956573486, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7091621160507202, + "num_tokens": 65533127.0, + "step": 2565 + }, + { + "epoch": 0.28179222490665495, + "grad_norm": 2.436304807662964, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.7052619457244873, + "num_tokens": 65554509.0, + "step": 2566 + }, + { + "epoch": 0.2819020426092686, + "grad_norm": 2.3654210567474365, + "learning_rate": 1e-06, + "loss": 1.0984, + "mean_token_accuracy": 0.6808323860168457, + "num_tokens": 65578203.0, + "step": 2567 + }, + { + "epoch": 0.2820118603118823, + "grad_norm": 1.9709806442260742, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6893006563186646, + "num_tokens": 65608336.0, + "step": 2568 + }, + { + "epoch": 0.28212167801449595, + "grad_norm": 2.4585280418395996, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7151432037353516, + "num_tokens": 65629255.0, + "step": 2569 + }, + { + "epoch": 0.2822314957171096, + "grad_norm": 2.256458282470703, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6972851753234863, + "num_tokens": 65653731.0, + "step": 2570 + }, + { + "epoch": 0.28234131341972324, + "grad_norm": 2.264859199523926, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.7058737874031067, + "num_tokens": 65677867.0, + "step": 2571 + }, + { + "epoch": 0.28245113112233694, + "grad_norm": 2.366602659225464, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.7005751132965088, + "num_tokens": 65701188.0, + "step": 2572 + }, + { + "epoch": 0.2825609488249506, + "grad_norm": 2.2162744998931885, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7285164594650269, + "num_tokens": 65724081.0, + "step": 2573 + }, + { + "epoch": 0.28267076652756423, + "grad_norm": 2.5579211711883545, + "learning_rate": 1e-06, + "loss": 1.0516, + "mean_token_accuracy": 0.6835985779762268, + "num_tokens": 65746254.0, + "step": 2574 + }, + { + "epoch": 0.2827805842301779, + "grad_norm": 2.1359641551971436, + "learning_rate": 1e-06, + "loss": 1.0221, + "mean_token_accuracy": 0.6896514892578125, + "num_tokens": 65772958.0, + "step": 2575 + }, + { + "epoch": 0.2828904019327916, + "grad_norm": 2.368537425994873, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7018353939056396, + "num_tokens": 65795861.0, + "step": 2576 + }, + { + "epoch": 0.2830002196354052, + "grad_norm": 2.392587661743164, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7068310976028442, + "num_tokens": 65818611.0, + "step": 2577 + }, + { + "epoch": 0.28311003733801887, + "grad_norm": 2.197749137878418, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.700994610786438, + "num_tokens": 65845445.0, + "step": 2578 + }, + { + "epoch": 0.2832198550406326, + "grad_norm": 2.383864402770996, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7044563889503479, + "num_tokens": 65869470.0, + "step": 2579 + }, + { + "epoch": 0.2833296727432462, + "grad_norm": 2.2984659671783447, + "learning_rate": 1e-06, + "loss": 1.0728, + "mean_token_accuracy": 0.6821913719177246, + "num_tokens": 65896125.0, + "step": 2580 + }, + { + "epoch": 0.28343949044585987, + "grad_norm": 1.955784797668457, + "learning_rate": 1e-06, + "loss": 1.0835, + "mean_token_accuracy": 0.6790758371353149, + "num_tokens": 65927862.0, + "step": 2581 + }, + { + "epoch": 0.2835493081484735, + "grad_norm": 2.1092231273651123, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.708039402961731, + "num_tokens": 65957499.0, + "step": 2582 + }, + { + "epoch": 0.2836591258510872, + "grad_norm": 2.2595374584198, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.716803252696991, + "num_tokens": 65981372.0, + "step": 2583 + }, + { + "epoch": 0.28376894355370086, + "grad_norm": 2.0893208980560303, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.706817090511322, + "num_tokens": 66008705.0, + "step": 2584 + }, + { + "epoch": 0.2838787612563145, + "grad_norm": 2.5554518699645996, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6939188241958618, + "num_tokens": 66029459.0, + "step": 2585 + }, + { + "epoch": 0.2839885789589282, + "grad_norm": 2.183048725128174, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7138186097145081, + "num_tokens": 66057540.0, + "step": 2586 + }, + { + "epoch": 0.28409839666154185, + "grad_norm": 1.8679386377334595, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7092380523681641, + "num_tokens": 66091337.0, + "step": 2587 + }, + { + "epoch": 0.2842082143641555, + "grad_norm": 2.3198294639587402, + "learning_rate": 1e-06, + "loss": 1.0302, + "mean_token_accuracy": 0.6947637796401978, + "num_tokens": 66115236.0, + "step": 2588 + }, + { + "epoch": 0.28431803206676914, + "grad_norm": 2.4107887744903564, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7002429366111755, + "num_tokens": 66137315.0, + "step": 2589 + }, + { + "epoch": 0.28442784976938285, + "grad_norm": 2.0707991123199463, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6998111605644226, + "num_tokens": 66167517.0, + "step": 2590 + }, + { + "epoch": 0.2845376674719965, + "grad_norm": 2.2046091556549072, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6943467855453491, + "num_tokens": 66192284.0, + "step": 2591 + }, + { + "epoch": 0.28464748517461014, + "grad_norm": 2.197638511657715, + "learning_rate": 1e-06, + "loss": 1.055, + "mean_token_accuracy": 0.6852083206176758, + "num_tokens": 66218554.0, + "step": 2592 + }, + { + "epoch": 0.2847573028772238, + "grad_norm": 2.2366433143615723, + "learning_rate": 1e-06, + "loss": 1.0954, + "mean_token_accuracy": 0.6752657294273376, + "num_tokens": 66248144.0, + "step": 2593 + }, + { + "epoch": 0.2848671205798375, + "grad_norm": 2.360703706741333, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.6971374750137329, + "num_tokens": 66271404.0, + "step": 2594 + }, + { + "epoch": 0.28497693828245113, + "grad_norm": 2.366598129272461, + "learning_rate": 1e-06, + "loss": 1.0475, + "mean_token_accuracy": 0.6931858658790588, + "num_tokens": 66295335.0, + "step": 2595 + }, + { + "epoch": 0.2850867559850648, + "grad_norm": 2.321204423904419, + "learning_rate": 1e-06, + "loss": 1.0239, + "mean_token_accuracy": 0.6943732500076294, + "num_tokens": 66319877.0, + "step": 2596 + }, + { + "epoch": 0.2851965736876785, + "grad_norm": 2.250859022140503, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7127933502197266, + "num_tokens": 66345146.0, + "step": 2597 + }, + { + "epoch": 0.2853063913902921, + "grad_norm": 2.1117517948150635, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.7003011703491211, + "num_tokens": 66373293.0, + "step": 2598 + }, + { + "epoch": 0.28541620909290577, + "grad_norm": 2.3956353664398193, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.712297797203064, + "num_tokens": 66396686.0, + "step": 2599 + }, + { + "epoch": 0.2855260267955194, + "grad_norm": 2.1969363689422607, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6898373365402222, + "num_tokens": 66424796.0, + "step": 2600 + }, + { + "epoch": 0.2856358444981331, + "grad_norm": 2.164583921432495, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7095227241516113, + "num_tokens": 66450969.0, + "step": 2601 + }, + { + "epoch": 0.28574566220074676, + "grad_norm": 2.2086029052734375, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.7020160555839539, + "num_tokens": 66475542.0, + "step": 2602 + }, + { + "epoch": 0.2858554799033604, + "grad_norm": 2.0789852142333984, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7263226509094238, + "num_tokens": 66502581.0, + "step": 2603 + }, + { + "epoch": 0.28596529760597406, + "grad_norm": 2.476576089859009, + "learning_rate": 1e-06, + "loss": 1.0947, + "mean_token_accuracy": 0.6788693070411682, + "num_tokens": 66525194.0, + "step": 2604 + }, + { + "epoch": 0.28607511530858776, + "grad_norm": 2.113957405090332, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.701548159122467, + "num_tokens": 66552974.0, + "step": 2605 + }, + { + "epoch": 0.2861849330112014, + "grad_norm": 2.3563098907470703, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7052792906761169, + "num_tokens": 66578503.0, + "step": 2606 + }, + { + "epoch": 0.28629475071381505, + "grad_norm": 2.2533411979675293, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7104818820953369, + "num_tokens": 66601596.0, + "step": 2607 + }, + { + "epoch": 0.28640456841642875, + "grad_norm": 2.3177454471588135, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6916441917419434, + "num_tokens": 66626626.0, + "step": 2608 + }, + { + "epoch": 0.2865143861190424, + "grad_norm": 2.4340338706970215, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7205073833465576, + "num_tokens": 66648107.0, + "step": 2609 + }, + { + "epoch": 0.28662420382165604, + "grad_norm": 2.524918556213379, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7072722315788269, + "num_tokens": 66669009.0, + "step": 2610 + }, + { + "epoch": 0.2867340215242697, + "grad_norm": 2.133169174194336, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6948721408843994, + "num_tokens": 66697457.0, + "step": 2611 + }, + { + "epoch": 0.2868438392268834, + "grad_norm": 1.958382487297058, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7074054479598999, + "num_tokens": 66726812.0, + "step": 2612 + }, + { + "epoch": 0.28695365692949704, + "grad_norm": 2.5361618995666504, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.707096517086029, + "num_tokens": 66747961.0, + "step": 2613 + }, + { + "epoch": 0.2870634746321107, + "grad_norm": 2.385424852371216, + "learning_rate": 1e-06, + "loss": 1.0427, + "mean_token_accuracy": 0.687201976776123, + "num_tokens": 66771553.0, + "step": 2614 + }, + { + "epoch": 0.2871732923347244, + "grad_norm": 2.337932825088501, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7017828226089478, + "num_tokens": 66796301.0, + "step": 2615 + }, + { + "epoch": 0.28728311003733803, + "grad_norm": 2.3686978816986084, + "learning_rate": 1e-06, + "loss": 1.1065, + "mean_token_accuracy": 0.6780087947845459, + "num_tokens": 66821067.0, + "step": 2616 + }, + { + "epoch": 0.2873929277399517, + "grad_norm": 2.6668312549591064, + "learning_rate": 1e-06, + "loss": 1.039, + "mean_token_accuracy": 0.6881182193756104, + "num_tokens": 66841021.0, + "step": 2617 + }, + { + "epoch": 0.2875027454425653, + "grad_norm": 2.499980926513672, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6903609037399292, + "num_tokens": 66862528.0, + "step": 2618 + }, + { + "epoch": 0.287612563145179, + "grad_norm": 2.605729818344116, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6944596767425537, + "num_tokens": 66885210.0, + "step": 2619 + }, + { + "epoch": 0.28772238084779267, + "grad_norm": 2.2424490451812744, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.686772882938385, + "num_tokens": 66910035.0, + "step": 2620 + }, + { + "epoch": 0.2878321985504063, + "grad_norm": 2.3984529972076416, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7176132202148438, + "num_tokens": 66931983.0, + "step": 2621 + }, + { + "epoch": 0.28794201625301996, + "grad_norm": 2.402249336242676, + "learning_rate": 1e-06, + "loss": 1.0905, + "mean_token_accuracy": 0.6789863109588623, + "num_tokens": 66958100.0, + "step": 2622 + }, + { + "epoch": 0.28805183395563366, + "grad_norm": 2.3525350093841553, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7229243516921997, + "num_tokens": 66980100.0, + "step": 2623 + }, + { + "epoch": 0.2881616516582473, + "grad_norm": 2.5108163356781006, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6931164860725403, + "num_tokens": 67003133.0, + "step": 2624 + }, + { + "epoch": 0.28827146936086095, + "grad_norm": 2.0849485397338867, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.702242374420166, + "num_tokens": 67030644.0, + "step": 2625 + }, + { + "epoch": 0.28838128706347466, + "grad_norm": 2.2856478691101074, + "learning_rate": 1e-06, + "loss": 1.0452, + "mean_token_accuracy": 0.6858713626861572, + "num_tokens": 67055656.0, + "step": 2626 + }, + { + "epoch": 0.2884911047660883, + "grad_norm": 2.3850135803222656, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.7078753113746643, + "num_tokens": 67078756.0, + "step": 2627 + }, + { + "epoch": 0.28860092246870195, + "grad_norm": 2.195037603378296, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.7055754661560059, + "num_tokens": 67104154.0, + "step": 2628 + }, + { + "epoch": 0.2887107401713156, + "grad_norm": 2.1484296321868896, + "learning_rate": 1e-06, + "loss": 1.0508, + "mean_token_accuracy": 0.6847285032272339, + "num_tokens": 67132291.0, + "step": 2629 + }, + { + "epoch": 0.2888205578739293, + "grad_norm": 2.462610960006714, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.714458703994751, + "num_tokens": 67152821.0, + "step": 2630 + }, + { + "epoch": 0.28893037557654294, + "grad_norm": 2.2070202827453613, + "learning_rate": 1e-06, + "loss": 1.0434, + "mean_token_accuracy": 0.6812155246734619, + "num_tokens": 67178958.0, + "step": 2631 + }, + { + "epoch": 0.2890401932791566, + "grad_norm": 2.2285048961639404, + "learning_rate": 1e-06, + "loss": 1.0242, + "mean_token_accuracy": 0.6950002908706665, + "num_tokens": 67205776.0, + "step": 2632 + }, + { + "epoch": 0.2891500109817703, + "grad_norm": 2.233074426651001, + "learning_rate": 1e-06, + "loss": 1.1082, + "mean_token_accuracy": 0.6760428547859192, + "num_tokens": 67233662.0, + "step": 2633 + }, + { + "epoch": 0.28925982868438394, + "grad_norm": 2.0515732765197754, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.7003658413887024, + "num_tokens": 67264455.0, + "step": 2634 + }, + { + "epoch": 0.2893696463869976, + "grad_norm": 2.2190377712249756, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.6980865597724915, + "num_tokens": 67290152.0, + "step": 2635 + }, + { + "epoch": 0.2894794640896112, + "grad_norm": 2.3537204265594482, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.6974407434463501, + "num_tokens": 67313180.0, + "step": 2636 + }, + { + "epoch": 0.28958928179222493, + "grad_norm": 2.3669943809509277, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6976613402366638, + "num_tokens": 67337350.0, + "step": 2637 + }, + { + "epoch": 0.2896990994948386, + "grad_norm": 2.4057388305664062, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7008596658706665, + "num_tokens": 67360942.0, + "step": 2638 + }, + { + "epoch": 0.2898089171974522, + "grad_norm": 2.197174310684204, + "learning_rate": 1e-06, + "loss": 1.0737, + "mean_token_accuracy": 0.6807409524917603, + "num_tokens": 67388728.0, + "step": 2639 + }, + { + "epoch": 0.28991873490006587, + "grad_norm": 2.2308826446533203, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.7014889717102051, + "num_tokens": 67414703.0, + "step": 2640 + }, + { + "epoch": 0.29002855260267957, + "grad_norm": 2.1979050636291504, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7004827857017517, + "num_tokens": 67440064.0, + "step": 2641 + }, + { + "epoch": 0.2901383703052932, + "grad_norm": 1.9272189140319824, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7214723825454712, + "num_tokens": 67472336.0, + "step": 2642 + }, + { + "epoch": 0.29024818800790686, + "grad_norm": 2.1982131004333496, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7201897501945496, + "num_tokens": 67497499.0, + "step": 2643 + }, + { + "epoch": 0.29035800571052056, + "grad_norm": 2.3458995819091797, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7109552621841431, + "num_tokens": 67519660.0, + "step": 2644 + }, + { + "epoch": 0.2904678234131342, + "grad_norm": 2.1675188541412354, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6959003806114197, + "num_tokens": 67547443.0, + "step": 2645 + }, + { + "epoch": 0.29057764111574785, + "grad_norm": 2.32273268699646, + "learning_rate": 1e-06, + "loss": 1.0722, + "mean_token_accuracy": 0.6823946237564087, + "num_tokens": 67571103.0, + "step": 2646 + }, + { + "epoch": 0.2906874588183615, + "grad_norm": 2.0823304653167725, + "learning_rate": 1e-06, + "loss": 1.073, + "mean_token_accuracy": 0.6850053668022156, + "num_tokens": 67601777.0, + "step": 2647 + }, + { + "epoch": 0.2907972765209752, + "grad_norm": 2.1519007682800293, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7090316414833069, + "num_tokens": 67627605.0, + "step": 2648 + }, + { + "epoch": 0.29090709422358885, + "grad_norm": 2.6623547077178955, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7046152949333191, + "num_tokens": 67648033.0, + "step": 2649 + }, + { + "epoch": 0.2910169119262025, + "grad_norm": 2.1328847408294678, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6896842122077942, + "num_tokens": 67675422.0, + "step": 2650 + }, + { + "epoch": 0.29112672962881614, + "grad_norm": 2.3986527919769287, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.688288152217865, + "num_tokens": 67700922.0, + "step": 2651 + }, + { + "epoch": 0.29123654733142984, + "grad_norm": 2.411430835723877, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.6971457004547119, + "num_tokens": 67724430.0, + "step": 2652 + }, + { + "epoch": 0.2913463650340435, + "grad_norm": 2.3863584995269775, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.7011771202087402, + "num_tokens": 67746112.0, + "step": 2653 + }, + { + "epoch": 0.29145618273665713, + "grad_norm": 2.5545380115509033, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.72374427318573, + "num_tokens": 67766135.0, + "step": 2654 + }, + { + "epoch": 0.29156600043927083, + "grad_norm": 2.1590659618377686, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7033926248550415, + "num_tokens": 67792287.0, + "step": 2655 + }, + { + "epoch": 0.2916758181418845, + "grad_norm": 2.637094497680664, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7041798830032349, + "num_tokens": 67812379.0, + "step": 2656 + }, + { + "epoch": 0.2917856358444981, + "grad_norm": 2.293261766433716, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.6925612688064575, + "num_tokens": 67837772.0, + "step": 2657 + }, + { + "epoch": 0.29189545354711177, + "grad_norm": 2.376258611679077, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.7009352445602417, + "num_tokens": 67859863.0, + "step": 2658 + }, + { + "epoch": 0.2920052712497255, + "grad_norm": 2.3615036010742188, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6958655118942261, + "num_tokens": 67882532.0, + "step": 2659 + }, + { + "epoch": 0.2921150889523391, + "grad_norm": 2.3059849739074707, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7194929122924805, + "num_tokens": 67904873.0, + "step": 2660 + }, + { + "epoch": 0.29222490665495277, + "grad_norm": 2.1715798377990723, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6862850189208984, + "num_tokens": 67928447.0, + "step": 2661 + }, + { + "epoch": 0.29233472435756647, + "grad_norm": 2.2921204566955566, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7115589380264282, + "num_tokens": 67953421.0, + "step": 2662 + }, + { + "epoch": 0.2924445420601801, + "grad_norm": 2.3272838592529297, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6901476383209229, + "num_tokens": 67977417.0, + "step": 2663 + }, + { + "epoch": 0.29255435976279376, + "grad_norm": 2.196873188018799, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7235991954803467, + "num_tokens": 68002434.0, + "step": 2664 + }, + { + "epoch": 0.2926641774654074, + "grad_norm": 2.4643375873565674, + "learning_rate": 1e-06, + "loss": 1.0316, + "mean_token_accuracy": 0.6984585523605347, + "num_tokens": 68024198.0, + "step": 2665 + }, + { + "epoch": 0.2927739951680211, + "grad_norm": 2.391955614089966, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7120429277420044, + "num_tokens": 68045467.0, + "step": 2666 + }, + { + "epoch": 0.29288381287063475, + "grad_norm": 2.084730386734009, + "learning_rate": 1e-06, + "loss": 1.0566, + "mean_token_accuracy": 0.6858909130096436, + "num_tokens": 68073329.0, + "step": 2667 + }, + { + "epoch": 0.2929936305732484, + "grad_norm": 2.1334667205810547, + "learning_rate": 1e-06, + "loss": 1.0582, + "mean_token_accuracy": 0.6916158199310303, + "num_tokens": 68101943.0, + "step": 2668 + }, + { + "epoch": 0.29310344827586204, + "grad_norm": 2.037642002105713, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6981891393661499, + "num_tokens": 68131675.0, + "step": 2669 + }, + { + "epoch": 0.29321326597847575, + "grad_norm": 2.723499298095703, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.7096786499023438, + "num_tokens": 68149515.0, + "step": 2670 + }, + { + "epoch": 0.2933230836810894, + "grad_norm": 2.1880924701690674, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7047109603881836, + "num_tokens": 68175007.0, + "step": 2671 + }, + { + "epoch": 0.29343290138370304, + "grad_norm": 2.090184211730957, + "learning_rate": 1e-06, + "loss": 1.0489, + "mean_token_accuracy": 0.6860401034355164, + "num_tokens": 68203592.0, + "step": 2672 + }, + { + "epoch": 0.29354271908631674, + "grad_norm": 2.539173126220703, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7045170068740845, + "num_tokens": 68221948.0, + "step": 2673 + }, + { + "epoch": 0.2936525367889304, + "grad_norm": 2.1022701263427734, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.7013460993766785, + "num_tokens": 68249277.0, + "step": 2674 + }, + { + "epoch": 0.29376235449154403, + "grad_norm": 2.2821080684661865, + "learning_rate": 1e-06, + "loss": 1.1123, + "mean_token_accuracy": 0.6729671359062195, + "num_tokens": 68275838.0, + "step": 2675 + }, + { + "epoch": 0.2938721721941577, + "grad_norm": 2.398378372192383, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6943076252937317, + "num_tokens": 68299336.0, + "step": 2676 + }, + { + "epoch": 0.2939819898967714, + "grad_norm": 2.211843252182007, + "learning_rate": 1e-06, + "loss": 1.0872, + "mean_token_accuracy": 0.6869089603424072, + "num_tokens": 68328109.0, + "step": 2677 + }, + { + "epoch": 0.294091807599385, + "grad_norm": 2.5615811347961426, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7054843902587891, + "num_tokens": 68348625.0, + "step": 2678 + }, + { + "epoch": 0.29420162530199867, + "grad_norm": 2.413796901702881, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7040696144104004, + "num_tokens": 68370199.0, + "step": 2679 + }, + { + "epoch": 0.2943114430046123, + "grad_norm": 2.517669200897217, + "learning_rate": 1e-06, + "loss": 1.0525, + "mean_token_accuracy": 0.6787403225898743, + "num_tokens": 68391639.0, + "step": 2680 + }, + { + "epoch": 0.294421260707226, + "grad_norm": 2.1459975242614746, + "learning_rate": 1e-06, + "loss": 1.0306, + "mean_token_accuracy": 0.6897411346435547, + "num_tokens": 68419917.0, + "step": 2681 + }, + { + "epoch": 0.29453107840983966, + "grad_norm": 2.487006425857544, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.690514862537384, + "num_tokens": 68439864.0, + "step": 2682 + }, + { + "epoch": 0.2946408961124533, + "grad_norm": 2.2168331146240234, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7082945108413696, + "num_tokens": 68465940.0, + "step": 2683 + }, + { + "epoch": 0.294750713815067, + "grad_norm": 2.2776856422424316, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6950376033782959, + "num_tokens": 68490238.0, + "step": 2684 + }, + { + "epoch": 0.29486053151768066, + "grad_norm": 2.3677921295166016, + "learning_rate": 1e-06, + "loss": 1.049, + "mean_token_accuracy": 0.673320472240448, + "num_tokens": 68512341.0, + "step": 2685 + }, + { + "epoch": 0.2949703492202943, + "grad_norm": 2.3445849418640137, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7202490568161011, + "num_tokens": 68535112.0, + "step": 2686 + }, + { + "epoch": 0.29508016692290795, + "grad_norm": 2.372492551803589, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7056829929351807, + "num_tokens": 68558242.0, + "step": 2687 + }, + { + "epoch": 0.29518998462552165, + "grad_norm": 2.226181983947754, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.69600510597229, + "num_tokens": 68583913.0, + "step": 2688 + }, + { + "epoch": 0.2952998023281353, + "grad_norm": 2.132622003555298, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7023522853851318, + "num_tokens": 68610517.0, + "step": 2689 + }, + { + "epoch": 0.29540962003074894, + "grad_norm": 2.7437121868133545, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.6950575113296509, + "num_tokens": 68628039.0, + "step": 2690 + }, + { + "epoch": 0.29551943773336264, + "grad_norm": 2.2090134620666504, + "learning_rate": 1e-06, + "loss": 1.0518, + "mean_token_accuracy": 0.6870585680007935, + "num_tokens": 68655559.0, + "step": 2691 + }, + { + "epoch": 0.2956292554359763, + "grad_norm": 2.2466530799865723, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7060555815696716, + "num_tokens": 68681342.0, + "step": 2692 + }, + { + "epoch": 0.29573907313858994, + "grad_norm": 2.155799150466919, + "learning_rate": 1e-06, + "loss": 1.056, + "mean_token_accuracy": 0.6832712888717651, + "num_tokens": 68710501.0, + "step": 2693 + }, + { + "epoch": 0.2958488908412036, + "grad_norm": 2.1959662437438965, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7012555003166199, + "num_tokens": 68737285.0, + "step": 2694 + }, + { + "epoch": 0.2959587085438173, + "grad_norm": 2.376525640487671, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.70184725522995, + "num_tokens": 68760136.0, + "step": 2695 + }, + { + "epoch": 0.29606852624643093, + "grad_norm": 2.252922296524048, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.727602481842041, + "num_tokens": 68784443.0, + "step": 2696 + }, + { + "epoch": 0.2961783439490446, + "grad_norm": 2.2479135990142822, + "learning_rate": 1e-06, + "loss": 1.0286, + "mean_token_accuracy": 0.6997740268707275, + "num_tokens": 68809715.0, + "step": 2697 + }, + { + "epoch": 0.2962881616516582, + "grad_norm": 2.0541718006134033, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6945858001708984, + "num_tokens": 68838736.0, + "step": 2698 + }, + { + "epoch": 0.2963979793542719, + "grad_norm": 2.7216169834136963, + "learning_rate": 1e-06, + "loss": 1.0463, + "mean_token_accuracy": 0.6769826412200928, + "num_tokens": 68858196.0, + "step": 2699 + }, + { + "epoch": 0.29650779705688557, + "grad_norm": 2.2055728435516357, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6962285041809082, + "num_tokens": 68883333.0, + "step": 2700 + }, + { + "epoch": 0.2966176147594992, + "grad_norm": 2.151276111602783, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6870418190956116, + "num_tokens": 68910596.0, + "step": 2701 + }, + { + "epoch": 0.2967274324621129, + "grad_norm": 2.238030195236206, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7171698808670044, + "num_tokens": 68935145.0, + "step": 2702 + }, + { + "epoch": 0.29683725016472656, + "grad_norm": 3.0660903453826904, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7279590368270874, + "num_tokens": 68949986.0, + "step": 2703 + }, + { + "epoch": 0.2969470678673402, + "grad_norm": 2.0602364540100098, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7199370265007019, + "num_tokens": 68975936.0, + "step": 2704 + }, + { + "epoch": 0.29705688556995385, + "grad_norm": 2.4655282497406006, + "learning_rate": 1e-06, + "loss": 1.0436, + "mean_token_accuracy": 0.683967113494873, + "num_tokens": 69000349.0, + "step": 2705 + }, + { + "epoch": 0.29716670327256756, + "grad_norm": 2.5990378856658936, + "learning_rate": 1e-06, + "loss": 1.0455, + "mean_token_accuracy": 0.6853499412536621, + "num_tokens": 69020827.0, + "step": 2706 + }, + { + "epoch": 0.2972765209751812, + "grad_norm": 2.3122987747192383, + "learning_rate": 1e-06, + "loss": 1.0632, + "mean_token_accuracy": 0.6851516962051392, + "num_tokens": 69046144.0, + "step": 2707 + }, + { + "epoch": 0.29738633867779485, + "grad_norm": 2.3340771198272705, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.7003465890884399, + "num_tokens": 69071603.0, + "step": 2708 + }, + { + "epoch": 0.29749615638040855, + "grad_norm": 2.4516103267669678, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7111814022064209, + "num_tokens": 69092517.0, + "step": 2709 + }, + { + "epoch": 0.2976059740830222, + "grad_norm": 2.20977520942688, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.6873461008071899, + "num_tokens": 69119873.0, + "step": 2710 + }, + { + "epoch": 0.29771579178563584, + "grad_norm": 2.1388113498687744, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.698512613773346, + "num_tokens": 69150332.0, + "step": 2711 + }, + { + "epoch": 0.2978256094882495, + "grad_norm": 2.0942301750183105, + "learning_rate": 1e-06, + "loss": 1.0296, + "mean_token_accuracy": 0.6953051090240479, + "num_tokens": 69178028.0, + "step": 2712 + }, + { + "epoch": 0.2979354271908632, + "grad_norm": 2.6278786659240723, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7175118923187256, + "num_tokens": 69195942.0, + "step": 2713 + }, + { + "epoch": 0.29804524489347684, + "grad_norm": 2.470656156539917, + "learning_rate": 1e-06, + "loss": 1.0706, + "mean_token_accuracy": 0.6825807094573975, + "num_tokens": 69218205.0, + "step": 2714 + }, + { + "epoch": 0.2981550625960905, + "grad_norm": 2.157209634780884, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7063180208206177, + "num_tokens": 69245179.0, + "step": 2715 + }, + { + "epoch": 0.2982648802987041, + "grad_norm": 2.5074236392974854, + "learning_rate": 1e-06, + "loss": 1.0613, + "mean_token_accuracy": 0.6804260611534119, + "num_tokens": 69269178.0, + "step": 2716 + }, + { + "epoch": 0.29837469800131783, + "grad_norm": 2.075143814086914, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.68708735704422, + "num_tokens": 69300376.0, + "step": 2717 + }, + { + "epoch": 0.2984845157039315, + "grad_norm": 2.571272373199463, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7126160264015198, + "num_tokens": 69321730.0, + "step": 2718 + }, + { + "epoch": 0.2985943334065451, + "grad_norm": 2.5701262950897217, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7122489213943481, + "num_tokens": 69342834.0, + "step": 2719 + }, + { + "epoch": 0.2987041511091588, + "grad_norm": 2.452514886856079, + "learning_rate": 1e-06, + "loss": 1.0506, + "mean_token_accuracy": 0.680874228477478, + "num_tokens": 69367008.0, + "step": 2720 + }, + { + "epoch": 0.29881396881177247, + "grad_norm": 2.249760866165161, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6961326599121094, + "num_tokens": 69394714.0, + "step": 2721 + }, + { + "epoch": 0.2989237865143861, + "grad_norm": 2.3890492916107178, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6919329166412354, + "num_tokens": 69419148.0, + "step": 2722 + }, + { + "epoch": 0.29903360421699976, + "grad_norm": 2.22881817817688, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7170619368553162, + "num_tokens": 69442239.0, + "step": 2723 + }, + { + "epoch": 0.29914342191961346, + "grad_norm": 2.162501573562622, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.714870274066925, + "num_tokens": 69468562.0, + "step": 2724 + }, + { + "epoch": 0.2992532396222271, + "grad_norm": 2.2275335788726807, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6998247504234314, + "num_tokens": 69496086.0, + "step": 2725 + }, + { + "epoch": 0.29936305732484075, + "grad_norm": 2.4905920028686523, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7000352144241333, + "num_tokens": 69518256.0, + "step": 2726 + }, + { + "epoch": 0.2994728750274544, + "grad_norm": 2.8272902965545654, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7141778469085693, + "num_tokens": 69535338.0, + "step": 2727 + }, + { + "epoch": 0.2995826927300681, + "grad_norm": 2.4240329265594482, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.6906934976577759, + "num_tokens": 69559988.0, + "step": 2728 + }, + { + "epoch": 0.29969251043268175, + "grad_norm": 2.5227670669555664, + "learning_rate": 1e-06, + "loss": 1.0141, + "mean_token_accuracy": 0.6926530599594116, + "num_tokens": 69581726.0, + "step": 2729 + }, + { + "epoch": 0.2998023281352954, + "grad_norm": 2.24155592918396, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7029293775558472, + "num_tokens": 69607108.0, + "step": 2730 + }, + { + "epoch": 0.2999121458379091, + "grad_norm": 2.2833330631256104, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.7036531567573547, + "num_tokens": 69633679.0, + "step": 2731 + }, + { + "epoch": 0.30002196354052274, + "grad_norm": 2.2800655364990234, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.7046208381652832, + "num_tokens": 69659856.0, + "step": 2732 + }, + { + "epoch": 0.3001317812431364, + "grad_norm": 2.4334654808044434, + "learning_rate": 1e-06, + "loss": 1.0373, + "mean_token_accuracy": 0.6969247460365295, + "num_tokens": 69683263.0, + "step": 2733 + }, + { + "epoch": 0.30024159894575003, + "grad_norm": 2.4926326274871826, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6956510543823242, + "num_tokens": 69707874.0, + "step": 2734 + }, + { + "epoch": 0.30035141664836373, + "grad_norm": 1.8542149066925049, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.7032744288444519, + "num_tokens": 69745004.0, + "step": 2735 + }, + { + "epoch": 0.3004612343509774, + "grad_norm": 2.1322238445281982, + "learning_rate": 1e-06, + "loss": 1.0586, + "mean_token_accuracy": 0.6875267624855042, + "num_tokens": 69774227.0, + "step": 2736 + }, + { + "epoch": 0.300571052053591, + "grad_norm": 2.05747389793396, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7297326326370239, + "num_tokens": 69802772.0, + "step": 2737 + }, + { + "epoch": 0.3006808697562047, + "grad_norm": 2.3544704914093018, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7265843749046326, + "num_tokens": 69824559.0, + "step": 2738 + }, + { + "epoch": 0.3007906874588184, + "grad_norm": 2.2528209686279297, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6983593106269836, + "num_tokens": 69848179.0, + "step": 2739 + }, + { + "epoch": 0.300900505161432, + "grad_norm": 2.158111095428467, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.6915028691291809, + "num_tokens": 69876001.0, + "step": 2740 + }, + { + "epoch": 0.30101032286404567, + "grad_norm": 2.211799383163452, + "learning_rate": 1e-06, + "loss": 1.0521, + "mean_token_accuracy": 0.6806520819664001, + "num_tokens": 69904734.0, + "step": 2741 + }, + { + "epoch": 0.30112014056665937, + "grad_norm": 2.276381015777588, + "learning_rate": 1e-06, + "loss": 1.0682, + "mean_token_accuracy": 0.6727930307388306, + "num_tokens": 69930423.0, + "step": 2742 + }, + { + "epoch": 0.301229958269273, + "grad_norm": 2.075869560241699, + "learning_rate": 1e-06, + "loss": 1.0321, + "mean_token_accuracy": 0.6904904842376709, + "num_tokens": 69958860.0, + "step": 2743 + }, + { + "epoch": 0.30133977597188666, + "grad_norm": 2.4041123390197754, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7058014869689941, + "num_tokens": 69981972.0, + "step": 2744 + }, + { + "epoch": 0.3014495936745003, + "grad_norm": 2.093778610229492, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6877337694168091, + "num_tokens": 70011432.0, + "step": 2745 + }, + { + "epoch": 0.301559411377114, + "grad_norm": 2.627441167831421, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.7157756090164185, + "num_tokens": 70031991.0, + "step": 2746 + }, + { + "epoch": 0.30166922907972765, + "grad_norm": 2.656019449234009, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7074301242828369, + "num_tokens": 70052101.0, + "step": 2747 + }, + { + "epoch": 0.3017790467823413, + "grad_norm": 2.5443408489227295, + "learning_rate": 1e-06, + "loss": 1.105, + "mean_token_accuracy": 0.6802688837051392, + "num_tokens": 70075984.0, + "step": 2748 + }, + { + "epoch": 0.301888864484955, + "grad_norm": 2.399033308029175, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.68631911277771, + "num_tokens": 70099432.0, + "step": 2749 + }, + { + "epoch": 0.30199868218756865, + "grad_norm": 2.1681466102600098, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6962806582450867, + "num_tokens": 70129005.0, + "step": 2750 + }, + { + "epoch": 0.3021084998901823, + "grad_norm": 2.3465843200683594, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.6826356649398804, + "num_tokens": 70152077.0, + "step": 2751 + }, + { + "epoch": 0.30221831759279594, + "grad_norm": 2.305755615234375, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.6855681538581848, + "num_tokens": 70175911.0, + "step": 2752 + }, + { + "epoch": 0.30232813529540964, + "grad_norm": 2.07474684715271, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.6933653354644775, + "num_tokens": 70207533.0, + "step": 2753 + }, + { + "epoch": 0.3024379529980233, + "grad_norm": 2.168208360671997, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.7096951603889465, + "num_tokens": 70234103.0, + "step": 2754 + }, + { + "epoch": 0.30254777070063693, + "grad_norm": 2.0305073261260986, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7175865173339844, + "num_tokens": 70263607.0, + "step": 2755 + }, + { + "epoch": 0.3026575884032506, + "grad_norm": 2.2449944019317627, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6936087608337402, + "num_tokens": 70289929.0, + "step": 2756 + }, + { + "epoch": 0.3027674061058643, + "grad_norm": 2.298994302749634, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6957857012748718, + "num_tokens": 70313669.0, + "step": 2757 + }, + { + "epoch": 0.3028772238084779, + "grad_norm": 1.9663939476013184, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.702843427658081, + "num_tokens": 70343649.0, + "step": 2758 + }, + { + "epoch": 0.30298704151109157, + "grad_norm": 2.144479990005493, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7176944017410278, + "num_tokens": 70372223.0, + "step": 2759 + }, + { + "epoch": 0.30309685921370527, + "grad_norm": 2.642706871032715, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7159541845321655, + "num_tokens": 70390203.0, + "step": 2760 + }, + { + "epoch": 0.3032066769163189, + "grad_norm": 2.2795825004577637, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.711073637008667, + "num_tokens": 70416006.0, + "step": 2761 + }, + { + "epoch": 0.30331649461893256, + "grad_norm": 2.2651851177215576, + "learning_rate": 1e-06, + "loss": 1.0682, + "mean_token_accuracy": 0.6859625577926636, + "num_tokens": 70442589.0, + "step": 2762 + }, + { + "epoch": 0.3034263123215462, + "grad_norm": 2.285759210586548, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7130746841430664, + "num_tokens": 70464704.0, + "step": 2763 + }, + { + "epoch": 0.3035361300241599, + "grad_norm": 2.374028205871582, + "learning_rate": 1e-06, + "loss": 1.0286, + "mean_token_accuracy": 0.6838072538375854, + "num_tokens": 70487360.0, + "step": 2764 + }, + { + "epoch": 0.30364594772677356, + "grad_norm": 2.223311424255371, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7107261419296265, + "num_tokens": 70512576.0, + "step": 2765 + }, + { + "epoch": 0.3037557654293872, + "grad_norm": 2.370788812637329, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.690938413143158, + "num_tokens": 70537355.0, + "step": 2766 + }, + { + "epoch": 0.3038655831320009, + "grad_norm": 2.156876802444458, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6960993409156799, + "num_tokens": 70566689.0, + "step": 2767 + }, + { + "epoch": 0.30397540083461455, + "grad_norm": 2.34798526763916, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7114284038543701, + "num_tokens": 70591605.0, + "step": 2768 + }, + { + "epoch": 0.3040852185372282, + "grad_norm": 2.3007988929748535, + "learning_rate": 1e-06, + "loss": 1.0353, + "mean_token_accuracy": 0.6951063871383667, + "num_tokens": 70617601.0, + "step": 2769 + }, + { + "epoch": 0.30419503623984184, + "grad_norm": 2.4765608310699463, + "learning_rate": 1e-06, + "loss": 1.0775, + "mean_token_accuracy": 0.6821599006652832, + "num_tokens": 70640189.0, + "step": 2770 + }, + { + "epoch": 0.30430485394245554, + "grad_norm": 2.6086459159851074, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.6991297006607056, + "num_tokens": 70661242.0, + "step": 2771 + }, + { + "epoch": 0.3044146716450692, + "grad_norm": 2.5203375816345215, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6872342824935913, + "num_tokens": 70683753.0, + "step": 2772 + }, + { + "epoch": 0.30452448934768284, + "grad_norm": 2.108686685562134, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.7044450044631958, + "num_tokens": 70712538.0, + "step": 2773 + }, + { + "epoch": 0.3046343070502965, + "grad_norm": 2.5669844150543213, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7069595456123352, + "num_tokens": 70734652.0, + "step": 2774 + }, + { + "epoch": 0.3047441247529102, + "grad_norm": 2.0246400833129883, + "learning_rate": 1e-06, + "loss": 1.0472, + "mean_token_accuracy": 0.6831682920455933, + "num_tokens": 70765429.0, + "step": 2775 + }, + { + "epoch": 0.30485394245552383, + "grad_norm": 1.9656888246536255, + "learning_rate": 1e-06, + "loss": 1.0726, + "mean_token_accuracy": 0.682645320892334, + "num_tokens": 70797638.0, + "step": 2776 + }, + { + "epoch": 0.3049637601581375, + "grad_norm": 2.684981107711792, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7257019877433777, + "num_tokens": 70816875.0, + "step": 2777 + }, + { + "epoch": 0.3050735778607512, + "grad_norm": 2.2625973224639893, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7035075426101685, + "num_tokens": 70841123.0, + "step": 2778 + }, + { + "epoch": 0.3051833955633648, + "grad_norm": 1.9731881618499756, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7027769088745117, + "num_tokens": 70870641.0, + "step": 2779 + }, + { + "epoch": 0.30529321326597847, + "grad_norm": 1.9189141988754272, + "learning_rate": 1e-06, + "loss": 1.0876, + "mean_token_accuracy": 0.6810458898544312, + "num_tokens": 70904607.0, + "step": 2780 + }, + { + "epoch": 0.3054030309685921, + "grad_norm": 2.1256141662597656, + "learning_rate": 1e-06, + "loss": 1.053, + "mean_token_accuracy": 0.6807062029838562, + "num_tokens": 70932377.0, + "step": 2781 + }, + { + "epoch": 0.3055128486712058, + "grad_norm": 2.233654737472534, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6899402141571045, + "num_tokens": 70959217.0, + "step": 2782 + }, + { + "epoch": 0.30562266637381946, + "grad_norm": 2.382497787475586, + "learning_rate": 1e-06, + "loss": 1.0495, + "mean_token_accuracy": 0.6859983801841736, + "num_tokens": 70982917.0, + "step": 2783 + }, + { + "epoch": 0.3057324840764331, + "grad_norm": 2.03078556060791, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7153318524360657, + "num_tokens": 71009370.0, + "step": 2784 + }, + { + "epoch": 0.30584230177904675, + "grad_norm": 2.0313093662261963, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7115245461463928, + "num_tokens": 71038318.0, + "step": 2785 + }, + { + "epoch": 0.30595211948166046, + "grad_norm": 1.8996250629425049, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.7043393850326538, + "num_tokens": 71073480.0, + "step": 2786 + }, + { + "epoch": 0.3060619371842741, + "grad_norm": 2.2573060989379883, + "learning_rate": 1e-06, + "loss": 1.0326, + "mean_token_accuracy": 0.6869920492172241, + "num_tokens": 71098450.0, + "step": 2787 + }, + { + "epoch": 0.30617175488688775, + "grad_norm": 2.1849896907806396, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7264347076416016, + "num_tokens": 71124506.0, + "step": 2788 + }, + { + "epoch": 0.30628157258950145, + "grad_norm": 2.3077850341796875, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6962807178497314, + "num_tokens": 71149939.0, + "step": 2789 + }, + { + "epoch": 0.3063913902921151, + "grad_norm": 2.261538505554199, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7283235788345337, + "num_tokens": 71173544.0, + "step": 2790 + }, + { + "epoch": 0.30650120799472874, + "grad_norm": 2.471348524093628, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7285385131835938, + "num_tokens": 71194465.0, + "step": 2791 + }, + { + "epoch": 0.3066110256973424, + "grad_norm": 2.1955976486206055, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6948147416114807, + "num_tokens": 71219332.0, + "step": 2792 + }, + { + "epoch": 0.3067208433999561, + "grad_norm": 2.2016892433166504, + "learning_rate": 1e-06, + "loss": 1.0421, + "mean_token_accuracy": 0.690566897392273, + "num_tokens": 71246201.0, + "step": 2793 + }, + { + "epoch": 0.30683066110256974, + "grad_norm": 2.425053358078003, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7111690640449524, + "num_tokens": 71267202.0, + "step": 2794 + }, + { + "epoch": 0.3069404788051834, + "grad_norm": 2.429929256439209, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.7030611634254456, + "num_tokens": 71289367.0, + "step": 2795 + }, + { + "epoch": 0.3070502965077971, + "grad_norm": 2.3353590965270996, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.7014667987823486, + "num_tokens": 71314710.0, + "step": 2796 + }, + { + "epoch": 0.30716011421041073, + "grad_norm": 2.2545690536499023, + "learning_rate": 1e-06, + "loss": 1.0685, + "mean_token_accuracy": 0.6813632845878601, + "num_tokens": 71341613.0, + "step": 2797 + }, + { + "epoch": 0.3072699319130244, + "grad_norm": 2.1781537532806396, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.7102408409118652, + "num_tokens": 71369448.0, + "step": 2798 + }, + { + "epoch": 0.307379749615638, + "grad_norm": 2.3395678997039795, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.7041651606559753, + "num_tokens": 71393550.0, + "step": 2799 + }, + { + "epoch": 0.3074895673182517, + "grad_norm": 2.1815927028656006, + "learning_rate": 1e-06, + "loss": 1.0805, + "mean_token_accuracy": 0.6899253726005554, + "num_tokens": 71422206.0, + "step": 2800 + }, + { + "epoch": 0.30759938502086537, + "grad_norm": 2.0600154399871826, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.697606086730957, + "num_tokens": 71453491.0, + "step": 2801 + }, + { + "epoch": 0.307709202723479, + "grad_norm": 2.1520681381225586, + "learning_rate": 1e-06, + "loss": 1.0668, + "mean_token_accuracy": 0.6778907775878906, + "num_tokens": 71481666.0, + "step": 2802 + }, + { + "epoch": 0.30781902042609266, + "grad_norm": 2.2950222492218018, + "learning_rate": 1e-06, + "loss": 1.1417, + "mean_token_accuracy": 0.6651310920715332, + "num_tokens": 71507998.0, + "step": 2803 + }, + { + "epoch": 0.30792883812870636, + "grad_norm": 2.0273284912109375, + "learning_rate": 1e-06, + "loss": 1.0724, + "mean_token_accuracy": 0.6811710000038147, + "num_tokens": 71538492.0, + "step": 2804 + }, + { + "epoch": 0.30803865583132, + "grad_norm": 2.4926939010620117, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7029013633728027, + "num_tokens": 71558558.0, + "step": 2805 + }, + { + "epoch": 0.30814847353393365, + "grad_norm": 2.0352375507354736, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.7014737725257874, + "num_tokens": 71588452.0, + "step": 2806 + }, + { + "epoch": 0.30825829123654735, + "grad_norm": 2.3470726013183594, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7238878011703491, + "num_tokens": 71610559.0, + "step": 2807 + }, + { + "epoch": 0.308368108939161, + "grad_norm": 2.058089017868042, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7281997799873352, + "num_tokens": 71637448.0, + "step": 2808 + }, + { + "epoch": 0.30847792664177465, + "grad_norm": 2.443009614944458, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6975101232528687, + "num_tokens": 71658122.0, + "step": 2809 + }, + { + "epoch": 0.3085877443443883, + "grad_norm": 2.2233643531799316, + "learning_rate": 1e-06, + "loss": 1.0313, + "mean_token_accuracy": 0.6927422881126404, + "num_tokens": 71683540.0, + "step": 2810 + }, + { + "epoch": 0.308697562047002, + "grad_norm": 2.1368961334228516, + "learning_rate": 1e-06, + "loss": 1.0486, + "mean_token_accuracy": 0.6878158450126648, + "num_tokens": 71711108.0, + "step": 2811 + }, + { + "epoch": 0.30880737974961564, + "grad_norm": 2.138138771057129, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7113282680511475, + "num_tokens": 71738871.0, + "step": 2812 + }, + { + "epoch": 0.3089171974522293, + "grad_norm": 2.1884047985076904, + "learning_rate": 1e-06, + "loss": 1.0712, + "mean_token_accuracy": 0.6769224405288696, + "num_tokens": 71770320.0, + "step": 2813 + }, + { + "epoch": 0.309027015154843, + "grad_norm": 1.9251868724822998, + "learning_rate": 1e-06, + "loss": 1.0923, + "mean_token_accuracy": 0.6745426058769226, + "num_tokens": 71805447.0, + "step": 2814 + }, + { + "epoch": 0.30913683285745663, + "grad_norm": 2.283637285232544, + "learning_rate": 1e-06, + "loss": 1.0803, + "mean_token_accuracy": 0.6798913478851318, + "num_tokens": 71829760.0, + "step": 2815 + }, + { + "epoch": 0.3092466505600703, + "grad_norm": 2.38572359085083, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.7053477764129639, + "num_tokens": 71855563.0, + "step": 2816 + }, + { + "epoch": 0.3093564682626839, + "grad_norm": 2.1895065307617188, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.7010993361473083, + "num_tokens": 71883235.0, + "step": 2817 + }, + { + "epoch": 0.3094662859652976, + "grad_norm": 2.4106605052948, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.6991031765937805, + "num_tokens": 71905098.0, + "step": 2818 + }, + { + "epoch": 0.3095761036679113, + "grad_norm": 2.3903443813323975, + "learning_rate": 1e-06, + "loss": 1.0813, + "mean_token_accuracy": 0.6809482574462891, + "num_tokens": 71929366.0, + "step": 2819 + }, + { + "epoch": 0.3096859213705249, + "grad_norm": 2.3173041343688965, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6939027309417725, + "num_tokens": 71953344.0, + "step": 2820 + }, + { + "epoch": 0.30979573907313857, + "grad_norm": 1.949440836906433, + "learning_rate": 1e-06, + "loss": 1.0492, + "mean_token_accuracy": 0.6880329847335815, + "num_tokens": 71987008.0, + "step": 2821 + }, + { + "epoch": 0.30990555677575227, + "grad_norm": 2.230257511138916, + "learning_rate": 1e-06, + "loss": 1.0636, + "mean_token_accuracy": 0.6771559119224548, + "num_tokens": 72013556.0, + "step": 2822 + }, + { + "epoch": 0.3100153744783659, + "grad_norm": 2.3396036624908447, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.714981734752655, + "num_tokens": 72036153.0, + "step": 2823 + }, + { + "epoch": 0.31012519218097956, + "grad_norm": 2.1392290592193604, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6973041296005249, + "num_tokens": 72061829.0, + "step": 2824 + }, + { + "epoch": 0.31023500988359326, + "grad_norm": 2.1673808097839355, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7062148451805115, + "num_tokens": 72089891.0, + "step": 2825 + }, + { + "epoch": 0.3103448275862069, + "grad_norm": 2.0192604064941406, + "learning_rate": 1e-06, + "loss": 1.0772, + "mean_token_accuracy": 0.673477053642273, + "num_tokens": 72121729.0, + "step": 2826 + }, + { + "epoch": 0.31045464528882055, + "grad_norm": 2.57517147064209, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7209755778312683, + "num_tokens": 72138468.0, + "step": 2827 + }, + { + "epoch": 0.3105644629914342, + "grad_norm": 2.1862833499908447, + "learning_rate": 1e-06, + "loss": 1.0626, + "mean_token_accuracy": 0.6835713386535645, + "num_tokens": 72166229.0, + "step": 2828 + }, + { + "epoch": 0.3106742806940479, + "grad_norm": 2.246029853820801, + "learning_rate": 1e-06, + "loss": 1.0425, + "mean_token_accuracy": 0.6891050338745117, + "num_tokens": 72190751.0, + "step": 2829 + }, + { + "epoch": 0.31078409839666155, + "grad_norm": 2.0920095443725586, + "learning_rate": 1e-06, + "loss": 1.0635, + "mean_token_accuracy": 0.6808620691299438, + "num_tokens": 72219608.0, + "step": 2830 + }, + { + "epoch": 0.3108939160992752, + "grad_norm": 2.272432565689087, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7086060643196106, + "num_tokens": 72244524.0, + "step": 2831 + }, + { + "epoch": 0.31100373380188884, + "grad_norm": 2.2091238498687744, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.7071372270584106, + "num_tokens": 72269396.0, + "step": 2832 + }, + { + "epoch": 0.31111355150450254, + "grad_norm": 2.2881977558135986, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6954325437545776, + "num_tokens": 72294940.0, + "step": 2833 + }, + { + "epoch": 0.3112233692071162, + "grad_norm": 2.3709282875061035, + "learning_rate": 1e-06, + "loss": 1.0306, + "mean_token_accuracy": 0.6864199638366699, + "num_tokens": 72319752.0, + "step": 2834 + }, + { + "epoch": 0.31133318690972983, + "grad_norm": 2.2897794246673584, + "learning_rate": 1e-06, + "loss": 1.0544, + "mean_token_accuracy": 0.678871214389801, + "num_tokens": 72344581.0, + "step": 2835 + }, + { + "epoch": 0.31144300461234353, + "grad_norm": 2.1737475395202637, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7143881320953369, + "num_tokens": 72370664.0, + "step": 2836 + }, + { + "epoch": 0.3115528223149572, + "grad_norm": 2.321204423904419, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.6940827369689941, + "num_tokens": 72395705.0, + "step": 2837 + }, + { + "epoch": 0.3116626400175708, + "grad_norm": 2.2903871536254883, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6922374963760376, + "num_tokens": 72420699.0, + "step": 2838 + }, + { + "epoch": 0.31177245772018447, + "grad_norm": 2.2095448970794678, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6957888007164001, + "num_tokens": 72446388.0, + "step": 2839 + }, + { + "epoch": 0.31188227542279817, + "grad_norm": 2.3266794681549072, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6967329978942871, + "num_tokens": 72468625.0, + "step": 2840 + }, + { + "epoch": 0.3119920931254118, + "grad_norm": 2.211660146713257, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.7036357522010803, + "num_tokens": 72493365.0, + "step": 2841 + }, + { + "epoch": 0.31210191082802546, + "grad_norm": 2.2142446041107178, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6983511447906494, + "num_tokens": 72520009.0, + "step": 2842 + }, + { + "epoch": 0.31221172853063917, + "grad_norm": 1.8566001653671265, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6936284303665161, + "num_tokens": 72553465.0, + "step": 2843 + }, + { + "epoch": 0.3123215462332528, + "grad_norm": 2.367354393005371, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6995819807052612, + "num_tokens": 72576386.0, + "step": 2844 + }, + { + "epoch": 0.31243136393586646, + "grad_norm": 2.1792244911193848, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.69291090965271, + "num_tokens": 72604686.0, + "step": 2845 + }, + { + "epoch": 0.3125411816384801, + "grad_norm": 2.208024501800537, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7115832567214966, + "num_tokens": 72628996.0, + "step": 2846 + }, + { + "epoch": 0.3126509993410938, + "grad_norm": 2.541231155395508, + "learning_rate": 1e-06, + "loss": 1.0322, + "mean_token_accuracy": 0.7006988525390625, + "num_tokens": 72649646.0, + "step": 2847 + }, + { + "epoch": 0.31276081704370745, + "grad_norm": 2.122854709625244, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7044691443443298, + "num_tokens": 72676916.0, + "step": 2848 + }, + { + "epoch": 0.3128706347463211, + "grad_norm": 2.2330193519592285, + "learning_rate": 1e-06, + "loss": 1.056, + "mean_token_accuracy": 0.6854828596115112, + "num_tokens": 72703315.0, + "step": 2849 + }, + { + "epoch": 0.31298045244893474, + "grad_norm": 2.2856953144073486, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6977990865707397, + "num_tokens": 72727600.0, + "step": 2850 + }, + { + "epoch": 0.31309027015154844, + "grad_norm": 2.2301127910614014, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7038338780403137, + "num_tokens": 72753071.0, + "step": 2851 + }, + { + "epoch": 0.3132000878541621, + "grad_norm": 1.980042815208435, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.7009342908859253, + "num_tokens": 72781279.0, + "step": 2852 + }, + { + "epoch": 0.31330990555677574, + "grad_norm": 2.331761121749878, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.6859035491943359, + "num_tokens": 72804981.0, + "step": 2853 + }, + { + "epoch": 0.31341972325938944, + "grad_norm": 2.143690347671509, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.6908683776855469, + "num_tokens": 72833813.0, + "step": 2854 + }, + { + "epoch": 0.3135295409620031, + "grad_norm": 2.0600426197052, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7152988314628601, + "num_tokens": 72863735.0, + "step": 2855 + }, + { + "epoch": 0.31363935866461673, + "grad_norm": 2.097076416015625, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6988096237182617, + "num_tokens": 72894370.0, + "step": 2856 + }, + { + "epoch": 0.3137491763672304, + "grad_norm": 2.282463550567627, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6940162181854248, + "num_tokens": 72919049.0, + "step": 2857 + }, + { + "epoch": 0.3138589940698441, + "grad_norm": 1.8999100923538208, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7180408239364624, + "num_tokens": 72950581.0, + "step": 2858 + }, + { + "epoch": 0.3139688117724577, + "grad_norm": 2.358869791030884, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7266573905944824, + "num_tokens": 72974756.0, + "step": 2859 + }, + { + "epoch": 0.31407862947507137, + "grad_norm": 2.322371482849121, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7072204351425171, + "num_tokens": 72997423.0, + "step": 2860 + }, + { + "epoch": 0.314188447177685, + "grad_norm": 2.0604183673858643, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.6944259405136108, + "num_tokens": 73027914.0, + "step": 2861 + }, + { + "epoch": 0.3142982648802987, + "grad_norm": 2.3580353260040283, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7059015035629272, + "num_tokens": 73050854.0, + "step": 2862 + }, + { + "epoch": 0.31440808258291236, + "grad_norm": 2.272491216659546, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6962356567382812, + "num_tokens": 73075872.0, + "step": 2863 + }, + { + "epoch": 0.314517900285526, + "grad_norm": 2.3045363426208496, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6875944137573242, + "num_tokens": 73100632.0, + "step": 2864 + }, + { + "epoch": 0.3146277179881397, + "grad_norm": 2.267754554748535, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7141565084457397, + "num_tokens": 73125674.0, + "step": 2865 + }, + { + "epoch": 0.31473753569075336, + "grad_norm": 2.284799575805664, + "learning_rate": 1e-06, + "loss": 1.0413, + "mean_token_accuracy": 0.680646538734436, + "num_tokens": 73150486.0, + "step": 2866 + }, + { + "epoch": 0.314847353393367, + "grad_norm": 2.336545705795288, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7153642773628235, + "num_tokens": 73173814.0, + "step": 2867 + }, + { + "epoch": 0.31495717109598065, + "grad_norm": 2.226806163787842, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6959528923034668, + "num_tokens": 73197118.0, + "step": 2868 + }, + { + "epoch": 0.31506698879859435, + "grad_norm": 2.610868215560913, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7023055553436279, + "num_tokens": 73217858.0, + "step": 2869 + }, + { + "epoch": 0.315176806501208, + "grad_norm": 2.2031784057617188, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6910587549209595, + "num_tokens": 73241796.0, + "step": 2870 + }, + { + "epoch": 0.31528662420382164, + "grad_norm": 1.8997069597244263, + "learning_rate": 1e-06, + "loss": 1.0838, + "mean_token_accuracy": 0.6787828207015991, + "num_tokens": 73276442.0, + "step": 2871 + }, + { + "epoch": 0.31539644190643534, + "grad_norm": 2.2297749519348145, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6898747086524963, + "num_tokens": 73302555.0, + "step": 2872 + }, + { + "epoch": 0.315506259609049, + "grad_norm": 2.1397037506103516, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7083735466003418, + "num_tokens": 73328585.0, + "step": 2873 + }, + { + "epoch": 0.31561607731166264, + "grad_norm": 2.070206642150879, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7179388999938965, + "num_tokens": 73356443.0, + "step": 2874 + }, + { + "epoch": 0.3157258950142763, + "grad_norm": 2.313213348388672, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7053691148757935, + "num_tokens": 73380304.0, + "step": 2875 + }, + { + "epoch": 0.31583571271689, + "grad_norm": 2.1593751907348633, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6949610114097595, + "num_tokens": 73405964.0, + "step": 2876 + }, + { + "epoch": 0.31594553041950363, + "grad_norm": 2.350848913192749, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6959192752838135, + "num_tokens": 73429823.0, + "step": 2877 + }, + { + "epoch": 0.3160553481221173, + "grad_norm": 2.183248519897461, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6897591352462769, + "num_tokens": 73457824.0, + "step": 2878 + }, + { + "epoch": 0.3161651658247309, + "grad_norm": 2.3668506145477295, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.723541259765625, + "num_tokens": 73481054.0, + "step": 2879 + }, + { + "epoch": 0.3162749835273446, + "grad_norm": 2.3739655017852783, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7286632657051086, + "num_tokens": 73502958.0, + "step": 2880 + }, + { + "epoch": 0.31638480122995827, + "grad_norm": 2.18900728225708, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7177353501319885, + "num_tokens": 73528090.0, + "step": 2881 + }, + { + "epoch": 0.3164946189325719, + "grad_norm": 2.1403095722198486, + "learning_rate": 1e-06, + "loss": 1.0736, + "mean_token_accuracy": 0.6763607263565063, + "num_tokens": 73557831.0, + "step": 2882 + }, + { + "epoch": 0.3166044366351856, + "grad_norm": 2.326246500015259, + "learning_rate": 1e-06, + "loss": 1.084, + "mean_token_accuracy": 0.6745043992996216, + "num_tokens": 73582355.0, + "step": 2883 + }, + { + "epoch": 0.31671425433779926, + "grad_norm": 2.319007396697998, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7167676687240601, + "num_tokens": 73605971.0, + "step": 2884 + }, + { + "epoch": 0.3168240720404129, + "grad_norm": 2.339463710784912, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.720165491104126, + "num_tokens": 73627642.0, + "step": 2885 + }, + { + "epoch": 0.31693388974302655, + "grad_norm": 2.138709783554077, + "learning_rate": 1e-06, + "loss": 1.0235, + "mean_token_accuracy": 0.6927840709686279, + "num_tokens": 73658997.0, + "step": 2886 + }, + { + "epoch": 0.31704370744564025, + "grad_norm": 2.244699001312256, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.7008925676345825, + "num_tokens": 73684466.0, + "step": 2887 + }, + { + "epoch": 0.3171535251482539, + "grad_norm": 2.294916868209839, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7094032764434814, + "num_tokens": 73709829.0, + "step": 2888 + }, + { + "epoch": 0.31726334285086755, + "grad_norm": 2.3135969638824463, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.691089391708374, + "num_tokens": 73733516.0, + "step": 2889 + }, + { + "epoch": 0.31737316055348125, + "grad_norm": 2.0131468772888184, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6927066445350647, + "num_tokens": 73764264.0, + "step": 2890 + }, + { + "epoch": 0.3174829782560949, + "grad_norm": 2.129626750946045, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.7002348899841309, + "num_tokens": 73790551.0, + "step": 2891 + }, + { + "epoch": 0.31759279595870854, + "grad_norm": 2.639995813369751, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.7066581845283508, + "num_tokens": 73810852.0, + "step": 2892 + }, + { + "epoch": 0.3177026136613222, + "grad_norm": 2.299346923828125, + "learning_rate": 1e-06, + "loss": 1.0316, + "mean_token_accuracy": 0.6848804950714111, + "num_tokens": 73839262.0, + "step": 2893 + }, + { + "epoch": 0.3178124313639359, + "grad_norm": 2.1346426010131836, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.7074968814849854, + "num_tokens": 73865086.0, + "step": 2894 + }, + { + "epoch": 0.31792224906654953, + "grad_norm": 2.3873748779296875, + "learning_rate": 1e-06, + "loss": 1.0525, + "mean_token_accuracy": 0.6860138177871704, + "num_tokens": 73891326.0, + "step": 2895 + }, + { + "epoch": 0.3180320667691632, + "grad_norm": 2.170931816101074, + "learning_rate": 1e-06, + "loss": 1.1146, + "mean_token_accuracy": 0.6781004071235657, + "num_tokens": 73921186.0, + "step": 2896 + }, + { + "epoch": 0.3181418844717768, + "grad_norm": 2.090287446975708, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6852161884307861, + "num_tokens": 73949131.0, + "step": 2897 + }, + { + "epoch": 0.3182517021743905, + "grad_norm": 2.149595022201538, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6961487531661987, + "num_tokens": 73974414.0, + "step": 2898 + }, + { + "epoch": 0.3183615198770042, + "grad_norm": 2.1984548568725586, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7153194546699524, + "num_tokens": 74000111.0, + "step": 2899 + }, + { + "epoch": 0.3184713375796178, + "grad_norm": 2.390190362930298, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.7035225033760071, + "num_tokens": 74023136.0, + "step": 2900 + }, + { + "epoch": 0.3185811552822315, + "grad_norm": 2.226086378097534, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7074536085128784, + "num_tokens": 74050220.0, + "step": 2901 + }, + { + "epoch": 0.31869097298484517, + "grad_norm": 2.2066028118133545, + "learning_rate": 1e-06, + "loss": 1.0356, + "mean_token_accuracy": 0.687863290309906, + "num_tokens": 74077314.0, + "step": 2902 + }, + { + "epoch": 0.3188007906874588, + "grad_norm": 2.232468605041504, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6948993802070618, + "num_tokens": 74103893.0, + "step": 2903 + }, + { + "epoch": 0.31891060839007246, + "grad_norm": 2.949810028076172, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6831741333007812, + "num_tokens": 74122383.0, + "step": 2904 + }, + { + "epoch": 0.31902042609268616, + "grad_norm": 2.442721128463745, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7008035182952881, + "num_tokens": 74145293.0, + "step": 2905 + }, + { + "epoch": 0.3191302437952998, + "grad_norm": 2.0244569778442383, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6959728002548218, + "num_tokens": 74173568.0, + "step": 2906 + }, + { + "epoch": 0.31924006149791345, + "grad_norm": 2.250354290008545, + "learning_rate": 1e-06, + "loss": 1.1096, + "mean_token_accuracy": 0.6640596985816956, + "num_tokens": 74202332.0, + "step": 2907 + }, + { + "epoch": 0.3193498792005271, + "grad_norm": 2.3331685066223145, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7215043306350708, + "num_tokens": 74226374.0, + "step": 2908 + }, + { + "epoch": 0.3194596969031408, + "grad_norm": 2.475928544998169, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.703887939453125, + "num_tokens": 74248172.0, + "step": 2909 + }, + { + "epoch": 0.31956951460575445, + "grad_norm": 2.386183023452759, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6987558007240295, + "num_tokens": 74271654.0, + "step": 2910 + }, + { + "epoch": 0.3196793323083681, + "grad_norm": 2.209007740020752, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7137740850448608, + "num_tokens": 74294785.0, + "step": 2911 + }, + { + "epoch": 0.3197891500109818, + "grad_norm": 2.372879981994629, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7074270248413086, + "num_tokens": 74317858.0, + "step": 2912 + }, + { + "epoch": 0.31989896771359544, + "grad_norm": 2.2366104125976562, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7083256840705872, + "num_tokens": 74342803.0, + "step": 2913 + }, + { + "epoch": 0.3200087854162091, + "grad_norm": 2.217315673828125, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.7002218961715698, + "num_tokens": 74370976.0, + "step": 2914 + }, + { + "epoch": 0.32011860311882273, + "grad_norm": 2.022146463394165, + "learning_rate": 1e-06, + "loss": 1.0926, + "mean_token_accuracy": 0.6737684011459351, + "num_tokens": 74400773.0, + "step": 2915 + }, + { + "epoch": 0.32022842082143643, + "grad_norm": 2.516218900680542, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7167956829071045, + "num_tokens": 74421340.0, + "step": 2916 + }, + { + "epoch": 0.3203382385240501, + "grad_norm": 2.338831901550293, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.708288311958313, + "num_tokens": 74445046.0, + "step": 2917 + }, + { + "epoch": 0.3204480562266637, + "grad_norm": 2.160034656524658, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7043156623840332, + "num_tokens": 74472602.0, + "step": 2918 + }, + { + "epoch": 0.3205578739292774, + "grad_norm": 2.025040626525879, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7214903831481934, + "num_tokens": 74499217.0, + "step": 2919 + }, + { + "epoch": 0.32066769163189107, + "grad_norm": 2.183093547821045, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6895966529846191, + "num_tokens": 74524960.0, + "step": 2920 + }, + { + "epoch": 0.3207775093345047, + "grad_norm": 2.4623324871063232, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6995318531990051, + "num_tokens": 74546021.0, + "step": 2921 + }, + { + "epoch": 0.32088732703711836, + "grad_norm": 1.9652308225631714, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7005870342254639, + "num_tokens": 74575981.0, + "step": 2922 + }, + { + "epoch": 0.32099714473973207, + "grad_norm": 2.005919933319092, + "learning_rate": 1e-06, + "loss": 1.052, + "mean_token_accuracy": 0.6946236491203308, + "num_tokens": 74607661.0, + "step": 2923 + }, + { + "epoch": 0.3211069624423457, + "grad_norm": 2.1870064735412598, + "learning_rate": 1e-06, + "loss": 1.0486, + "mean_token_accuracy": 0.6853514313697815, + "num_tokens": 74634551.0, + "step": 2924 + }, + { + "epoch": 0.32121678014495936, + "grad_norm": 2.0392940044403076, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.70338374376297, + "num_tokens": 74663796.0, + "step": 2925 + }, + { + "epoch": 0.321326597847573, + "grad_norm": 2.33427357673645, + "learning_rate": 1e-06, + "loss": 1.13, + "mean_token_accuracy": 0.6656098365783691, + "num_tokens": 74690956.0, + "step": 2926 + }, + { + "epoch": 0.3214364155501867, + "grad_norm": 2.5094072818756104, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7070823311805725, + "num_tokens": 74711001.0, + "step": 2927 + }, + { + "epoch": 0.32154623325280035, + "grad_norm": 2.1268320083618164, + "learning_rate": 1e-06, + "loss": 1.0968, + "mean_token_accuracy": 0.6745189428329468, + "num_tokens": 74740701.0, + "step": 2928 + }, + { + "epoch": 0.321656050955414, + "grad_norm": 2.418830633163452, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7270234823226929, + "num_tokens": 74764048.0, + "step": 2929 + }, + { + "epoch": 0.3217658686580277, + "grad_norm": 2.3651540279388428, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.7049672603607178, + "num_tokens": 74788024.0, + "step": 2930 + }, + { + "epoch": 0.32187568636064134, + "grad_norm": 2.302020788192749, + "learning_rate": 1e-06, + "loss": 1.0708, + "mean_token_accuracy": 0.6883442401885986, + "num_tokens": 74812536.0, + "step": 2931 + }, + { + "epoch": 0.321985504063255, + "grad_norm": 2.4580090045928955, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.6994144916534424, + "num_tokens": 74833123.0, + "step": 2932 + }, + { + "epoch": 0.32209532176586864, + "grad_norm": 2.6008522510528564, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7058818340301514, + "num_tokens": 74852334.0, + "step": 2933 + }, + { + "epoch": 0.32220513946848234, + "grad_norm": 2.2804830074310303, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7091891765594482, + "num_tokens": 74875913.0, + "step": 2934 + }, + { + "epoch": 0.322314957171096, + "grad_norm": 2.4490842819213867, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7115033268928528, + "num_tokens": 74896906.0, + "step": 2935 + }, + { + "epoch": 0.32242477487370963, + "grad_norm": 2.211291551589966, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.700429379940033, + "num_tokens": 74924777.0, + "step": 2936 + }, + { + "epoch": 0.3225345925763233, + "grad_norm": 2.0181028842926025, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6879768371582031, + "num_tokens": 74957731.0, + "step": 2937 + }, + { + "epoch": 0.322644410278937, + "grad_norm": 2.1942176818847656, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.696997880935669, + "num_tokens": 74984129.0, + "step": 2938 + }, + { + "epoch": 0.3227542279815506, + "grad_norm": 2.180954933166504, + "learning_rate": 1e-06, + "loss": 1.0614, + "mean_token_accuracy": 0.6821814775466919, + "num_tokens": 75013850.0, + "step": 2939 + }, + { + "epoch": 0.32286404568416427, + "grad_norm": 2.229408025741577, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7187552452087402, + "num_tokens": 75038095.0, + "step": 2940 + }, + { + "epoch": 0.32297386338677797, + "grad_norm": 2.356395959854126, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7144204378128052, + "num_tokens": 75060919.0, + "step": 2941 + }, + { + "epoch": 0.3230836810893916, + "grad_norm": 2.124289035797119, + "learning_rate": 1e-06, + "loss": 1.051, + "mean_token_accuracy": 0.6850963234901428, + "num_tokens": 75088692.0, + "step": 2942 + }, + { + "epoch": 0.32319349879200526, + "grad_norm": 2.128971815109253, + "learning_rate": 1e-06, + "loss": 1.1232, + "mean_token_accuracy": 0.6650660037994385, + "num_tokens": 75119825.0, + "step": 2943 + }, + { + "epoch": 0.3233033164946189, + "grad_norm": 2.521024227142334, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7053583860397339, + "num_tokens": 75140853.0, + "step": 2944 + }, + { + "epoch": 0.3234131341972326, + "grad_norm": 2.509493112564087, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6956315040588379, + "num_tokens": 75162483.0, + "step": 2945 + }, + { + "epoch": 0.32352295189984626, + "grad_norm": 2.106001138687134, + "learning_rate": 1e-06, + "loss": 1.0395, + "mean_token_accuracy": 0.6920617818832397, + "num_tokens": 75188158.0, + "step": 2946 + }, + { + "epoch": 0.3236327696024599, + "grad_norm": 2.1586430072784424, + "learning_rate": 1e-06, + "loss": 1.0628, + "mean_token_accuracy": 0.6836901903152466, + "num_tokens": 75218664.0, + "step": 2947 + }, + { + "epoch": 0.3237425873050736, + "grad_norm": 2.5900988578796387, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7035125494003296, + "num_tokens": 75238505.0, + "step": 2948 + }, + { + "epoch": 0.32385240500768725, + "grad_norm": 2.479619264602661, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7040752172470093, + "num_tokens": 75260684.0, + "step": 2949 + }, + { + "epoch": 0.3239622227103009, + "grad_norm": 2.2889373302459717, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.691885232925415, + "num_tokens": 75285603.0, + "step": 2950 + }, + { + "epoch": 0.32407204041291454, + "grad_norm": 2.513026714324951, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6859474182128906, + "num_tokens": 75306418.0, + "step": 2951 + }, + { + "epoch": 0.32418185811552824, + "grad_norm": 2.5660057067871094, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.6959542036056519, + "num_tokens": 75326902.0, + "step": 2952 + }, + { + "epoch": 0.3242916758181419, + "grad_norm": 2.12431001663208, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7063695192337036, + "num_tokens": 75356650.0, + "step": 2953 + }, + { + "epoch": 0.32440149352075554, + "grad_norm": 2.222451686859131, + "learning_rate": 1e-06, + "loss": 1.043, + "mean_token_accuracy": 0.680941104888916, + "num_tokens": 75384055.0, + "step": 2954 + }, + { + "epoch": 0.3245113112233692, + "grad_norm": 2.095700263977051, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7066766023635864, + "num_tokens": 75413752.0, + "step": 2955 + }, + { + "epoch": 0.3246211289259829, + "grad_norm": 2.1191256046295166, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7098386287689209, + "num_tokens": 75440959.0, + "step": 2956 + }, + { + "epoch": 0.32473094662859653, + "grad_norm": 1.9424675703048706, + "learning_rate": 1e-06, + "loss": 1.0668, + "mean_token_accuracy": 0.6888373494148254, + "num_tokens": 75470622.0, + "step": 2957 + }, + { + "epoch": 0.3248407643312102, + "grad_norm": 2.1614019870758057, + "learning_rate": 1e-06, + "loss": 1.0864, + "mean_token_accuracy": 0.6754241585731506, + "num_tokens": 75499634.0, + "step": 2958 + }, + { + "epoch": 0.3249505820338239, + "grad_norm": 2.4973697662353516, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7037010192871094, + "num_tokens": 75519476.0, + "step": 2959 + }, + { + "epoch": 0.3250603997364375, + "grad_norm": 2.2865614891052246, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7068606615066528, + "num_tokens": 75543308.0, + "step": 2960 + }, + { + "epoch": 0.32517021743905117, + "grad_norm": 2.6537952423095703, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7020536065101624, + "num_tokens": 75563257.0, + "step": 2961 + }, + { + "epoch": 0.3252800351416648, + "grad_norm": 1.9816844463348389, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6963782906532288, + "num_tokens": 75592932.0, + "step": 2962 + }, + { + "epoch": 0.3253898528442785, + "grad_norm": 2.1021552085876465, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.6980781555175781, + "num_tokens": 75619995.0, + "step": 2963 + }, + { + "epoch": 0.32549967054689216, + "grad_norm": 2.1026785373687744, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6885417103767395, + "num_tokens": 75648256.0, + "step": 2964 + }, + { + "epoch": 0.3256094882495058, + "grad_norm": 2.170340061187744, + "learning_rate": 1e-06, + "loss": 1.0566, + "mean_token_accuracy": 0.6792135238647461, + "num_tokens": 75676278.0, + "step": 2965 + }, + { + "epoch": 0.3257193059521195, + "grad_norm": 2.0129494667053223, + "learning_rate": 1e-06, + "loss": 1.0703, + "mean_token_accuracy": 0.6865679025650024, + "num_tokens": 75707432.0, + "step": 2966 + }, + { + "epoch": 0.32582912365473315, + "grad_norm": 2.1782023906707764, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7299415469169617, + "num_tokens": 75731961.0, + "step": 2967 + }, + { + "epoch": 0.3259389413573468, + "grad_norm": 2.2016520500183105, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7392126321792603, + "num_tokens": 75757201.0, + "step": 2968 + }, + { + "epoch": 0.32604875905996045, + "grad_norm": 2.2568819522857666, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7155046463012695, + "num_tokens": 75781743.0, + "step": 2969 + }, + { + "epoch": 0.32615857676257415, + "grad_norm": 2.1655867099761963, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.7056723833084106, + "num_tokens": 75808917.0, + "step": 2970 + }, + { + "epoch": 0.3262683944651878, + "grad_norm": 2.2104532718658447, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7230163812637329, + "num_tokens": 75832279.0, + "step": 2971 + }, + { + "epoch": 0.32637821216780144, + "grad_norm": 2.5086872577667236, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6962504386901855, + "num_tokens": 75852894.0, + "step": 2972 + }, + { + "epoch": 0.3264880298704151, + "grad_norm": 2.133234739303589, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.7090681195259094, + "num_tokens": 75879265.0, + "step": 2973 + }, + { + "epoch": 0.3265978475730288, + "grad_norm": 2.3198437690734863, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6954121589660645, + "num_tokens": 75901077.0, + "step": 2974 + }, + { + "epoch": 0.32670766527564243, + "grad_norm": 2.105548620223999, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.6976016163825989, + "num_tokens": 75928095.0, + "step": 2975 + }, + { + "epoch": 0.3268174829782561, + "grad_norm": 2.7498810291290283, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.691196084022522, + "num_tokens": 75948327.0, + "step": 2976 + }, + { + "epoch": 0.3269273006808698, + "grad_norm": 2.419058084487915, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.692181408405304, + "num_tokens": 75971217.0, + "step": 2977 + }, + { + "epoch": 0.3270371183834834, + "grad_norm": 2.3209149837493896, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.7006995677947998, + "num_tokens": 75997383.0, + "step": 2978 + }, + { + "epoch": 0.3271469360860971, + "grad_norm": 2.394672393798828, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.6965998411178589, + "num_tokens": 76018387.0, + "step": 2979 + }, + { + "epoch": 0.3272567537887107, + "grad_norm": 2.3991127014160156, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6895204782485962, + "num_tokens": 76041183.0, + "step": 2980 + }, + { + "epoch": 0.3273665714913244, + "grad_norm": 2.16933012008667, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6966540217399597, + "num_tokens": 76065939.0, + "step": 2981 + }, + { + "epoch": 0.32747638919393807, + "grad_norm": 2.1740241050720215, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.7026515007019043, + "num_tokens": 76092380.0, + "step": 2982 + }, + { + "epoch": 0.3275862068965517, + "grad_norm": 1.9783979654312134, + "learning_rate": 1e-06, + "loss": 1.0598, + "mean_token_accuracy": 0.68068528175354, + "num_tokens": 76125187.0, + "step": 2983 + }, + { + "epoch": 0.32769602459916536, + "grad_norm": 2.3229100704193115, + "learning_rate": 1e-06, + "loss": 1.0472, + "mean_token_accuracy": 0.6830629110336304, + "num_tokens": 76148804.0, + "step": 2984 + }, + { + "epoch": 0.32780584230177906, + "grad_norm": 2.225348711013794, + "learning_rate": 1e-06, + "loss": 1.0621, + "mean_token_accuracy": 0.6851456761360168, + "num_tokens": 76173556.0, + "step": 2985 + }, + { + "epoch": 0.3279156600043927, + "grad_norm": 2.2121872901916504, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7054308652877808, + "num_tokens": 76197562.0, + "step": 2986 + }, + { + "epoch": 0.32802547770700635, + "grad_norm": 1.9697076082229614, + "learning_rate": 1e-06, + "loss": 1.0585, + "mean_token_accuracy": 0.6846174597740173, + "num_tokens": 76230639.0, + "step": 2987 + }, + { + "epoch": 0.32813529540962005, + "grad_norm": 2.1895751953125, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.6995316743850708, + "num_tokens": 76255049.0, + "step": 2988 + }, + { + "epoch": 0.3282451131122337, + "grad_norm": 2.4186112880706787, + "learning_rate": 1e-06, + "loss": 1.0479, + "mean_token_accuracy": 0.6867955327033997, + "num_tokens": 76277833.0, + "step": 2989 + }, + { + "epoch": 0.32835493081484735, + "grad_norm": 2.1807973384857178, + "learning_rate": 1e-06, + "loss": 1.0423, + "mean_token_accuracy": 0.6918387413024902, + "num_tokens": 76302749.0, + "step": 2990 + }, + { + "epoch": 0.328464748517461, + "grad_norm": 2.2787094116210938, + "learning_rate": 1e-06, + "loss": 1.0852, + "mean_token_accuracy": 0.6772637963294983, + "num_tokens": 76326713.0, + "step": 2991 + }, + { + "epoch": 0.3285745662200747, + "grad_norm": 2.2184269428253174, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.7063579559326172, + "num_tokens": 76351210.0, + "step": 2992 + }, + { + "epoch": 0.32868438392268834, + "grad_norm": 2.320211172103882, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7264739274978638, + "num_tokens": 76371805.0, + "step": 2993 + }, + { + "epoch": 0.328794201625302, + "grad_norm": 2.2116942405700684, + "learning_rate": 1e-06, + "loss": 1.0794, + "mean_token_accuracy": 0.6791232824325562, + "num_tokens": 76399516.0, + "step": 2994 + }, + { + "epoch": 0.3289040193279157, + "grad_norm": 2.2048466205596924, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7068724632263184, + "num_tokens": 76424930.0, + "step": 2995 + }, + { + "epoch": 0.32901383703052933, + "grad_norm": 2.138627052307129, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6910833120346069, + "num_tokens": 76452265.0, + "step": 2996 + }, + { + "epoch": 0.329123654733143, + "grad_norm": 2.14389967918396, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7018320560455322, + "num_tokens": 76479082.0, + "step": 2997 + }, + { + "epoch": 0.3292334724357566, + "grad_norm": 2.298722982406616, + "learning_rate": 1e-06, + "loss": 1.0198, + "mean_token_accuracy": 0.6901966333389282, + "num_tokens": 76502356.0, + "step": 2998 + }, + { + "epoch": 0.3293432901383703, + "grad_norm": 2.5800669193267822, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.6927008032798767, + "num_tokens": 76524111.0, + "step": 2999 + }, + { + "epoch": 0.32945310784098397, + "grad_norm": 2.7672958374023438, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7090115547180176, + "num_tokens": 76542857.0, + "step": 3000 + }, + { + "epoch": 0.3295629255435976, + "grad_norm": 2.2710988521575928, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7071307897567749, + "num_tokens": 76567105.0, + "step": 3001 + }, + { + "epoch": 0.32967274324621126, + "grad_norm": 2.0077195167541504, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6870424747467041, + "num_tokens": 76596935.0, + "step": 3002 + }, + { + "epoch": 0.32978256094882497, + "grad_norm": 2.280283212661743, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.713283121585846, + "num_tokens": 76622906.0, + "step": 3003 + }, + { + "epoch": 0.3298923786514386, + "grad_norm": 2.263519763946533, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6881883144378662, + "num_tokens": 76647422.0, + "step": 3004 + }, + { + "epoch": 0.33000219635405226, + "grad_norm": 2.086357593536377, + "learning_rate": 1e-06, + "loss": 1.068, + "mean_token_accuracy": 0.6939781904220581, + "num_tokens": 76674975.0, + "step": 3005 + }, + { + "epoch": 0.33011201405666596, + "grad_norm": 2.1300411224365234, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.6947672367095947, + "num_tokens": 76703737.0, + "step": 3006 + }, + { + "epoch": 0.3302218317592796, + "grad_norm": 2.1707894802093506, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6893297433853149, + "num_tokens": 76728903.0, + "step": 3007 + }, + { + "epoch": 0.33033164946189325, + "grad_norm": 2.1833252906799316, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.710052490234375, + "num_tokens": 76756076.0, + "step": 3008 + }, + { + "epoch": 0.3304414671645069, + "grad_norm": 2.4330639839172363, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7037264108657837, + "num_tokens": 76778183.0, + "step": 3009 + }, + { + "epoch": 0.3305512848671206, + "grad_norm": 2.0517611503601074, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6954894661903381, + "num_tokens": 76806804.0, + "step": 3010 + }, + { + "epoch": 0.33066110256973424, + "grad_norm": 2.1679539680480957, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6936790943145752, + "num_tokens": 76833836.0, + "step": 3011 + }, + { + "epoch": 0.3307709202723479, + "grad_norm": 2.243319034576416, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.6952770948410034, + "num_tokens": 76860773.0, + "step": 3012 + }, + { + "epoch": 0.33088073797496154, + "grad_norm": 2.48129940032959, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.6926953196525574, + "num_tokens": 76880766.0, + "step": 3013 + }, + { + "epoch": 0.33099055567757524, + "grad_norm": 2.457770586013794, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7030858993530273, + "num_tokens": 76902415.0, + "step": 3014 + }, + { + "epoch": 0.3311003733801889, + "grad_norm": 2.1430485248565674, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7204005718231201, + "num_tokens": 76927926.0, + "step": 3015 + }, + { + "epoch": 0.33121019108280253, + "grad_norm": 2.236588716506958, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6963624358177185, + "num_tokens": 76954215.0, + "step": 3016 + }, + { + "epoch": 0.33132000878541623, + "grad_norm": 2.4679062366485596, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7237284183502197, + "num_tokens": 76976832.0, + "step": 3017 + }, + { + "epoch": 0.3314298264880299, + "grad_norm": 2.239936113357544, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.7016528844833374, + "num_tokens": 77002986.0, + "step": 3018 + }, + { + "epoch": 0.3315396441906435, + "grad_norm": 2.150052547454834, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7109103202819824, + "num_tokens": 77030671.0, + "step": 3019 + }, + { + "epoch": 0.33164946189325717, + "grad_norm": 2.1669836044311523, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6932161450386047, + "num_tokens": 77058512.0, + "step": 3020 + }, + { + "epoch": 0.33175927959587087, + "grad_norm": 2.131458044052124, + "learning_rate": 1e-06, + "loss": 1.0254, + "mean_token_accuracy": 0.691336989402771, + "num_tokens": 77085989.0, + "step": 3021 + }, + { + "epoch": 0.3318690972984845, + "grad_norm": 2.2497899532318115, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7119517922401428, + "num_tokens": 77111154.0, + "step": 3022 + }, + { + "epoch": 0.33197891500109816, + "grad_norm": 2.2641284465789795, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.6893624067306519, + "num_tokens": 77136583.0, + "step": 3023 + }, + { + "epoch": 0.33208873270371186, + "grad_norm": 2.3514437675476074, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7032284140586853, + "num_tokens": 77162108.0, + "step": 3024 + }, + { + "epoch": 0.3321985504063255, + "grad_norm": 2.1825428009033203, + "learning_rate": 1e-06, + "loss": 1.0573, + "mean_token_accuracy": 0.6861175894737244, + "num_tokens": 77190451.0, + "step": 3025 + }, + { + "epoch": 0.33230836810893916, + "grad_norm": 2.0468027591705322, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7014762163162231, + "num_tokens": 77217354.0, + "step": 3026 + }, + { + "epoch": 0.3324181858115528, + "grad_norm": 2.2400689125061035, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6984042525291443, + "num_tokens": 77242841.0, + "step": 3027 + }, + { + "epoch": 0.3325280035141665, + "grad_norm": 2.1484463214874268, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7093825936317444, + "num_tokens": 77269572.0, + "step": 3028 + }, + { + "epoch": 0.33263782121678015, + "grad_norm": 1.920168399810791, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.7022831439971924, + "num_tokens": 77303506.0, + "step": 3029 + }, + { + "epoch": 0.3327476389193938, + "grad_norm": 2.19578218460083, + "learning_rate": 1e-06, + "loss": 1.0647, + "mean_token_accuracy": 0.6852547526359558, + "num_tokens": 77330621.0, + "step": 3030 + }, + { + "epoch": 0.33285745662200744, + "grad_norm": 2.30318284034729, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6996181607246399, + "num_tokens": 77354103.0, + "step": 3031 + }, + { + "epoch": 0.33296727432462114, + "grad_norm": 2.2328689098358154, + "learning_rate": 1e-06, + "loss": 1.0488, + "mean_token_accuracy": 0.6826189756393433, + "num_tokens": 77381572.0, + "step": 3032 + }, + { + "epoch": 0.3330770920272348, + "grad_norm": 2.0516273975372314, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.7012987732887268, + "num_tokens": 77411821.0, + "step": 3033 + }, + { + "epoch": 0.33318690972984844, + "grad_norm": 2.2587170600891113, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.6986637115478516, + "num_tokens": 77437024.0, + "step": 3034 + }, + { + "epoch": 0.33329672743246214, + "grad_norm": 2.084822416305542, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6942689418792725, + "num_tokens": 77465200.0, + "step": 3035 + }, + { + "epoch": 0.3334065451350758, + "grad_norm": 2.0931732654571533, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7108036279678345, + "num_tokens": 77493510.0, + "step": 3036 + }, + { + "epoch": 0.33351636283768943, + "grad_norm": 2.2658472061157227, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7056179046630859, + "num_tokens": 77518217.0, + "step": 3037 + }, + { + "epoch": 0.3336261805403031, + "grad_norm": 2.439764976501465, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6995654702186584, + "num_tokens": 77541243.0, + "step": 3038 + }, + { + "epoch": 0.3337359982429168, + "grad_norm": 2.4677276611328125, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.683510959148407, + "num_tokens": 77563164.0, + "step": 3039 + }, + { + "epoch": 0.3338458159455304, + "grad_norm": 2.1020877361297607, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7009720206260681, + "num_tokens": 77593358.0, + "step": 3040 + }, + { + "epoch": 0.33395563364814407, + "grad_norm": 1.9873361587524414, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.6998340487480164, + "num_tokens": 77625225.0, + "step": 3041 + }, + { + "epoch": 0.33406545135075777, + "grad_norm": 2.100308418273926, + "learning_rate": 1e-06, + "loss": 1.0995, + "mean_token_accuracy": 0.670417845249176, + "num_tokens": 77653361.0, + "step": 3042 + }, + { + "epoch": 0.3341752690533714, + "grad_norm": 2.339723825454712, + "learning_rate": 1e-06, + "loss": 1.0633, + "mean_token_accuracy": 0.6842397451400757, + "num_tokens": 77680250.0, + "step": 3043 + }, + { + "epoch": 0.33428508675598506, + "grad_norm": 2.023837089538574, + "learning_rate": 1e-06, + "loss": 1.0445, + "mean_token_accuracy": 0.6873432993888855, + "num_tokens": 77712176.0, + "step": 3044 + }, + { + "epoch": 0.3343949044585987, + "grad_norm": 2.3482093811035156, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6973326206207275, + "num_tokens": 77735517.0, + "step": 3045 + }, + { + "epoch": 0.3345047221612124, + "grad_norm": 2.726517677307129, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.6963421702384949, + "num_tokens": 77754105.0, + "step": 3046 + }, + { + "epoch": 0.33461453986382605, + "grad_norm": 2.2486820220947266, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.6921882629394531, + "num_tokens": 77783234.0, + "step": 3047 + }, + { + "epoch": 0.3347243575664397, + "grad_norm": 2.387209177017212, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7251087427139282, + "num_tokens": 77805563.0, + "step": 3048 + }, + { + "epoch": 0.33483417526905335, + "grad_norm": 2.19020676612854, + "learning_rate": 1e-06, + "loss": 1.0531, + "mean_token_accuracy": 0.676570475101471, + "num_tokens": 77830547.0, + "step": 3049 + }, + { + "epoch": 0.33494399297166705, + "grad_norm": 1.988895058631897, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6891356706619263, + "num_tokens": 77862047.0, + "step": 3050 + }, + { + "epoch": 0.3350538106742807, + "grad_norm": 2.3883073329925537, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7248073220252991, + "num_tokens": 77883573.0, + "step": 3051 + }, + { + "epoch": 0.33516362837689434, + "grad_norm": 2.462372303009033, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.6899311542510986, + "num_tokens": 77907315.0, + "step": 3052 + }, + { + "epoch": 0.33527344607950804, + "grad_norm": 2.1382150650024414, + "learning_rate": 1e-06, + "loss": 1.0737, + "mean_token_accuracy": 0.6839607954025269, + "num_tokens": 77935126.0, + "step": 3053 + }, + { + "epoch": 0.3353832637821217, + "grad_norm": 2.274444341659546, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6982098817825317, + "num_tokens": 77958297.0, + "step": 3054 + }, + { + "epoch": 0.33549308148473533, + "grad_norm": 2.1691794395446777, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7157158851623535, + "num_tokens": 77983882.0, + "step": 3055 + }, + { + "epoch": 0.335602899187349, + "grad_norm": 2.422497510910034, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7201019525527954, + "num_tokens": 78008982.0, + "step": 3056 + }, + { + "epoch": 0.3357127168899627, + "grad_norm": 2.0910656452178955, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7127290368080139, + "num_tokens": 78036084.0, + "step": 3057 + }, + { + "epoch": 0.3358225345925763, + "grad_norm": 2.0355966091156006, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.6911423206329346, + "num_tokens": 78066252.0, + "step": 3058 + }, + { + "epoch": 0.33593235229519, + "grad_norm": 2.3338637351989746, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7096616625785828, + "num_tokens": 78089476.0, + "step": 3059 + }, + { + "epoch": 0.3360421699978036, + "grad_norm": 2.2178051471710205, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6925629377365112, + "num_tokens": 78114163.0, + "step": 3060 + }, + { + "epoch": 0.3361519877004173, + "grad_norm": 2.5316219329833984, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7158089876174927, + "num_tokens": 78135758.0, + "step": 3061 + }, + { + "epoch": 0.33626180540303097, + "grad_norm": 1.9938056468963623, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7065843343734741, + "num_tokens": 78165165.0, + "step": 3062 + }, + { + "epoch": 0.3363716231056446, + "grad_norm": 2.389131546020508, + "learning_rate": 1e-06, + "loss": 1.0608, + "mean_token_accuracy": 0.680919349193573, + "num_tokens": 78188764.0, + "step": 3063 + }, + { + "epoch": 0.3364814408082583, + "grad_norm": 2.272573471069336, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.68500155210495, + "num_tokens": 78213154.0, + "step": 3064 + }, + { + "epoch": 0.33659125851087196, + "grad_norm": 1.9425357580184937, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.6844078898429871, + "num_tokens": 78243949.0, + "step": 3065 + }, + { + "epoch": 0.3367010762134856, + "grad_norm": 2.0109000205993652, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6968202590942383, + "num_tokens": 78273792.0, + "step": 3066 + }, + { + "epoch": 0.33681089391609925, + "grad_norm": 2.208564281463623, + "learning_rate": 1e-06, + "loss": 1.0451, + "mean_token_accuracy": 0.6873644590377808, + "num_tokens": 78298704.0, + "step": 3067 + }, + { + "epoch": 0.33692071161871295, + "grad_norm": 2.3873138427734375, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7090094089508057, + "num_tokens": 78320026.0, + "step": 3068 + }, + { + "epoch": 0.3370305293213266, + "grad_norm": 2.181511163711548, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7189445495605469, + "num_tokens": 78346679.0, + "step": 3069 + }, + { + "epoch": 0.33714034702394025, + "grad_norm": 2.2929694652557373, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7252719402313232, + "num_tokens": 78369357.0, + "step": 3070 + }, + { + "epoch": 0.33725016472655395, + "grad_norm": 2.323843479156494, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6892716884613037, + "num_tokens": 78393629.0, + "step": 3071 + }, + { + "epoch": 0.3373599824291676, + "grad_norm": 2.0852153301239014, + "learning_rate": 1e-06, + "loss": 1.1025, + "mean_token_accuracy": 0.6677690744400024, + "num_tokens": 78423921.0, + "step": 3072 + }, + { + "epoch": 0.33746980013178124, + "grad_norm": 2.156592845916748, + "learning_rate": 1e-06, + "loss": 1.1005, + "mean_token_accuracy": 0.6778237819671631, + "num_tokens": 78451092.0, + "step": 3073 + }, + { + "epoch": 0.3375796178343949, + "grad_norm": 2.734502077102661, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7185056209564209, + "num_tokens": 78471213.0, + "step": 3074 + }, + { + "epoch": 0.3376894355370086, + "grad_norm": 1.9737114906311035, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7184220552444458, + "num_tokens": 78500546.0, + "step": 3075 + }, + { + "epoch": 0.33779925323962223, + "grad_norm": 2.150189161300659, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6925793886184692, + "num_tokens": 78526364.0, + "step": 3076 + }, + { + "epoch": 0.3379090709422359, + "grad_norm": 2.3032259941101074, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6933841705322266, + "num_tokens": 78554701.0, + "step": 3077 + }, + { + "epoch": 0.3380188886448495, + "grad_norm": 2.2229223251342773, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.6980853080749512, + "num_tokens": 78581425.0, + "step": 3078 + }, + { + "epoch": 0.3381287063474632, + "grad_norm": 2.2922115325927734, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7181311249732971, + "num_tokens": 78603539.0, + "step": 3079 + }, + { + "epoch": 0.33823852405007687, + "grad_norm": 2.1580312252044678, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.698158860206604, + "num_tokens": 78630837.0, + "step": 3080 + }, + { + "epoch": 0.3383483417526905, + "grad_norm": 2.2710273265838623, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6895909309387207, + "num_tokens": 78656479.0, + "step": 3081 + }, + { + "epoch": 0.3384581594553042, + "grad_norm": 2.578305244445801, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6958763599395752, + "num_tokens": 78676455.0, + "step": 3082 + }, + { + "epoch": 0.33856797715791787, + "grad_norm": 2.0987966060638428, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7114342451095581, + "num_tokens": 78706084.0, + "step": 3083 + }, + { + "epoch": 0.3386777948605315, + "grad_norm": 2.0344762802124023, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7060532569885254, + "num_tokens": 78735998.0, + "step": 3084 + }, + { + "epoch": 0.33878761256314516, + "grad_norm": 2.3169915676116943, + "learning_rate": 1e-06, + "loss": 0.9913, + "mean_token_accuracy": 0.70421302318573, + "num_tokens": 78759014.0, + "step": 3085 + }, + { + "epoch": 0.33889743026575886, + "grad_norm": 2.150043249130249, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7016342878341675, + "num_tokens": 78786093.0, + "step": 3086 + }, + { + "epoch": 0.3390072479683725, + "grad_norm": 2.4220166206359863, + "learning_rate": 1e-06, + "loss": 1.0781, + "mean_token_accuracy": 0.6829594373703003, + "num_tokens": 78808732.0, + "step": 3087 + }, + { + "epoch": 0.33911706567098615, + "grad_norm": 2.2189855575561523, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7181648015975952, + "num_tokens": 78834385.0, + "step": 3088 + }, + { + "epoch": 0.3392268833735998, + "grad_norm": 2.220533609390259, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7042839527130127, + "num_tokens": 78860521.0, + "step": 3089 + }, + { + "epoch": 0.3393367010762135, + "grad_norm": 2.4685280323028564, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.69051194190979, + "num_tokens": 78880715.0, + "step": 3090 + }, + { + "epoch": 0.33944651877882714, + "grad_norm": 2.232219696044922, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7146037817001343, + "num_tokens": 78906701.0, + "step": 3091 + }, + { + "epoch": 0.3395563364814408, + "grad_norm": 2.266561269760132, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6925737857818604, + "num_tokens": 78929966.0, + "step": 3092 + }, + { + "epoch": 0.3396661541840545, + "grad_norm": 2.1935274600982666, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.693446159362793, + "num_tokens": 78955595.0, + "step": 3093 + }, + { + "epoch": 0.33977597188666814, + "grad_norm": 2.0925121307373047, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.7097997665405273, + "num_tokens": 78982334.0, + "step": 3094 + }, + { + "epoch": 0.3398857895892818, + "grad_norm": 2.4690942764282227, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.700875461101532, + "num_tokens": 79002724.0, + "step": 3095 + }, + { + "epoch": 0.33999560729189543, + "grad_norm": 2.2590701580047607, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.6966065168380737, + "num_tokens": 79027549.0, + "step": 3096 + }, + { + "epoch": 0.34010542499450913, + "grad_norm": 2.478774070739746, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6958844065666199, + "num_tokens": 79051394.0, + "step": 3097 + }, + { + "epoch": 0.3402152426971228, + "grad_norm": 2.226890802383423, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6922587752342224, + "num_tokens": 79076089.0, + "step": 3098 + }, + { + "epoch": 0.3403250603997364, + "grad_norm": 1.965400218963623, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7076319456100464, + "num_tokens": 79109491.0, + "step": 3099 + }, + { + "epoch": 0.3404348781023501, + "grad_norm": 2.0109126567840576, + "learning_rate": 1e-06, + "loss": 1.1574, + "mean_token_accuracy": 0.6540035009384155, + "num_tokens": 79141649.0, + "step": 3100 + }, + { + "epoch": 0.34054469580496377, + "grad_norm": 2.356647253036499, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7115529775619507, + "num_tokens": 79164603.0, + "step": 3101 + }, + { + "epoch": 0.3406545135075774, + "grad_norm": 2.3034017086029053, + "learning_rate": 1e-06, + "loss": 1.054, + "mean_token_accuracy": 0.6838307976722717, + "num_tokens": 79187915.0, + "step": 3102 + }, + { + "epoch": 0.34076433121019106, + "grad_norm": 2.275766134262085, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6918323040008545, + "num_tokens": 79212926.0, + "step": 3103 + }, + { + "epoch": 0.34087414891280476, + "grad_norm": 2.2331459522247314, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.6978940367698669, + "num_tokens": 79237043.0, + "step": 3104 + }, + { + "epoch": 0.3409839666154184, + "grad_norm": 2.213182210922241, + "learning_rate": 1e-06, + "loss": 1.0224, + "mean_token_accuracy": 0.6940160393714905, + "num_tokens": 79262769.0, + "step": 3105 + }, + { + "epoch": 0.34109378431803206, + "grad_norm": 2.4500977993011475, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7221164107322693, + "num_tokens": 79283624.0, + "step": 3106 + }, + { + "epoch": 0.3412036020206457, + "grad_norm": 2.315277338027954, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6966150999069214, + "num_tokens": 79308240.0, + "step": 3107 + }, + { + "epoch": 0.3413134197232594, + "grad_norm": 2.3535945415496826, + "learning_rate": 1e-06, + "loss": 1.0773, + "mean_token_accuracy": 0.677527666091919, + "num_tokens": 79334956.0, + "step": 3108 + }, + { + "epoch": 0.34142323742587305, + "grad_norm": 2.288224458694458, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.6902831792831421, + "num_tokens": 79360751.0, + "step": 3109 + }, + { + "epoch": 0.3415330551284867, + "grad_norm": 2.012774705886841, + "learning_rate": 1e-06, + "loss": 1.1194, + "mean_token_accuracy": 0.6655200719833374, + "num_tokens": 79392290.0, + "step": 3110 + }, + { + "epoch": 0.3416428728311004, + "grad_norm": 2.3100006580352783, + "learning_rate": 1e-06, + "loss": 1.1044, + "mean_token_accuracy": 0.6699320077896118, + "num_tokens": 79416136.0, + "step": 3111 + }, + { + "epoch": 0.34175269053371404, + "grad_norm": 2.361194133758545, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7229270935058594, + "num_tokens": 79437825.0, + "step": 3112 + }, + { + "epoch": 0.3418625082363277, + "grad_norm": 2.1506597995758057, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.6873509287834167, + "num_tokens": 79466689.0, + "step": 3113 + }, + { + "epoch": 0.34197232593894134, + "grad_norm": 2.5931739807128906, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7243880033493042, + "num_tokens": 79485223.0, + "step": 3114 + }, + { + "epoch": 0.34208214364155504, + "grad_norm": 2.317310333251953, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7098749876022339, + "num_tokens": 79506302.0, + "step": 3115 + }, + { + "epoch": 0.3421919613441687, + "grad_norm": 2.1009156703948975, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6898735165596008, + "num_tokens": 79533445.0, + "step": 3116 + }, + { + "epoch": 0.34230177904678233, + "grad_norm": 2.0449020862579346, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6877723932266235, + "num_tokens": 79561371.0, + "step": 3117 + }, + { + "epoch": 0.34241159674939603, + "grad_norm": 2.1627073287963867, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7089942693710327, + "num_tokens": 79586710.0, + "step": 3118 + }, + { + "epoch": 0.3425214144520097, + "grad_norm": 2.1703386306762695, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7001269459724426, + "num_tokens": 79612234.0, + "step": 3119 + }, + { + "epoch": 0.3426312321546233, + "grad_norm": 2.1647355556488037, + "learning_rate": 1e-06, + "loss": 1.0677, + "mean_token_accuracy": 0.6818020343780518, + "num_tokens": 79639187.0, + "step": 3120 + }, + { + "epoch": 0.34274104985723697, + "grad_norm": 1.949537992477417, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7074311971664429, + "num_tokens": 79670137.0, + "step": 3121 + }, + { + "epoch": 0.34285086755985067, + "grad_norm": 2.3127481937408447, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.6977376937866211, + "num_tokens": 79695311.0, + "step": 3122 + }, + { + "epoch": 0.3429606852624643, + "grad_norm": 1.9559581279754639, + "learning_rate": 1e-06, + "loss": 1.0674, + "mean_token_accuracy": 0.6777087450027466, + "num_tokens": 79726820.0, + "step": 3123 + }, + { + "epoch": 0.34307050296507796, + "grad_norm": 2.375483512878418, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7083655595779419, + "num_tokens": 79750566.0, + "step": 3124 + }, + { + "epoch": 0.3431803206676916, + "grad_norm": 1.9985463619232178, + "learning_rate": 1e-06, + "loss": 1.0399, + "mean_token_accuracy": 0.691201388835907, + "num_tokens": 79781984.0, + "step": 3125 + }, + { + "epoch": 0.3432901383703053, + "grad_norm": 2.4129745960235596, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6975420713424683, + "num_tokens": 79805752.0, + "step": 3126 + }, + { + "epoch": 0.34339995607291895, + "grad_norm": 2.5891988277435303, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7096966505050659, + "num_tokens": 79825783.0, + "step": 3127 + }, + { + "epoch": 0.3435097737755326, + "grad_norm": 2.0990233421325684, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.6952159404754639, + "num_tokens": 79854787.0, + "step": 3128 + }, + { + "epoch": 0.3436195914781463, + "grad_norm": 2.352545738220215, + "learning_rate": 1e-06, + "loss": 1.0234, + "mean_token_accuracy": 0.7012203931808472, + "num_tokens": 79877888.0, + "step": 3129 + }, + { + "epoch": 0.34372940918075995, + "grad_norm": 2.0527732372283936, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.6970048546791077, + "num_tokens": 79905382.0, + "step": 3130 + }, + { + "epoch": 0.3438392268833736, + "grad_norm": 2.427461624145508, + "learning_rate": 1e-06, + "loss": 1.1011, + "mean_token_accuracy": 0.6713627576828003, + "num_tokens": 79929072.0, + "step": 3131 + }, + { + "epoch": 0.34394904458598724, + "grad_norm": 2.2485246658325195, + "learning_rate": 1e-06, + "loss": 1.066, + "mean_token_accuracy": 0.6839120388031006, + "num_tokens": 79955580.0, + "step": 3132 + }, + { + "epoch": 0.34405886228860094, + "grad_norm": 2.274810552597046, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.703181803226471, + "num_tokens": 79978984.0, + "step": 3133 + }, + { + "epoch": 0.3441686799912146, + "grad_norm": 2.1335270404815674, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.701439380645752, + "num_tokens": 80005614.0, + "step": 3134 + }, + { + "epoch": 0.34427849769382823, + "grad_norm": 2.465150833129883, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.731263279914856, + "num_tokens": 80025056.0, + "step": 3135 + }, + { + "epoch": 0.3443883153964419, + "grad_norm": 2.4021389484405518, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.6958539485931396, + "num_tokens": 80047923.0, + "step": 3136 + }, + { + "epoch": 0.3444981330990556, + "grad_norm": 2.5750067234039307, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7027814388275146, + "num_tokens": 80069087.0, + "step": 3137 + }, + { + "epoch": 0.3446079508016692, + "grad_norm": 2.4831631183624268, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7098953723907471, + "num_tokens": 80089585.0, + "step": 3138 + }, + { + "epoch": 0.3447177685042829, + "grad_norm": 2.029625177383423, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.6909044981002808, + "num_tokens": 80117906.0, + "step": 3139 + }, + { + "epoch": 0.3448275862068966, + "grad_norm": 2.0815839767456055, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7063621878623962, + "num_tokens": 80145941.0, + "step": 3140 + }, + { + "epoch": 0.3449374039095102, + "grad_norm": 2.3432371616363525, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6886031627655029, + "num_tokens": 80169978.0, + "step": 3141 + }, + { + "epoch": 0.34504722161212387, + "grad_norm": 2.197542667388916, + "learning_rate": 1e-06, + "loss": 1.1072, + "mean_token_accuracy": 0.6651889085769653, + "num_tokens": 80197624.0, + "step": 3142 + }, + { + "epoch": 0.3451570393147375, + "grad_norm": 2.1095235347747803, + "learning_rate": 1e-06, + "loss": 1.0293, + "mean_token_accuracy": 0.6859544515609741, + "num_tokens": 80224417.0, + "step": 3143 + }, + { + "epoch": 0.3452668570173512, + "grad_norm": 2.3448944091796875, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7313414812088013, + "num_tokens": 80246089.0, + "step": 3144 + }, + { + "epoch": 0.34537667471996486, + "grad_norm": 2.4047915935516357, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7151776552200317, + "num_tokens": 80267238.0, + "step": 3145 + }, + { + "epoch": 0.3454864924225785, + "grad_norm": 2.225651264190674, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7242757678031921, + "num_tokens": 80292022.0, + "step": 3146 + }, + { + "epoch": 0.3455963101251922, + "grad_norm": 2.104261875152588, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7192026376724243, + "num_tokens": 80318084.0, + "step": 3147 + }, + { + "epoch": 0.34570612782780585, + "grad_norm": 2.547976016998291, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7331578731536865, + "num_tokens": 80337205.0, + "step": 3148 + }, + { + "epoch": 0.3458159455304195, + "grad_norm": 2.125652551651001, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6967246532440186, + "num_tokens": 80364749.0, + "step": 3149 + }, + { + "epoch": 0.34592576323303315, + "grad_norm": 2.2442781925201416, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7043086290359497, + "num_tokens": 80389908.0, + "step": 3150 + }, + { + "epoch": 0.34603558093564685, + "grad_norm": 2.3644161224365234, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6990388631820679, + "num_tokens": 80413617.0, + "step": 3151 + }, + { + "epoch": 0.3461453986382605, + "grad_norm": 2.27152419090271, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7013123035430908, + "num_tokens": 80439649.0, + "step": 3152 + }, + { + "epoch": 0.34625521634087414, + "grad_norm": 1.9949594736099243, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6905210018157959, + "num_tokens": 80470157.0, + "step": 3153 + }, + { + "epoch": 0.3463650340434878, + "grad_norm": 2.399169445037842, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7189162969589233, + "num_tokens": 80492974.0, + "step": 3154 + }, + { + "epoch": 0.3464748517461015, + "grad_norm": 2.2601511478424072, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7212575674057007, + "num_tokens": 80517935.0, + "step": 3155 + }, + { + "epoch": 0.34658466944871513, + "grad_norm": 2.2708640098571777, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.6858475804328918, + "num_tokens": 80542712.0, + "step": 3156 + }, + { + "epoch": 0.3466944871513288, + "grad_norm": 2.0479745864868164, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7137086987495422, + "num_tokens": 80571025.0, + "step": 3157 + }, + { + "epoch": 0.3468043048539425, + "grad_norm": 2.6176886558532715, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6934173107147217, + "num_tokens": 80590949.0, + "step": 3158 + }, + { + "epoch": 0.3469141225565561, + "grad_norm": 2.0532386302948, + "learning_rate": 1e-06, + "loss": 1.0376, + "mean_token_accuracy": 0.6908758878707886, + "num_tokens": 80621441.0, + "step": 3159 + }, + { + "epoch": 0.34702394025916977, + "grad_norm": 2.0930161476135254, + "learning_rate": 1e-06, + "loss": 1.0698, + "mean_token_accuracy": 0.6806182265281677, + "num_tokens": 80651999.0, + "step": 3160 + }, + { + "epoch": 0.3471337579617834, + "grad_norm": 2.071706771850586, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7044384479522705, + "num_tokens": 80680910.0, + "step": 3161 + }, + { + "epoch": 0.3472435756643971, + "grad_norm": 2.1347312927246094, + "learning_rate": 1e-06, + "loss": 1.096, + "mean_token_accuracy": 0.6754002571105957, + "num_tokens": 80711769.0, + "step": 3162 + }, + { + "epoch": 0.34735339336701077, + "grad_norm": 1.9734011888504028, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6940197348594666, + "num_tokens": 80743880.0, + "step": 3163 + }, + { + "epoch": 0.3474632110696244, + "grad_norm": 2.3239245414733887, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.710861325263977, + "num_tokens": 80766650.0, + "step": 3164 + }, + { + "epoch": 0.34757302877223806, + "grad_norm": 1.8375413417816162, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.6997298002243042, + "num_tokens": 80801941.0, + "step": 3165 + }, + { + "epoch": 0.34768284647485176, + "grad_norm": 2.094649076461792, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7079726457595825, + "num_tokens": 80828101.0, + "step": 3166 + }, + { + "epoch": 0.3477926641774654, + "grad_norm": 2.227311134338379, + "learning_rate": 1e-06, + "loss": 1.0431, + "mean_token_accuracy": 0.6870481371879578, + "num_tokens": 80855811.0, + "step": 3167 + }, + { + "epoch": 0.34790248188007905, + "grad_norm": 2.507958173751831, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7135498523712158, + "num_tokens": 80876617.0, + "step": 3168 + }, + { + "epoch": 0.34801229958269275, + "grad_norm": 2.008251428604126, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7155411839485168, + "num_tokens": 80903348.0, + "step": 3169 + }, + { + "epoch": 0.3481221172853064, + "grad_norm": 2.5765905380249023, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7107399702072144, + "num_tokens": 80922273.0, + "step": 3170 + }, + { + "epoch": 0.34823193498792004, + "grad_norm": 2.3107337951660156, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7058565616607666, + "num_tokens": 80944488.0, + "step": 3171 + }, + { + "epoch": 0.3483417526905337, + "grad_norm": 2.326702117919922, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7130136489868164, + "num_tokens": 80968072.0, + "step": 3172 + }, + { + "epoch": 0.3484515703931474, + "grad_norm": 2.1364078521728516, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7231102585792542, + "num_tokens": 80995059.0, + "step": 3173 + }, + { + "epoch": 0.34856138809576104, + "grad_norm": 2.437901496887207, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7006722688674927, + "num_tokens": 81015547.0, + "step": 3174 + }, + { + "epoch": 0.3486712057983747, + "grad_norm": 2.2886180877685547, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7014002799987793, + "num_tokens": 81040037.0, + "step": 3175 + }, + { + "epoch": 0.3487810235009884, + "grad_norm": 2.1806466579437256, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7061436772346497, + "num_tokens": 81064235.0, + "step": 3176 + }, + { + "epoch": 0.34889084120360203, + "grad_norm": 2.345482587814331, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6954267024993896, + "num_tokens": 81087669.0, + "step": 3177 + }, + { + "epoch": 0.3490006589062157, + "grad_norm": 2.1518778800964355, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.698486328125, + "num_tokens": 81113216.0, + "step": 3178 + }, + { + "epoch": 0.3491104766088293, + "grad_norm": 2.496373414993286, + "learning_rate": 1e-06, + "loss": 1.0441, + "mean_token_accuracy": 0.6802664995193481, + "num_tokens": 81135126.0, + "step": 3179 + }, + { + "epoch": 0.349220294311443, + "grad_norm": 2.472304105758667, + "learning_rate": 1e-06, + "loss": 1.0515, + "mean_token_accuracy": 0.6881661415100098, + "num_tokens": 81158409.0, + "step": 3180 + }, + { + "epoch": 0.34933011201405667, + "grad_norm": 2.4105234146118164, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7044917941093445, + "num_tokens": 81179915.0, + "step": 3181 + }, + { + "epoch": 0.3494399297166703, + "grad_norm": 2.2785754203796387, + "learning_rate": 1e-06, + "loss": 1.0547, + "mean_token_accuracy": 0.6830509901046753, + "num_tokens": 81205915.0, + "step": 3182 + }, + { + "epoch": 0.34954974741928396, + "grad_norm": 2.2508068084716797, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.6928337812423706, + "num_tokens": 81232321.0, + "step": 3183 + }, + { + "epoch": 0.34965956512189766, + "grad_norm": 1.9127916097640991, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.6950343251228333, + "num_tokens": 81264961.0, + "step": 3184 + }, + { + "epoch": 0.3497693828245113, + "grad_norm": 2.385772466659546, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.6986459493637085, + "num_tokens": 81288629.0, + "step": 3185 + }, + { + "epoch": 0.34987920052712496, + "grad_norm": 2.5885398387908936, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7075730562210083, + "num_tokens": 81308308.0, + "step": 3186 + }, + { + "epoch": 0.34998901822973866, + "grad_norm": 2.3326778411865234, + "learning_rate": 1e-06, + "loss": 1.0313, + "mean_token_accuracy": 0.7014881372451782, + "num_tokens": 81333029.0, + "step": 3187 + }, + { + "epoch": 0.3500988359323523, + "grad_norm": 2.4891765117645264, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7039767503738403, + "num_tokens": 81354741.0, + "step": 3188 + }, + { + "epoch": 0.35020865363496595, + "grad_norm": 2.3685648441314697, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7158339619636536, + "num_tokens": 81377198.0, + "step": 3189 + }, + { + "epoch": 0.3503184713375796, + "grad_norm": 2.5096824169158936, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6988486051559448, + "num_tokens": 81397195.0, + "step": 3190 + }, + { + "epoch": 0.3504282890401933, + "grad_norm": 2.0830938816070557, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.7043092846870422, + "num_tokens": 81424782.0, + "step": 3191 + }, + { + "epoch": 0.35053810674280694, + "grad_norm": 2.2803940773010254, + "learning_rate": 1e-06, + "loss": 1.0413, + "mean_token_accuracy": 0.6940625905990601, + "num_tokens": 81448759.0, + "step": 3192 + }, + { + "epoch": 0.3506479244454206, + "grad_norm": 2.2819862365722656, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7030461430549622, + "num_tokens": 81473235.0, + "step": 3193 + }, + { + "epoch": 0.3507577421480343, + "grad_norm": 2.299379825592041, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6941303610801697, + "num_tokens": 81499589.0, + "step": 3194 + }, + { + "epoch": 0.35086755985064794, + "grad_norm": 2.18272066116333, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6915881633758545, + "num_tokens": 81527928.0, + "step": 3195 + }, + { + "epoch": 0.3509773775532616, + "grad_norm": 2.685157537460327, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7252891063690186, + "num_tokens": 81544907.0, + "step": 3196 + }, + { + "epoch": 0.35108719525587523, + "grad_norm": 2.401953935623169, + "learning_rate": 1e-06, + "loss": 1.0406, + "mean_token_accuracy": 0.6943053007125854, + "num_tokens": 81567380.0, + "step": 3197 + }, + { + "epoch": 0.35119701295848893, + "grad_norm": 2.237065076828003, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.7000364065170288, + "num_tokens": 81592973.0, + "step": 3198 + }, + { + "epoch": 0.3513068306611026, + "grad_norm": 2.2059035301208496, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6896249055862427, + "num_tokens": 81618119.0, + "step": 3199 + }, + { + "epoch": 0.3514166483637162, + "grad_norm": 2.1502976417541504, + "learning_rate": 1e-06, + "loss": 1.0556, + "mean_token_accuracy": 0.6895219087600708, + "num_tokens": 81645650.0, + "step": 3200 + }, + { + "epoch": 0.35152646606632987, + "grad_norm": 2.530107021331787, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7167482376098633, + "num_tokens": 81666439.0, + "step": 3201 + }, + { + "epoch": 0.35163628376894357, + "grad_norm": 2.14241361618042, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7102735042572021, + "num_tokens": 81693991.0, + "step": 3202 + }, + { + "epoch": 0.3517461014715572, + "grad_norm": 2.1143088340759277, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6844925284385681, + "num_tokens": 81722214.0, + "step": 3203 + }, + { + "epoch": 0.35185591917417086, + "grad_norm": 2.2226052284240723, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6896981596946716, + "num_tokens": 81747781.0, + "step": 3204 + }, + { + "epoch": 0.35196573687678456, + "grad_norm": 2.3486740589141846, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7097960710525513, + "num_tokens": 81772260.0, + "step": 3205 + }, + { + "epoch": 0.3520755545793982, + "grad_norm": 1.8905540704727173, + "learning_rate": 1e-06, + "loss": 1.0761, + "mean_token_accuracy": 0.6788771748542786, + "num_tokens": 81808218.0, + "step": 3206 + }, + { + "epoch": 0.35218537228201185, + "grad_norm": 2.2614548206329346, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.7054593563079834, + "num_tokens": 81832450.0, + "step": 3207 + }, + { + "epoch": 0.3522951899846255, + "grad_norm": 2.2044923305511475, + "learning_rate": 1e-06, + "loss": 1.0435, + "mean_token_accuracy": 0.6849696636199951, + "num_tokens": 81859103.0, + "step": 3208 + }, + { + "epoch": 0.3524050076872392, + "grad_norm": 1.989321231842041, + "learning_rate": 1e-06, + "loss": 1.0777, + "mean_token_accuracy": 0.683525025844574, + "num_tokens": 81889833.0, + "step": 3209 + }, + { + "epoch": 0.35251482538985285, + "grad_norm": 2.366063356399536, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7178959846496582, + "num_tokens": 81910045.0, + "step": 3210 + }, + { + "epoch": 0.3526246430924665, + "grad_norm": 2.236301898956299, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6946964859962463, + "num_tokens": 81934317.0, + "step": 3211 + }, + { + "epoch": 0.35273446079508014, + "grad_norm": 2.1693241596221924, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7092570662498474, + "num_tokens": 81959794.0, + "step": 3212 + }, + { + "epoch": 0.35284427849769384, + "grad_norm": 1.9391260147094727, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7021474242210388, + "num_tokens": 81990146.0, + "step": 3213 + }, + { + "epoch": 0.3529540962003075, + "grad_norm": 2.3154897689819336, + "learning_rate": 1e-06, + "loss": 1.0808, + "mean_token_accuracy": 0.6830143332481384, + "num_tokens": 82015604.0, + "step": 3214 + }, + { + "epoch": 0.35306391390292113, + "grad_norm": 2.309537887573242, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6934123635292053, + "num_tokens": 82038937.0, + "step": 3215 + }, + { + "epoch": 0.35317373160553484, + "grad_norm": 2.004082441329956, + "learning_rate": 1e-06, + "loss": 1.069, + "mean_token_accuracy": 0.6775304079055786, + "num_tokens": 82070695.0, + "step": 3216 + }, + { + "epoch": 0.3532835493081485, + "grad_norm": 2.319195508956909, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.703885555267334, + "num_tokens": 82095741.0, + "step": 3217 + }, + { + "epoch": 0.3533933670107621, + "grad_norm": 2.296022653579712, + "learning_rate": 1e-06, + "loss": 1.0244, + "mean_token_accuracy": 0.6953970193862915, + "num_tokens": 82119806.0, + "step": 3218 + }, + { + "epoch": 0.3535031847133758, + "grad_norm": 2.326018810272217, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.6950827836990356, + "num_tokens": 82141960.0, + "step": 3219 + }, + { + "epoch": 0.3536130024159895, + "grad_norm": 2.0304770469665527, + "learning_rate": 1e-06, + "loss": 1.039, + "mean_token_accuracy": 0.686150848865509, + "num_tokens": 82172094.0, + "step": 3220 + }, + { + "epoch": 0.3537228201186031, + "grad_norm": 2.356052875518799, + "learning_rate": 1e-06, + "loss": 1.0828, + "mean_token_accuracy": 0.6810625791549683, + "num_tokens": 82196214.0, + "step": 3221 + }, + { + "epoch": 0.35383263782121677, + "grad_norm": 2.1037611961364746, + "learning_rate": 1e-06, + "loss": 1.107, + "mean_token_accuracy": 0.6700757145881653, + "num_tokens": 82226222.0, + "step": 3222 + }, + { + "epoch": 0.35394245552383047, + "grad_norm": 2.376300573348999, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.6945509314537048, + "num_tokens": 82247701.0, + "step": 3223 + }, + { + "epoch": 0.3540522732264441, + "grad_norm": 2.373243808746338, + "learning_rate": 1e-06, + "loss": 1.0733, + "mean_token_accuracy": 0.6784613132476807, + "num_tokens": 82272524.0, + "step": 3224 + }, + { + "epoch": 0.35416209092905776, + "grad_norm": 2.4229726791381836, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7244877815246582, + "num_tokens": 82295039.0, + "step": 3225 + }, + { + "epoch": 0.3542719086316714, + "grad_norm": 2.4253673553466797, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7224087715148926, + "num_tokens": 82316247.0, + "step": 3226 + }, + { + "epoch": 0.3543817263342851, + "grad_norm": 1.9829018115997314, + "learning_rate": 1e-06, + "loss": 1.0735, + "mean_token_accuracy": 0.6865171790122986, + "num_tokens": 82347792.0, + "step": 3227 + }, + { + "epoch": 0.35449154403689875, + "grad_norm": 2.3086342811584473, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7145288586616516, + "num_tokens": 82370790.0, + "step": 3228 + }, + { + "epoch": 0.3546013617395124, + "grad_norm": 2.3494060039520264, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7383197546005249, + "num_tokens": 82391936.0, + "step": 3229 + }, + { + "epoch": 0.35471117944212605, + "grad_norm": 2.2406561374664307, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.7076107263565063, + "num_tokens": 82415685.0, + "step": 3230 + }, + { + "epoch": 0.35482099714473975, + "grad_norm": 2.579052686691284, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.7009162306785583, + "num_tokens": 82437171.0, + "step": 3231 + }, + { + "epoch": 0.3549308148473534, + "grad_norm": 2.1266000270843506, + "learning_rate": 1e-06, + "loss": 1.038, + "mean_token_accuracy": 0.6850367784500122, + "num_tokens": 82466594.0, + "step": 3232 + }, + { + "epoch": 0.35504063254996704, + "grad_norm": 2.2888026237487793, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7182016372680664, + "num_tokens": 82490346.0, + "step": 3233 + }, + { + "epoch": 0.35515045025258074, + "grad_norm": 2.464740753173828, + "learning_rate": 1e-06, + "loss": 1.0217, + "mean_token_accuracy": 0.6941198706626892, + "num_tokens": 82513424.0, + "step": 3234 + }, + { + "epoch": 0.3552602679551944, + "grad_norm": 2.2852206230163574, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.715864896774292, + "num_tokens": 82537884.0, + "step": 3235 + }, + { + "epoch": 0.35537008565780803, + "grad_norm": 2.127190351486206, + "learning_rate": 1e-06, + "loss": 1.0684, + "mean_token_accuracy": 0.6788151264190674, + "num_tokens": 82565331.0, + "step": 3236 + }, + { + "epoch": 0.3554799033604217, + "grad_norm": 2.1826515197753906, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6982815265655518, + "num_tokens": 82593145.0, + "step": 3237 + }, + { + "epoch": 0.3555897210630354, + "grad_norm": 2.0907626152038574, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7211748361587524, + "num_tokens": 82619733.0, + "step": 3238 + }, + { + "epoch": 0.355699538765649, + "grad_norm": 2.3572003841400146, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7037802934646606, + "num_tokens": 82642494.0, + "step": 3239 + }, + { + "epoch": 0.35580935646826267, + "grad_norm": 1.9744980335235596, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7088630199432373, + "num_tokens": 82675518.0, + "step": 3240 + }, + { + "epoch": 0.3559191741708763, + "grad_norm": 1.8614983558654785, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7162832617759705, + "num_tokens": 82707028.0, + "step": 3241 + }, + { + "epoch": 0.35602899187349, + "grad_norm": 2.287445545196533, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7093720436096191, + "num_tokens": 82730148.0, + "step": 3242 + }, + { + "epoch": 0.35613880957610367, + "grad_norm": 2.5343735218048096, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7043629288673401, + "num_tokens": 82751950.0, + "step": 3243 + }, + { + "epoch": 0.3562486272787173, + "grad_norm": 2.505474805831909, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.6986452341079712, + "num_tokens": 82773826.0, + "step": 3244 + }, + { + "epoch": 0.356358444981331, + "grad_norm": 2.017429828643799, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.6983619928359985, + "num_tokens": 82802699.0, + "step": 3245 + }, + { + "epoch": 0.35646826268394466, + "grad_norm": 2.5809824466705322, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6977347135543823, + "num_tokens": 82823398.0, + "step": 3246 + }, + { + "epoch": 0.3565780803865583, + "grad_norm": 2.034153699874878, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7133938670158386, + "num_tokens": 82853145.0, + "step": 3247 + }, + { + "epoch": 0.35668789808917195, + "grad_norm": 2.3678016662597656, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7145369052886963, + "num_tokens": 82873322.0, + "step": 3248 + }, + { + "epoch": 0.35679771579178565, + "grad_norm": 2.026517868041992, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7019819021224976, + "num_tokens": 82903536.0, + "step": 3249 + }, + { + "epoch": 0.3569075334943993, + "grad_norm": 2.256643533706665, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.696388840675354, + "num_tokens": 82929571.0, + "step": 3250 + }, + { + "epoch": 0.35701735119701294, + "grad_norm": 2.552067995071411, + "learning_rate": 1e-06, + "loss": 1.0438, + "mean_token_accuracy": 0.6910947561264038, + "num_tokens": 82949926.0, + "step": 3251 + }, + { + "epoch": 0.35712716889962665, + "grad_norm": 1.9659292697906494, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7143252491950989, + "num_tokens": 82979760.0, + "step": 3252 + }, + { + "epoch": 0.3572369866022403, + "grad_norm": 2.2100865840911865, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6949010491371155, + "num_tokens": 83004134.0, + "step": 3253 + }, + { + "epoch": 0.35734680430485394, + "grad_norm": 2.3863818645477295, + "learning_rate": 1e-06, + "loss": 1.0426, + "mean_token_accuracy": 0.6910713911056519, + "num_tokens": 83026877.0, + "step": 3254 + }, + { + "epoch": 0.3574566220074676, + "grad_norm": 2.338124990463257, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.7014676332473755, + "num_tokens": 83053332.0, + "step": 3255 + }, + { + "epoch": 0.3575664397100813, + "grad_norm": 2.205369234085083, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7039146423339844, + "num_tokens": 83079163.0, + "step": 3256 + }, + { + "epoch": 0.35767625741269493, + "grad_norm": 2.14571213722229, + "learning_rate": 1e-06, + "loss": 1.0597, + "mean_token_accuracy": 0.6914315223693848, + "num_tokens": 83108756.0, + "step": 3257 + }, + { + "epoch": 0.3577860751153086, + "grad_norm": 2.282787322998047, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.6979427933692932, + "num_tokens": 83132863.0, + "step": 3258 + }, + { + "epoch": 0.3578958928179222, + "grad_norm": 2.613152503967285, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.7033973932266235, + "num_tokens": 83152072.0, + "step": 3259 + }, + { + "epoch": 0.3580057105205359, + "grad_norm": 2.022460460662842, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7118340134620667, + "num_tokens": 83181260.0, + "step": 3260 + }, + { + "epoch": 0.35811552822314957, + "grad_norm": 2.400641679763794, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.704512894153595, + "num_tokens": 83202967.0, + "step": 3261 + }, + { + "epoch": 0.3582253459257632, + "grad_norm": 2.2618680000305176, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.6997728943824768, + "num_tokens": 83227782.0, + "step": 3262 + }, + { + "epoch": 0.3583351636283769, + "grad_norm": 2.336799383163452, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.7003498077392578, + "num_tokens": 83252108.0, + "step": 3263 + }, + { + "epoch": 0.35844498133099056, + "grad_norm": 2.532363176345825, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7176237106323242, + "num_tokens": 83273456.0, + "step": 3264 + }, + { + "epoch": 0.3585547990336042, + "grad_norm": 2.198917865753174, + "learning_rate": 1e-06, + "loss": 1.0551, + "mean_token_accuracy": 0.6797052621841431, + "num_tokens": 83300411.0, + "step": 3265 + }, + { + "epoch": 0.35866461673621786, + "grad_norm": 2.018822431564331, + "learning_rate": 1e-06, + "loss": 1.1086, + "mean_token_accuracy": 0.6740248203277588, + "num_tokens": 83332884.0, + "step": 3266 + }, + { + "epoch": 0.35877443443883156, + "grad_norm": 2.467357873916626, + "learning_rate": 1e-06, + "loss": 1.0421, + "mean_token_accuracy": 0.6812597513198853, + "num_tokens": 83355217.0, + "step": 3267 + }, + { + "epoch": 0.3588842521414452, + "grad_norm": 2.181962728500366, + "learning_rate": 1e-06, + "loss": 1.0654, + "mean_token_accuracy": 0.6782405376434326, + "num_tokens": 83383843.0, + "step": 3268 + }, + { + "epoch": 0.35899406984405885, + "grad_norm": 2.293462038040161, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.7025574445724487, + "num_tokens": 83407921.0, + "step": 3269 + }, + { + "epoch": 0.35910388754667255, + "grad_norm": 2.016728162765503, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6988911628723145, + "num_tokens": 83437495.0, + "step": 3270 + }, + { + "epoch": 0.3592137052492862, + "grad_norm": 2.430307149887085, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6937682628631592, + "num_tokens": 83460411.0, + "step": 3271 + }, + { + "epoch": 0.35932352295189984, + "grad_norm": 2.1812260150909424, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6924296617507935, + "num_tokens": 83485070.0, + "step": 3272 + }, + { + "epoch": 0.3594333406545135, + "grad_norm": 2.580488920211792, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.707179069519043, + "num_tokens": 83504216.0, + "step": 3273 + }, + { + "epoch": 0.3595431583571272, + "grad_norm": 2.1630661487579346, + "learning_rate": 1e-06, + "loss": 1.0542, + "mean_token_accuracy": 0.6848450899124146, + "num_tokens": 83531236.0, + "step": 3274 + }, + { + "epoch": 0.35965297605974084, + "grad_norm": 2.4237735271453857, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6958536505699158, + "num_tokens": 83553976.0, + "step": 3275 + }, + { + "epoch": 0.3597627937623545, + "grad_norm": 2.2117037773132324, + "learning_rate": 1e-06, + "loss": 1.0708, + "mean_token_accuracy": 0.6791444420814514, + "num_tokens": 83582480.0, + "step": 3276 + }, + { + "epoch": 0.35987261146496813, + "grad_norm": 2.4538166522979736, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6933853030204773, + "num_tokens": 83604078.0, + "step": 3277 + }, + { + "epoch": 0.35998242916758183, + "grad_norm": 2.0592265129089355, + "learning_rate": 1e-06, + "loss": 1.0588, + "mean_token_accuracy": 0.68776535987854, + "num_tokens": 83633080.0, + "step": 3278 + }, + { + "epoch": 0.3600922468701955, + "grad_norm": 2.1992807388305664, + "learning_rate": 1e-06, + "loss": 1.0967, + "mean_token_accuracy": 0.6792370080947876, + "num_tokens": 83661346.0, + "step": 3279 + }, + { + "epoch": 0.3602020645728091, + "grad_norm": 2.5578231811523438, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7075163125991821, + "num_tokens": 83681816.0, + "step": 3280 + }, + { + "epoch": 0.3603118822754228, + "grad_norm": 2.3548688888549805, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7073636054992676, + "num_tokens": 83704046.0, + "step": 3281 + }, + { + "epoch": 0.36042169997803647, + "grad_norm": 2.4006741046905518, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.710350751876831, + "num_tokens": 83726250.0, + "step": 3282 + }, + { + "epoch": 0.3605315176806501, + "grad_norm": 2.170480489730835, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.716704785823822, + "num_tokens": 83751182.0, + "step": 3283 + }, + { + "epoch": 0.36064133538326376, + "grad_norm": 2.102985143661499, + "learning_rate": 1e-06, + "loss": 1.1205, + "mean_token_accuracy": 0.6644508838653564, + "num_tokens": 83780450.0, + "step": 3284 + }, + { + "epoch": 0.36075115308587746, + "grad_norm": 2.4243004322052, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7111567258834839, + "num_tokens": 83802570.0, + "step": 3285 + }, + { + "epoch": 0.3608609707884911, + "grad_norm": 2.6083743572235107, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7072349190711975, + "num_tokens": 83821770.0, + "step": 3286 + }, + { + "epoch": 0.36097078849110475, + "grad_norm": 2.7465450763702393, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7080615758895874, + "num_tokens": 83839061.0, + "step": 3287 + }, + { + "epoch": 0.3610806061937184, + "grad_norm": 1.9311219453811646, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6818764209747314, + "num_tokens": 83869944.0, + "step": 3288 + }, + { + "epoch": 0.3611904238963321, + "grad_norm": 2.1428332328796387, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7109077572822571, + "num_tokens": 83894841.0, + "step": 3289 + }, + { + "epoch": 0.36130024159894575, + "grad_norm": 2.2678627967834473, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7280082702636719, + "num_tokens": 83917194.0, + "step": 3290 + }, + { + "epoch": 0.3614100593015594, + "grad_norm": 2.1414835453033447, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7064608931541443, + "num_tokens": 83944645.0, + "step": 3291 + }, + { + "epoch": 0.3615198770041731, + "grad_norm": 2.2701425552368164, + "learning_rate": 1e-06, + "loss": 1.0688, + "mean_token_accuracy": 0.6773251891136169, + "num_tokens": 83970473.0, + "step": 3292 + }, + { + "epoch": 0.36162969470678674, + "grad_norm": 2.4223132133483887, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.7050780057907104, + "num_tokens": 83994765.0, + "step": 3293 + }, + { + "epoch": 0.3617395124094004, + "grad_norm": 2.1900155544281006, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7002699375152588, + "num_tokens": 84022655.0, + "step": 3294 + }, + { + "epoch": 0.36184933011201403, + "grad_norm": 2.2608253955841064, + "learning_rate": 1e-06, + "loss": 1.1099, + "mean_token_accuracy": 0.6709502935409546, + "num_tokens": 84051744.0, + "step": 3295 + }, + { + "epoch": 0.36195914781462774, + "grad_norm": 2.0454986095428467, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7028446197509766, + "num_tokens": 84082019.0, + "step": 3296 + }, + { + "epoch": 0.3620689655172414, + "grad_norm": 1.9058654308319092, + "learning_rate": 1e-06, + "loss": 1.0605, + "mean_token_accuracy": 0.6811982989311218, + "num_tokens": 84118460.0, + "step": 3297 + }, + { + "epoch": 0.362178783219855, + "grad_norm": 2.1369009017944336, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.713981032371521, + "num_tokens": 84144499.0, + "step": 3298 + }, + { + "epoch": 0.36228860092246873, + "grad_norm": 1.8855938911437988, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6895003914833069, + "num_tokens": 84178120.0, + "step": 3299 + }, + { + "epoch": 0.3623984186250824, + "grad_norm": 2.1735687255859375, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7005281448364258, + "num_tokens": 84204143.0, + "step": 3300 + }, + { + "epoch": 0.362508236327696, + "grad_norm": 2.2078328132629395, + "learning_rate": 1e-06, + "loss": 1.094, + "mean_token_accuracy": 0.6771883964538574, + "num_tokens": 84234983.0, + "step": 3301 + }, + { + "epoch": 0.36261805403030967, + "grad_norm": 2.5946218967437744, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7247799634933472, + "num_tokens": 84255012.0, + "step": 3302 + }, + { + "epoch": 0.36272787173292337, + "grad_norm": 2.2651374340057373, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7155261039733887, + "num_tokens": 84279722.0, + "step": 3303 + }, + { + "epoch": 0.362837689435537, + "grad_norm": 2.3019161224365234, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.7067791223526001, + "num_tokens": 84305255.0, + "step": 3304 + }, + { + "epoch": 0.36294750713815066, + "grad_norm": 2.2379441261291504, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.716429591178894, + "num_tokens": 84328010.0, + "step": 3305 + }, + { + "epoch": 0.3630573248407643, + "grad_norm": 2.290287971496582, + "learning_rate": 1e-06, + "loss": 1.056, + "mean_token_accuracy": 0.6940820217132568, + "num_tokens": 84352778.0, + "step": 3306 + }, + { + "epoch": 0.363167142543378, + "grad_norm": 2.5778701305389404, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6967750787734985, + "num_tokens": 84376387.0, + "step": 3307 + }, + { + "epoch": 0.36327696024599165, + "grad_norm": 2.4600651264190674, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.7014885544776917, + "num_tokens": 84399950.0, + "step": 3308 + }, + { + "epoch": 0.3633867779486053, + "grad_norm": 2.5519275665283203, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6887743473052979, + "num_tokens": 84420271.0, + "step": 3309 + }, + { + "epoch": 0.363496595651219, + "grad_norm": 2.335360050201416, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.6983442902565002, + "num_tokens": 84443902.0, + "step": 3310 + }, + { + "epoch": 0.36360641335383265, + "grad_norm": 2.2499589920043945, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6969355344772339, + "num_tokens": 84468013.0, + "step": 3311 + }, + { + "epoch": 0.3637162310564463, + "grad_norm": 2.218120813369751, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7075124382972717, + "num_tokens": 84495454.0, + "step": 3312 + }, + { + "epoch": 0.36382604875905994, + "grad_norm": 2.171616554260254, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6993473172187805, + "num_tokens": 84523520.0, + "step": 3313 + }, + { + "epoch": 0.36393586646167364, + "grad_norm": 2.0229341983795166, + "learning_rate": 1e-06, + "loss": 1.0656, + "mean_token_accuracy": 0.6887802481651306, + "num_tokens": 84555680.0, + "step": 3314 + }, + { + "epoch": 0.3640456841642873, + "grad_norm": 1.9436098337173462, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6883975863456726, + "num_tokens": 84587180.0, + "step": 3315 + }, + { + "epoch": 0.36415550186690093, + "grad_norm": 2.38558030128479, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7040244340896606, + "num_tokens": 84609522.0, + "step": 3316 + }, + { + "epoch": 0.3642653195695146, + "grad_norm": 2.337376832962036, + "learning_rate": 1e-06, + "loss": 1.1006, + "mean_token_accuracy": 0.6701737642288208, + "num_tokens": 84634763.0, + "step": 3317 + }, + { + "epoch": 0.3643751372721283, + "grad_norm": 2.0435397624969482, + "learning_rate": 1e-06, + "loss": 1.0737, + "mean_token_accuracy": 0.6830998659133911, + "num_tokens": 84664034.0, + "step": 3318 + }, + { + "epoch": 0.3644849549747419, + "grad_norm": 1.964630365371704, + "learning_rate": 1e-06, + "loss": 1.0557, + "mean_token_accuracy": 0.6813734769821167, + "num_tokens": 84695020.0, + "step": 3319 + }, + { + "epoch": 0.36459477267735557, + "grad_norm": 2.3514134883880615, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7265281081199646, + "num_tokens": 84718762.0, + "step": 3320 + }, + { + "epoch": 0.3647045903799693, + "grad_norm": 2.3677306175231934, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.7019249200820923, + "num_tokens": 84743159.0, + "step": 3321 + }, + { + "epoch": 0.3648144080825829, + "grad_norm": 1.9394601583480835, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6968013644218445, + "num_tokens": 84775842.0, + "step": 3322 + }, + { + "epoch": 0.36492422578519657, + "grad_norm": 2.538668155670166, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7034316658973694, + "num_tokens": 84796577.0, + "step": 3323 + }, + { + "epoch": 0.3650340434878102, + "grad_norm": 2.248821258544922, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7081679105758667, + "num_tokens": 84822496.0, + "step": 3324 + }, + { + "epoch": 0.3651438611904239, + "grad_norm": 2.0855743885040283, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7099055647850037, + "num_tokens": 84849396.0, + "step": 3325 + }, + { + "epoch": 0.36525367889303756, + "grad_norm": 2.5807549953460693, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6960196495056152, + "num_tokens": 84870532.0, + "step": 3326 + }, + { + "epoch": 0.3653634965956512, + "grad_norm": 2.4987127780914307, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7064070701599121, + "num_tokens": 84891548.0, + "step": 3327 + }, + { + "epoch": 0.3654733142982649, + "grad_norm": 2.229240655899048, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7097968459129333, + "num_tokens": 84918217.0, + "step": 3328 + }, + { + "epoch": 0.36558313200087855, + "grad_norm": 2.416717767715454, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7087063193321228, + "num_tokens": 84941430.0, + "step": 3329 + }, + { + "epoch": 0.3656929497034922, + "grad_norm": 1.950790524482727, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7065685391426086, + "num_tokens": 84972832.0, + "step": 3330 + }, + { + "epoch": 0.36580276740610584, + "grad_norm": 2.241809129714966, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.694858729839325, + "num_tokens": 85001447.0, + "step": 3331 + }, + { + "epoch": 0.36591258510871955, + "grad_norm": 2.5249178409576416, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7107605934143066, + "num_tokens": 85020842.0, + "step": 3332 + }, + { + "epoch": 0.3660224028113332, + "grad_norm": 2.685535430908203, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6945412158966064, + "num_tokens": 85040663.0, + "step": 3333 + }, + { + "epoch": 0.36613222051394684, + "grad_norm": 1.9818860292434692, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7150838375091553, + "num_tokens": 85070386.0, + "step": 3334 + }, + { + "epoch": 0.3662420382165605, + "grad_norm": 2.071155548095703, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.717602550983429, + "num_tokens": 85098075.0, + "step": 3335 + }, + { + "epoch": 0.3663518559191742, + "grad_norm": 2.3778326511383057, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.698788583278656, + "num_tokens": 85122349.0, + "step": 3336 + }, + { + "epoch": 0.36646167362178783, + "grad_norm": 2.2784173488616943, + "learning_rate": 1e-06, + "loss": 1.0949, + "mean_token_accuracy": 0.6765972375869751, + "num_tokens": 85149905.0, + "step": 3337 + }, + { + "epoch": 0.3665714913244015, + "grad_norm": 2.062554359436035, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7030658721923828, + "num_tokens": 85177266.0, + "step": 3338 + }, + { + "epoch": 0.3666813090270152, + "grad_norm": 2.2451813220977783, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.6932722330093384, + "num_tokens": 85201483.0, + "step": 3339 + }, + { + "epoch": 0.3667911267296288, + "grad_norm": 2.379639148712158, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.7030014395713806, + "num_tokens": 85224185.0, + "step": 3340 + }, + { + "epoch": 0.36690094443224247, + "grad_norm": 2.309943675994873, + "learning_rate": 1e-06, + "loss": 1.0586, + "mean_token_accuracy": 0.6881603002548218, + "num_tokens": 85249427.0, + "step": 3341 + }, + { + "epoch": 0.3670107621348561, + "grad_norm": 2.0325236320495605, + "learning_rate": 1e-06, + "loss": 1.0513, + "mean_token_accuracy": 0.689953088760376, + "num_tokens": 85278906.0, + "step": 3342 + }, + { + "epoch": 0.3671205798374698, + "grad_norm": 2.800048828125, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7415587306022644, + "num_tokens": 85295261.0, + "step": 3343 + }, + { + "epoch": 0.36723039754008346, + "grad_norm": 2.4491264820098877, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6902077198028564, + "num_tokens": 85316564.0, + "step": 3344 + }, + { + "epoch": 0.3673402152426971, + "grad_norm": 2.2011399269104004, + "learning_rate": 1e-06, + "loss": 1.0959, + "mean_token_accuracy": 0.6700693368911743, + "num_tokens": 85344649.0, + "step": 3345 + }, + { + "epoch": 0.3674500329453108, + "grad_norm": 2.045671224594116, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.6983605027198792, + "num_tokens": 85372593.0, + "step": 3346 + }, + { + "epoch": 0.36755985064792446, + "grad_norm": 2.2332217693328857, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7036740779876709, + "num_tokens": 85399403.0, + "step": 3347 + }, + { + "epoch": 0.3676696683505381, + "grad_norm": 2.517559766769409, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7051829695701599, + "num_tokens": 85422441.0, + "step": 3348 + }, + { + "epoch": 0.36777948605315175, + "grad_norm": 2.6064257621765137, + "learning_rate": 1e-06, + "loss": 1.041, + "mean_token_accuracy": 0.68531334400177, + "num_tokens": 85442630.0, + "step": 3349 + }, + { + "epoch": 0.36788930375576545, + "grad_norm": 2.407320261001587, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7035785913467407, + "num_tokens": 85464392.0, + "step": 3350 + }, + { + "epoch": 0.3679991214583791, + "grad_norm": 2.276076078414917, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.698444128036499, + "num_tokens": 85488460.0, + "step": 3351 + }, + { + "epoch": 0.36810893916099274, + "grad_norm": 2.2867581844329834, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7075037956237793, + "num_tokens": 85513140.0, + "step": 3352 + }, + { + "epoch": 0.3682187568636064, + "grad_norm": 2.305565357208252, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7108404636383057, + "num_tokens": 85536890.0, + "step": 3353 + }, + { + "epoch": 0.3683285745662201, + "grad_norm": 2.0708389282226562, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6916321516036987, + "num_tokens": 85566293.0, + "step": 3354 + }, + { + "epoch": 0.36843839226883374, + "grad_norm": 2.3112075328826904, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.6999052166938782, + "num_tokens": 85589045.0, + "step": 3355 + }, + { + "epoch": 0.3685482099714474, + "grad_norm": 2.145864725112915, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7060689926147461, + "num_tokens": 85618155.0, + "step": 3356 + }, + { + "epoch": 0.3686580276740611, + "grad_norm": 2.1855709552764893, + "learning_rate": 1e-06, + "loss": 1.1138, + "mean_token_accuracy": 0.6700416803359985, + "num_tokens": 85647516.0, + "step": 3357 + }, + { + "epoch": 0.36876784537667473, + "grad_norm": 2.0868043899536133, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7102487087249756, + "num_tokens": 85674499.0, + "step": 3358 + }, + { + "epoch": 0.3688776630792884, + "grad_norm": 2.4448041915893555, + "learning_rate": 1e-06, + "loss": 1.0195, + "mean_token_accuracy": 0.6865265369415283, + "num_tokens": 85695443.0, + "step": 3359 + }, + { + "epoch": 0.368987480781902, + "grad_norm": 2.3089706897735596, + "learning_rate": 1e-06, + "loss": 1.0245, + "mean_token_accuracy": 0.6867275238037109, + "num_tokens": 85719990.0, + "step": 3360 + }, + { + "epoch": 0.3690972984845157, + "grad_norm": 2.472437858581543, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7008993029594421, + "num_tokens": 85740544.0, + "step": 3361 + }, + { + "epoch": 0.36920711618712937, + "grad_norm": 1.9930213689804077, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.6991162896156311, + "num_tokens": 85769807.0, + "step": 3362 + }, + { + "epoch": 0.369316933889743, + "grad_norm": 2.222963333129883, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.6865048408508301, + "num_tokens": 85797327.0, + "step": 3363 + }, + { + "epoch": 0.36942675159235666, + "grad_norm": 2.4890921115875244, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.691556453704834, + "num_tokens": 85818280.0, + "step": 3364 + }, + { + "epoch": 0.36953656929497036, + "grad_norm": 2.4471542835235596, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6984583139419556, + "num_tokens": 85841746.0, + "step": 3365 + }, + { + "epoch": 0.369646386997584, + "grad_norm": 2.323143243789673, + "learning_rate": 1e-06, + "loss": 1.0877, + "mean_token_accuracy": 0.6778491735458374, + "num_tokens": 85866786.0, + "step": 3366 + }, + { + "epoch": 0.36975620470019765, + "grad_norm": 2.156403064727783, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7206506133079529, + "num_tokens": 85893836.0, + "step": 3367 + }, + { + "epoch": 0.36986602240281136, + "grad_norm": 2.0909218788146973, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6950877904891968, + "num_tokens": 85924860.0, + "step": 3368 + }, + { + "epoch": 0.369975840105425, + "grad_norm": 2.3547024726867676, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.6998194456100464, + "num_tokens": 85948265.0, + "step": 3369 + }, + { + "epoch": 0.37008565780803865, + "grad_norm": 2.1599068641662598, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7005361318588257, + "num_tokens": 85974441.0, + "step": 3370 + }, + { + "epoch": 0.3701954755106523, + "grad_norm": 2.317272901535034, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7148809432983398, + "num_tokens": 86001777.0, + "step": 3371 + }, + { + "epoch": 0.370305293213266, + "grad_norm": 2.214871883392334, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.700283408164978, + "num_tokens": 86028222.0, + "step": 3372 + }, + { + "epoch": 0.37041511091587964, + "grad_norm": 2.1216485500335693, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.6890464425086975, + "num_tokens": 86057872.0, + "step": 3373 + }, + { + "epoch": 0.3705249286184933, + "grad_norm": 2.164721727371216, + "learning_rate": 1e-06, + "loss": 1.0165, + "mean_token_accuracy": 0.6992990970611572, + "num_tokens": 86085516.0, + "step": 3374 + }, + { + "epoch": 0.370634746321107, + "grad_norm": 2.2260830402374268, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.7007794380187988, + "num_tokens": 86110928.0, + "step": 3375 + }, + { + "epoch": 0.37074456402372064, + "grad_norm": 2.488629102706909, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6989433765411377, + "num_tokens": 86131334.0, + "step": 3376 + }, + { + "epoch": 0.3708543817263343, + "grad_norm": 2.0635218620300293, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7193640470504761, + "num_tokens": 86159198.0, + "step": 3377 + }, + { + "epoch": 0.3709641994289479, + "grad_norm": 2.377688407897949, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7242761254310608, + "num_tokens": 86179821.0, + "step": 3378 + }, + { + "epoch": 0.37107401713156163, + "grad_norm": 2.450192928314209, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7163708209991455, + "num_tokens": 86201002.0, + "step": 3379 + }, + { + "epoch": 0.3711838348341753, + "grad_norm": 1.9961795806884766, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6938194036483765, + "num_tokens": 86231473.0, + "step": 3380 + }, + { + "epoch": 0.3712936525367889, + "grad_norm": 2.2157816886901855, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.700827419757843, + "num_tokens": 86256856.0, + "step": 3381 + }, + { + "epoch": 0.37140347023940257, + "grad_norm": 2.339437961578369, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7071883082389832, + "num_tokens": 86280273.0, + "step": 3382 + }, + { + "epoch": 0.37151328794201627, + "grad_norm": 2.0599265098571777, + "learning_rate": 1e-06, + "loss": 0.7584, + "mean_token_accuracy": 0.7596758008003235, + "num_tokens": 86304167.0, + "step": 3383 + }, + { + "epoch": 0.3716231056446299, + "grad_norm": 2.1894900798797607, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6992565393447876, + "num_tokens": 86329833.0, + "step": 3384 + }, + { + "epoch": 0.37173292334724356, + "grad_norm": 2.508293628692627, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6918079257011414, + "num_tokens": 86352997.0, + "step": 3385 + }, + { + "epoch": 0.37184274104985726, + "grad_norm": 1.9958922863006592, + "learning_rate": 1e-06, + "loss": 1.097, + "mean_token_accuracy": 0.6785314083099365, + "num_tokens": 86389136.0, + "step": 3386 + }, + { + "epoch": 0.3719525587524709, + "grad_norm": 2.4784138202667236, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7074903249740601, + "num_tokens": 86413230.0, + "step": 3387 + }, + { + "epoch": 0.37206237645508455, + "grad_norm": 2.6187543869018555, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.6999140977859497, + "num_tokens": 86432174.0, + "step": 3388 + }, + { + "epoch": 0.3721721941576982, + "grad_norm": 2.4794554710388184, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7034092545509338, + "num_tokens": 86452733.0, + "step": 3389 + }, + { + "epoch": 0.3722820118603119, + "grad_norm": 2.204359531402588, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7064522504806519, + "num_tokens": 86477728.0, + "step": 3390 + }, + { + "epoch": 0.37239182956292555, + "grad_norm": 2.367841958999634, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7103665471076965, + "num_tokens": 86500155.0, + "step": 3391 + }, + { + "epoch": 0.3725016472655392, + "grad_norm": 2.1479837894439697, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6919629573822021, + "num_tokens": 86529605.0, + "step": 3392 + }, + { + "epoch": 0.37261146496815284, + "grad_norm": 2.3354461193084717, + "learning_rate": 1e-06, + "loss": 1.0948, + "mean_token_accuracy": 0.6779848337173462, + "num_tokens": 86553107.0, + "step": 3393 + }, + { + "epoch": 0.37272128267076654, + "grad_norm": 2.4145898818969727, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.6988759636878967, + "num_tokens": 86573488.0, + "step": 3394 + }, + { + "epoch": 0.3728311003733802, + "grad_norm": 2.0490450859069824, + "learning_rate": 1e-06, + "loss": 1.0926, + "mean_token_accuracy": 0.6755725145339966, + "num_tokens": 86603746.0, + "step": 3395 + }, + { + "epoch": 0.37294091807599383, + "grad_norm": 2.2419111728668213, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.6987823843955994, + "num_tokens": 86629314.0, + "step": 3396 + }, + { + "epoch": 0.37305073577860753, + "grad_norm": 2.0159497261047363, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.6903063654899597, + "num_tokens": 86658818.0, + "step": 3397 + }, + { + "epoch": 0.3731605534812212, + "grad_norm": 2.1104767322540283, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7104525566101074, + "num_tokens": 86684734.0, + "step": 3398 + }, + { + "epoch": 0.3732703711838348, + "grad_norm": 2.254610776901245, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7083086967468262, + "num_tokens": 86709256.0, + "step": 3399 + }, + { + "epoch": 0.37338018888644847, + "grad_norm": 2.126155138015747, + "learning_rate": 1e-06, + "loss": 1.0371, + "mean_token_accuracy": 0.6876053810119629, + "num_tokens": 86738859.0, + "step": 3400 + }, + { + "epoch": 0.3734900065890622, + "grad_norm": 2.0188193321228027, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.7010210752487183, + "num_tokens": 86770524.0, + "step": 3401 + }, + { + "epoch": 0.3735998242916758, + "grad_norm": 2.2126898765563965, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7177162170410156, + "num_tokens": 86795883.0, + "step": 3402 + }, + { + "epoch": 0.37370964199428947, + "grad_norm": 2.135115146636963, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7207805514335632, + "num_tokens": 86820248.0, + "step": 3403 + }, + { + "epoch": 0.37381945969690317, + "grad_norm": 2.060293674468994, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.696702778339386, + "num_tokens": 86850309.0, + "step": 3404 + }, + { + "epoch": 0.3739292773995168, + "grad_norm": 2.1761958599090576, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7198914289474487, + "num_tokens": 86878285.0, + "step": 3405 + }, + { + "epoch": 0.37403909510213046, + "grad_norm": 2.4522156715393066, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7027381658554077, + "num_tokens": 86899731.0, + "step": 3406 + }, + { + "epoch": 0.3741489128047441, + "grad_norm": 2.202542304992676, + "learning_rate": 1e-06, + "loss": 1.0932, + "mean_token_accuracy": 0.6758346557617188, + "num_tokens": 86930334.0, + "step": 3407 + }, + { + "epoch": 0.3742587305073578, + "grad_norm": 2.0355985164642334, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.7006922960281372, + "num_tokens": 86960701.0, + "step": 3408 + }, + { + "epoch": 0.37436854820997145, + "grad_norm": 2.0915169715881348, + "learning_rate": 1e-06, + "loss": 1.0961, + "mean_token_accuracy": 0.6729983687400818, + "num_tokens": 86992394.0, + "step": 3409 + }, + { + "epoch": 0.3744783659125851, + "grad_norm": 1.960878849029541, + "learning_rate": 1e-06, + "loss": 1.0217, + "mean_token_accuracy": 0.6924242973327637, + "num_tokens": 87026484.0, + "step": 3410 + }, + { + "epoch": 0.37458818361519874, + "grad_norm": 2.6357743740081787, + "learning_rate": 1e-06, + "loss": 1.0334, + "mean_token_accuracy": 0.7004563808441162, + "num_tokens": 87046908.0, + "step": 3411 + }, + { + "epoch": 0.37469800131781245, + "grad_norm": 2.3192996978759766, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.6993484497070312, + "num_tokens": 87071892.0, + "step": 3412 + }, + { + "epoch": 0.3748078190204261, + "grad_norm": 2.034651279449463, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7196325063705444, + "num_tokens": 87101057.0, + "step": 3413 + }, + { + "epoch": 0.37491763672303974, + "grad_norm": 2.0021355152130127, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.7004963159561157, + "num_tokens": 87132065.0, + "step": 3414 + }, + { + "epoch": 0.37502745442565344, + "grad_norm": 2.30901837348938, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7121721506118774, + "num_tokens": 87155549.0, + "step": 3415 + }, + { + "epoch": 0.3751372721282671, + "grad_norm": 2.7677364349365234, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.7001345753669739, + "num_tokens": 87176715.0, + "step": 3416 + }, + { + "epoch": 0.37524708983088073, + "grad_norm": 2.3354334831237793, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7146448493003845, + "num_tokens": 87200717.0, + "step": 3417 + }, + { + "epoch": 0.3753569075334944, + "grad_norm": 2.461460590362549, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6963310241699219, + "num_tokens": 87225465.0, + "step": 3418 + }, + { + "epoch": 0.3754667252361081, + "grad_norm": 2.472022533416748, + "learning_rate": 1e-06, + "loss": 1.0761, + "mean_token_accuracy": 0.6852325797080994, + "num_tokens": 87249242.0, + "step": 3419 + }, + { + "epoch": 0.3755765429387217, + "grad_norm": 2.283029794692993, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.7005458474159241, + "num_tokens": 87272436.0, + "step": 3420 + }, + { + "epoch": 0.37568636064133537, + "grad_norm": 2.1931326389312744, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6880059242248535, + "num_tokens": 87298129.0, + "step": 3421 + }, + { + "epoch": 0.37579617834394907, + "grad_norm": 2.349850654602051, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7141856551170349, + "num_tokens": 87322213.0, + "step": 3422 + }, + { + "epoch": 0.3759059960465627, + "grad_norm": 2.085822105407715, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7088826894760132, + "num_tokens": 87350443.0, + "step": 3423 + }, + { + "epoch": 0.37601581374917636, + "grad_norm": 1.9585487842559814, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7003759145736694, + "num_tokens": 87379857.0, + "step": 3424 + }, + { + "epoch": 0.37612563145179, + "grad_norm": 2.0167126655578613, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.6927611827850342, + "num_tokens": 87410319.0, + "step": 3425 + }, + { + "epoch": 0.3762354491544037, + "grad_norm": 2.2072911262512207, + "learning_rate": 1e-06, + "loss": 1.0464, + "mean_token_accuracy": 0.6868066787719727, + "num_tokens": 87437014.0, + "step": 3426 + }, + { + "epoch": 0.37634526685701736, + "grad_norm": 2.327831268310547, + "learning_rate": 1e-06, + "loss": 1.0447, + "mean_token_accuracy": 0.6903603076934814, + "num_tokens": 87462003.0, + "step": 3427 + }, + { + "epoch": 0.376455084559631, + "grad_norm": 2.1457512378692627, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.7013605833053589, + "num_tokens": 87489243.0, + "step": 3428 + }, + { + "epoch": 0.37656490226224465, + "grad_norm": 2.411097764968872, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7207935452461243, + "num_tokens": 87510806.0, + "step": 3429 + }, + { + "epoch": 0.37667471996485835, + "grad_norm": 2.0862653255462646, + "learning_rate": 1e-06, + "loss": 1.0347, + "mean_token_accuracy": 0.684171736240387, + "num_tokens": 87540848.0, + "step": 3430 + }, + { + "epoch": 0.376784537667472, + "grad_norm": 2.00431489944458, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.7121555805206299, + "num_tokens": 87570705.0, + "step": 3431 + }, + { + "epoch": 0.37689435537008564, + "grad_norm": 2.585594654083252, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7081932425498962, + "num_tokens": 87590604.0, + "step": 3432 + }, + { + "epoch": 0.37700417307269934, + "grad_norm": 2.0877981185913086, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6901722550392151, + "num_tokens": 87619146.0, + "step": 3433 + }, + { + "epoch": 0.377113990775313, + "grad_norm": 2.2893242835998535, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.6940464973449707, + "num_tokens": 87642782.0, + "step": 3434 + }, + { + "epoch": 0.37722380847792664, + "grad_norm": 2.432734489440918, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7082358598709106, + "num_tokens": 87664252.0, + "step": 3435 + }, + { + "epoch": 0.3773336261805403, + "grad_norm": 2.3396191596984863, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6933584809303284, + "num_tokens": 87686958.0, + "step": 3436 + }, + { + "epoch": 0.377443443883154, + "grad_norm": 2.2987377643585205, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7121070027351379, + "num_tokens": 87710950.0, + "step": 3437 + }, + { + "epoch": 0.37755326158576763, + "grad_norm": 2.016934394836426, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7088661193847656, + "num_tokens": 87741412.0, + "step": 3438 + }, + { + "epoch": 0.3776630792883813, + "grad_norm": 2.472508668899536, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7155781388282776, + "num_tokens": 87763982.0, + "step": 3439 + }, + { + "epoch": 0.3777728969909949, + "grad_norm": 1.8701750040054321, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6885284185409546, + "num_tokens": 87796445.0, + "step": 3440 + }, + { + "epoch": 0.3778827146936086, + "grad_norm": 2.0882043838500977, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6941287517547607, + "num_tokens": 87824710.0, + "step": 3441 + }, + { + "epoch": 0.37799253239622227, + "grad_norm": 2.393608808517456, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6925175189971924, + "num_tokens": 87848388.0, + "step": 3442 + }, + { + "epoch": 0.3781023500988359, + "grad_norm": 2.2707605361938477, + "learning_rate": 1e-06, + "loss": 1.0658, + "mean_token_accuracy": 0.6883253455162048, + "num_tokens": 87874271.0, + "step": 3443 + }, + { + "epoch": 0.3782121678014496, + "grad_norm": 2.3665897846221924, + "learning_rate": 1e-06, + "loss": 1.0222, + "mean_token_accuracy": 0.6946886777877808, + "num_tokens": 87897368.0, + "step": 3444 + }, + { + "epoch": 0.37832198550406326, + "grad_norm": 2.1079604625701904, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.6915847063064575, + "num_tokens": 87923523.0, + "step": 3445 + }, + { + "epoch": 0.3784318032066769, + "grad_norm": 2.186889886856079, + "learning_rate": 1e-06, + "loss": 1.0476, + "mean_token_accuracy": 0.6797589063644409, + "num_tokens": 87951245.0, + "step": 3446 + }, + { + "epoch": 0.37854162090929055, + "grad_norm": 2.2974088191986084, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.6964592933654785, + "num_tokens": 87974378.0, + "step": 3447 + }, + { + "epoch": 0.37865143861190426, + "grad_norm": 2.232222318649292, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.6954810619354248, + "num_tokens": 88001324.0, + "step": 3448 + }, + { + "epoch": 0.3787612563145179, + "grad_norm": 2.4715027809143066, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7406880259513855, + "num_tokens": 88024351.0, + "step": 3449 + }, + { + "epoch": 0.37887107401713155, + "grad_norm": 2.3422842025756836, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.702802836894989, + "num_tokens": 88047411.0, + "step": 3450 + }, + { + "epoch": 0.37898089171974525, + "grad_norm": 2.2517666816711426, + "learning_rate": 1e-06, + "loss": 1.0758, + "mean_token_accuracy": 0.6796492338180542, + "num_tokens": 88072163.0, + "step": 3451 + }, + { + "epoch": 0.3790907094223589, + "grad_norm": 1.9938257932662964, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.704226553440094, + "num_tokens": 88102330.0, + "step": 3452 + }, + { + "epoch": 0.37920052712497254, + "grad_norm": 2.031522035598755, + "learning_rate": 1e-06, + "loss": 1.0915, + "mean_token_accuracy": 0.675830066204071, + "num_tokens": 88132915.0, + "step": 3453 + }, + { + "epoch": 0.3793103448275862, + "grad_norm": 2.7026548385620117, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7042929530143738, + "num_tokens": 88153619.0, + "step": 3454 + }, + { + "epoch": 0.3794201625301999, + "grad_norm": 2.4121367931365967, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.712830126285553, + "num_tokens": 88175484.0, + "step": 3455 + }, + { + "epoch": 0.37952998023281354, + "grad_norm": 2.311035633087158, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7238038778305054, + "num_tokens": 88199426.0, + "step": 3456 + }, + { + "epoch": 0.3796397979354272, + "grad_norm": 2.54563045501709, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7121366262435913, + "num_tokens": 88220166.0, + "step": 3457 + }, + { + "epoch": 0.3797496156380408, + "grad_norm": 2.3458354473114014, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7023929357528687, + "num_tokens": 88243932.0, + "step": 3458 + }, + { + "epoch": 0.37985943334065453, + "grad_norm": 2.5932955741882324, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.7062157392501831, + "num_tokens": 88266030.0, + "step": 3459 + }, + { + "epoch": 0.3799692510432682, + "grad_norm": 2.4772863388061523, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7134029865264893, + "num_tokens": 88288509.0, + "step": 3460 + }, + { + "epoch": 0.3800790687458818, + "grad_norm": 2.309607982635498, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6900684833526611, + "num_tokens": 88315515.0, + "step": 3461 + }, + { + "epoch": 0.3801888864484955, + "grad_norm": 1.9995927810668945, + "learning_rate": 1e-06, + "loss": 1.0649, + "mean_token_accuracy": 0.6807774305343628, + "num_tokens": 88347688.0, + "step": 3462 + }, + { + "epoch": 0.38029870415110917, + "grad_norm": 2.3888862133026123, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7238076329231262, + "num_tokens": 88371880.0, + "step": 3463 + }, + { + "epoch": 0.3804085218537228, + "grad_norm": 2.6697397232055664, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7058159112930298, + "num_tokens": 88392284.0, + "step": 3464 + }, + { + "epoch": 0.38051833955633646, + "grad_norm": 2.230973482131958, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7258270978927612, + "num_tokens": 88418909.0, + "step": 3465 + }, + { + "epoch": 0.38062815725895016, + "grad_norm": 2.1905972957611084, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7195709347724915, + "num_tokens": 88443487.0, + "step": 3466 + }, + { + "epoch": 0.3807379749615638, + "grad_norm": 2.0588760375976562, + "learning_rate": 1e-06, + "loss": 1.025, + "mean_token_accuracy": 0.6834119558334351, + "num_tokens": 88474415.0, + "step": 3467 + }, + { + "epoch": 0.38084779266417745, + "grad_norm": 2.4347894191741943, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.7042962312698364, + "num_tokens": 88497820.0, + "step": 3468 + }, + { + "epoch": 0.3809576103667911, + "grad_norm": 2.0782229900360107, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.720955491065979, + "num_tokens": 88524014.0, + "step": 3469 + }, + { + "epoch": 0.3810674280694048, + "grad_norm": 1.9448648691177368, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7380504608154297, + "num_tokens": 88550466.0, + "step": 3470 + }, + { + "epoch": 0.38117724577201845, + "grad_norm": 2.2615585327148438, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7146291732788086, + "num_tokens": 88574063.0, + "step": 3471 + }, + { + "epoch": 0.3812870634746321, + "grad_norm": 2.30941104888916, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6937231421470642, + "num_tokens": 88598532.0, + "step": 3472 + }, + { + "epoch": 0.3813968811772458, + "grad_norm": 2.22469162940979, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7190719842910767, + "num_tokens": 88621521.0, + "step": 3473 + }, + { + "epoch": 0.38150669887985944, + "grad_norm": 2.3848507404327393, + "learning_rate": 1e-06, + "loss": 1.0209, + "mean_token_accuracy": 0.6984531283378601, + "num_tokens": 88646077.0, + "step": 3474 + }, + { + "epoch": 0.3816165165824731, + "grad_norm": 2.115128755569458, + "learning_rate": 1e-06, + "loss": 1.0663, + "mean_token_accuracy": 0.6859637498855591, + "num_tokens": 88675148.0, + "step": 3475 + }, + { + "epoch": 0.38172633428508673, + "grad_norm": 2.04416823387146, + "learning_rate": 1e-06, + "loss": 1.0541, + "mean_token_accuracy": 0.6886026859283447, + "num_tokens": 88704328.0, + "step": 3476 + }, + { + "epoch": 0.38183615198770043, + "grad_norm": 2.4348843097686768, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.7006194591522217, + "num_tokens": 88726179.0, + "step": 3477 + }, + { + "epoch": 0.3819459696903141, + "grad_norm": 2.1634881496429443, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7210690379142761, + "num_tokens": 88752823.0, + "step": 3478 + }, + { + "epoch": 0.3820557873929277, + "grad_norm": 2.3844118118286133, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7212824821472168, + "num_tokens": 88774261.0, + "step": 3479 + }, + { + "epoch": 0.3821656050955414, + "grad_norm": 2.227667808532715, + "learning_rate": 1e-06, + "loss": 1.0736, + "mean_token_accuracy": 0.6884820461273193, + "num_tokens": 88799484.0, + "step": 3480 + }, + { + "epoch": 0.3822754227981551, + "grad_norm": 2.0729105472564697, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7035959959030151, + "num_tokens": 88827479.0, + "step": 3481 + }, + { + "epoch": 0.3823852405007687, + "grad_norm": 2.3925483226776123, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7054609060287476, + "num_tokens": 88849330.0, + "step": 3482 + }, + { + "epoch": 0.38249505820338237, + "grad_norm": 2.400757312774658, + "learning_rate": 1e-06, + "loss": 1.0704, + "mean_token_accuracy": 0.6799368858337402, + "num_tokens": 88874051.0, + "step": 3483 + }, + { + "epoch": 0.38260487590599607, + "grad_norm": 2.0771608352661133, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7049452066421509, + "num_tokens": 88901627.0, + "step": 3484 + }, + { + "epoch": 0.3827146936086097, + "grad_norm": 2.2839324474334717, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7236374020576477, + "num_tokens": 88924506.0, + "step": 3485 + }, + { + "epoch": 0.38282451131122336, + "grad_norm": 2.0570385456085205, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6983098983764648, + "num_tokens": 88952604.0, + "step": 3486 + }, + { + "epoch": 0.382934329013837, + "grad_norm": 2.0447373390197754, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7145571112632751, + "num_tokens": 88980744.0, + "step": 3487 + }, + { + "epoch": 0.3830441467164507, + "grad_norm": 1.9953160285949707, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.6965844631195068, + "num_tokens": 89012855.0, + "step": 3488 + }, + { + "epoch": 0.38315396441906435, + "grad_norm": 2.2090415954589844, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6991734504699707, + "num_tokens": 89039409.0, + "step": 3489 + }, + { + "epoch": 0.383263782121678, + "grad_norm": 2.159883499145508, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7043219208717346, + "num_tokens": 89065159.0, + "step": 3490 + }, + { + "epoch": 0.3833735998242917, + "grad_norm": 2.3250722885131836, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7239021062850952, + "num_tokens": 89087304.0, + "step": 3491 + }, + { + "epoch": 0.38348341752690535, + "grad_norm": 2.066532850265503, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7327505350112915, + "num_tokens": 89117083.0, + "step": 3492 + }, + { + "epoch": 0.383593235229519, + "grad_norm": 2.243745803833008, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.718737006187439, + "num_tokens": 89141555.0, + "step": 3493 + }, + { + "epoch": 0.38370305293213264, + "grad_norm": 2.08026123046875, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.7077574133872986, + "num_tokens": 89171254.0, + "step": 3494 + }, + { + "epoch": 0.38381287063474634, + "grad_norm": 2.056492805480957, + "learning_rate": 1e-06, + "loss": 1.0625, + "mean_token_accuracy": 0.6777878403663635, + "num_tokens": 89201950.0, + "step": 3495 + }, + { + "epoch": 0.38392268833736, + "grad_norm": 2.1646206378936768, + "learning_rate": 1e-06, + "loss": 1.0251, + "mean_token_accuracy": 0.6925961375236511, + "num_tokens": 89227475.0, + "step": 3496 + }, + { + "epoch": 0.38403250603997363, + "grad_norm": 2.356147289276123, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.7026307582855225, + "num_tokens": 89250272.0, + "step": 3497 + }, + { + "epoch": 0.38414232374258733, + "grad_norm": 2.3868248462677, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7299866080284119, + "num_tokens": 89272220.0, + "step": 3498 + }, + { + "epoch": 0.384252141445201, + "grad_norm": 2.670543909072876, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7094963788986206, + "num_tokens": 89290912.0, + "step": 3499 + }, + { + "epoch": 0.3843619591478146, + "grad_norm": 2.2275516986846924, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7014179229736328, + "num_tokens": 89316827.0, + "step": 3500 + }, + { + "epoch": 0.38447177685042827, + "grad_norm": 2.070131778717041, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.7133298516273499, + "num_tokens": 89345160.0, + "step": 3501 + }, + { + "epoch": 0.38458159455304197, + "grad_norm": 2.452723979949951, + "learning_rate": 1e-06, + "loss": 1.0773, + "mean_token_accuracy": 0.6796510219573975, + "num_tokens": 89368289.0, + "step": 3502 + }, + { + "epoch": 0.3846914122556556, + "grad_norm": 2.174743413925171, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7171514630317688, + "num_tokens": 89393498.0, + "step": 3503 + }, + { + "epoch": 0.38480122995826926, + "grad_norm": 1.930920958518982, + "learning_rate": 1e-06, + "loss": 1.0945, + "mean_token_accuracy": 0.6705272197723389, + "num_tokens": 89429898.0, + "step": 3504 + }, + { + "epoch": 0.3849110476608829, + "grad_norm": 2.3325836658477783, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7204048037528992, + "num_tokens": 89453080.0, + "step": 3505 + }, + { + "epoch": 0.3850208653634966, + "grad_norm": 2.3063900470733643, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7081243991851807, + "num_tokens": 89478247.0, + "step": 3506 + }, + { + "epoch": 0.38513068306611026, + "grad_norm": 2.572493553161621, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.717600405216217, + "num_tokens": 89499177.0, + "step": 3507 + }, + { + "epoch": 0.3852405007687239, + "grad_norm": 2.0409116744995117, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7139603495597839, + "num_tokens": 89527787.0, + "step": 3508 + }, + { + "epoch": 0.3853503184713376, + "grad_norm": 2.608132839202881, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7436195015907288, + "num_tokens": 89548301.0, + "step": 3509 + }, + { + "epoch": 0.38546013617395125, + "grad_norm": 2.5136008262634277, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7120214700698853, + "num_tokens": 89567723.0, + "step": 3510 + }, + { + "epoch": 0.3855699538765649, + "grad_norm": 2.364351272583008, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7036116123199463, + "num_tokens": 89590286.0, + "step": 3511 + }, + { + "epoch": 0.38567977157917854, + "grad_norm": 2.385012626647949, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.7026822566986084, + "num_tokens": 89612130.0, + "step": 3512 + }, + { + "epoch": 0.38578958928179224, + "grad_norm": 2.3021159172058105, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.710820198059082, + "num_tokens": 89634229.0, + "step": 3513 + }, + { + "epoch": 0.3858994069844059, + "grad_norm": 2.1366255283355713, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6945544481277466, + "num_tokens": 89660640.0, + "step": 3514 + }, + { + "epoch": 0.38600922468701954, + "grad_norm": 2.2331860065460205, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.707216739654541, + "num_tokens": 89688751.0, + "step": 3515 + }, + { + "epoch": 0.3861190423896332, + "grad_norm": 2.3355791568756104, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7078540325164795, + "num_tokens": 89711275.0, + "step": 3516 + }, + { + "epoch": 0.3862288600922469, + "grad_norm": 2.717874526977539, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7119054794311523, + "num_tokens": 89729663.0, + "step": 3517 + }, + { + "epoch": 0.38633867779486053, + "grad_norm": 2.0805017948150635, + "learning_rate": 1e-06, + "loss": 1.0171, + "mean_token_accuracy": 0.6945101618766785, + "num_tokens": 89759376.0, + "step": 3518 + }, + { + "epoch": 0.3864484954974742, + "grad_norm": 2.575883150100708, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7106614112854004, + "num_tokens": 89777410.0, + "step": 3519 + }, + { + "epoch": 0.3865583132000879, + "grad_norm": 2.441127300262451, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7331455945968628, + "num_tokens": 89798212.0, + "step": 3520 + }, + { + "epoch": 0.3866681309027015, + "grad_norm": 1.8952780961990356, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7205379009246826, + "num_tokens": 89832305.0, + "step": 3521 + }, + { + "epoch": 0.38677794860531517, + "grad_norm": 2.437690019607544, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.6968307495117188, + "num_tokens": 89855259.0, + "step": 3522 + }, + { + "epoch": 0.3868877663079288, + "grad_norm": 2.242319345474243, + "learning_rate": 1e-06, + "loss": 1.0687, + "mean_token_accuracy": 0.6792464256286621, + "num_tokens": 89881413.0, + "step": 3523 + }, + { + "epoch": 0.3869975840105425, + "grad_norm": 2.049530029296875, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7082706093788147, + "num_tokens": 89912087.0, + "step": 3524 + }, + { + "epoch": 0.38710740171315616, + "grad_norm": 2.2465503215789795, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7032164931297302, + "num_tokens": 89934311.0, + "step": 3525 + }, + { + "epoch": 0.3872172194157698, + "grad_norm": 1.921673059463501, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.701011598110199, + "num_tokens": 89965561.0, + "step": 3526 + }, + { + "epoch": 0.3873270371183835, + "grad_norm": 1.994235873222351, + "learning_rate": 1e-06, + "loss": 1.139, + "mean_token_accuracy": 0.6611608266830444, + "num_tokens": 89996572.0, + "step": 3527 + }, + { + "epoch": 0.38743685482099716, + "grad_norm": 2.5503408908843994, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6891441345214844, + "num_tokens": 90017703.0, + "step": 3528 + }, + { + "epoch": 0.3875466725236108, + "grad_norm": 2.4772469997406006, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.6863168478012085, + "num_tokens": 90040235.0, + "step": 3529 + }, + { + "epoch": 0.38765649022622445, + "grad_norm": 2.3956985473632812, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7080313563346863, + "num_tokens": 90064459.0, + "step": 3530 + }, + { + "epoch": 0.38776630792883815, + "grad_norm": 2.2314224243164062, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7214468717575073, + "num_tokens": 90090627.0, + "step": 3531 + }, + { + "epoch": 0.3878761256314518, + "grad_norm": 2.4849185943603516, + "learning_rate": 1e-06, + "loss": 1.0362, + "mean_token_accuracy": 0.6887827515602112, + "num_tokens": 90113917.0, + "step": 3532 + }, + { + "epoch": 0.38798594333406544, + "grad_norm": 2.0679473876953125, + "learning_rate": 1e-06, + "loss": 1.0653, + "mean_token_accuracy": 0.6844909191131592, + "num_tokens": 90144983.0, + "step": 3533 + }, + { + "epoch": 0.3880957610366791, + "grad_norm": 2.378851890563965, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7255909442901611, + "num_tokens": 90167111.0, + "step": 3534 + }, + { + "epoch": 0.3882055787392928, + "grad_norm": 2.4930548667907715, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7249103784561157, + "num_tokens": 90187163.0, + "step": 3535 + }, + { + "epoch": 0.38831539644190644, + "grad_norm": 2.2228145599365234, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.700992226600647, + "num_tokens": 90213973.0, + "step": 3536 + }, + { + "epoch": 0.3884252141445201, + "grad_norm": 2.1263442039489746, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.7001845836639404, + "num_tokens": 90240150.0, + "step": 3537 + }, + { + "epoch": 0.3885350318471338, + "grad_norm": 2.4353601932525635, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6994943618774414, + "num_tokens": 90262281.0, + "step": 3538 + }, + { + "epoch": 0.38864484954974743, + "grad_norm": 2.2757537364959717, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6952313184738159, + "num_tokens": 90287220.0, + "step": 3539 + }, + { + "epoch": 0.3887546672523611, + "grad_norm": 1.794887661933899, + "learning_rate": 1e-06, + "loss": 1.1045, + "mean_token_accuracy": 0.6721774339675903, + "num_tokens": 90326963.0, + "step": 3540 + }, + { + "epoch": 0.3888644849549747, + "grad_norm": 2.157074213027954, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.7096216082572937, + "num_tokens": 90353818.0, + "step": 3541 + }, + { + "epoch": 0.3889743026575884, + "grad_norm": 1.8399754762649536, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.6901167631149292, + "num_tokens": 90386655.0, + "step": 3542 + }, + { + "epoch": 0.38908412036020207, + "grad_norm": 2.449924945831299, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7212339043617249, + "num_tokens": 90406681.0, + "step": 3543 + }, + { + "epoch": 0.3891939380628157, + "grad_norm": 2.093060255050659, + "learning_rate": 1e-06, + "loss": 1.048, + "mean_token_accuracy": 0.6860766410827637, + "num_tokens": 90434679.0, + "step": 3544 + }, + { + "epoch": 0.38930375576542936, + "grad_norm": 2.066948652267456, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6907086968421936, + "num_tokens": 90463191.0, + "step": 3545 + }, + { + "epoch": 0.38941357346804306, + "grad_norm": 2.242889165878296, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7125463485717773, + "num_tokens": 90487753.0, + "step": 3546 + }, + { + "epoch": 0.3895233911706567, + "grad_norm": 2.0777571201324463, + "learning_rate": 1e-06, + "loss": 1.0739, + "mean_token_accuracy": 0.6786460280418396, + "num_tokens": 90516637.0, + "step": 3547 + }, + { + "epoch": 0.38963320887327035, + "grad_norm": 2.2667043209075928, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7223681211471558, + "num_tokens": 90539989.0, + "step": 3548 + }, + { + "epoch": 0.38974302657588406, + "grad_norm": 2.09466814994812, + "learning_rate": 1e-06, + "loss": 1.0398, + "mean_token_accuracy": 0.6882651448249817, + "num_tokens": 90569244.0, + "step": 3549 + }, + { + "epoch": 0.3898528442784977, + "grad_norm": 2.213965892791748, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.7028396129608154, + "num_tokens": 90595875.0, + "step": 3550 + }, + { + "epoch": 0.38996266198111135, + "grad_norm": 1.9941776990890503, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6946197748184204, + "num_tokens": 90625518.0, + "step": 3551 + }, + { + "epoch": 0.390072479683725, + "grad_norm": 2.1091601848602295, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6985762119293213, + "num_tokens": 90652971.0, + "step": 3552 + }, + { + "epoch": 0.3901822973863387, + "grad_norm": 2.570070743560791, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7217990159988403, + "num_tokens": 90671723.0, + "step": 3553 + }, + { + "epoch": 0.39029211508895234, + "grad_norm": 2.606295585632324, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7394354343414307, + "num_tokens": 90690357.0, + "step": 3554 + }, + { + "epoch": 0.390401932791566, + "grad_norm": 2.2063376903533936, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7170839905738831, + "num_tokens": 90715116.0, + "step": 3555 + }, + { + "epoch": 0.3905117504941797, + "grad_norm": 2.25216007232666, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7185152769088745, + "num_tokens": 90738854.0, + "step": 3556 + }, + { + "epoch": 0.39062156819679333, + "grad_norm": 2.2966811656951904, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6919700503349304, + "num_tokens": 90762367.0, + "step": 3557 + }, + { + "epoch": 0.390731385899407, + "grad_norm": 2.3685667514801025, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6969865560531616, + "num_tokens": 90784986.0, + "step": 3558 + }, + { + "epoch": 0.3908412036020206, + "grad_norm": 2.0271472930908203, + "learning_rate": 1e-06, + "loss": 1.0592, + "mean_token_accuracy": 0.6870063543319702, + "num_tokens": 90815099.0, + "step": 3559 + }, + { + "epoch": 0.3909510213046343, + "grad_norm": 2.1486892700195312, + "learning_rate": 1e-06, + "loss": 1.0664, + "mean_token_accuracy": 0.6783856749534607, + "num_tokens": 90843868.0, + "step": 3560 + }, + { + "epoch": 0.391060839007248, + "grad_norm": 2.6787612438201904, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7305809855461121, + "num_tokens": 90861416.0, + "step": 3561 + }, + { + "epoch": 0.3911706567098616, + "grad_norm": 2.302506923675537, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7035574913024902, + "num_tokens": 90884282.0, + "step": 3562 + }, + { + "epoch": 0.39128047441247527, + "grad_norm": 2.3156049251556396, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7105838060379028, + "num_tokens": 90908229.0, + "step": 3563 + }, + { + "epoch": 0.39139029211508897, + "grad_norm": 2.2795281410217285, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7221792936325073, + "num_tokens": 90932413.0, + "step": 3564 + }, + { + "epoch": 0.3915001098177026, + "grad_norm": 2.4506752490997314, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7051935195922852, + "num_tokens": 90954183.0, + "step": 3565 + }, + { + "epoch": 0.39160992752031626, + "grad_norm": 2.1185953617095947, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.693640947341919, + "num_tokens": 90982725.0, + "step": 3566 + }, + { + "epoch": 0.39171974522292996, + "grad_norm": 2.1322968006134033, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7098783254623413, + "num_tokens": 91009703.0, + "step": 3567 + }, + { + "epoch": 0.3918295629255436, + "grad_norm": 1.960448145866394, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.6933570504188538, + "num_tokens": 91043078.0, + "step": 3568 + }, + { + "epoch": 0.39193938062815725, + "grad_norm": 2.2680933475494385, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7178421020507812, + "num_tokens": 91066839.0, + "step": 3569 + }, + { + "epoch": 0.3920491983307709, + "grad_norm": 2.388197183609009, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.690726637840271, + "num_tokens": 91089079.0, + "step": 3570 + }, + { + "epoch": 0.3921590160333846, + "grad_norm": 2.505671262741089, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7038879990577698, + "num_tokens": 91110113.0, + "step": 3571 + }, + { + "epoch": 0.39226883373599825, + "grad_norm": 2.238326072692871, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7028160095214844, + "num_tokens": 91134877.0, + "step": 3572 + }, + { + "epoch": 0.3923786514386119, + "grad_norm": 2.500380039215088, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7058697938919067, + "num_tokens": 91157031.0, + "step": 3573 + }, + { + "epoch": 0.3924884691412256, + "grad_norm": 2.587954044342041, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7083033323287964, + "num_tokens": 91175995.0, + "step": 3574 + }, + { + "epoch": 0.39259828684383924, + "grad_norm": 2.28708815574646, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.7024043798446655, + "num_tokens": 91200071.0, + "step": 3575 + }, + { + "epoch": 0.3927081045464529, + "grad_norm": 2.6191720962524414, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7282897233963013, + "num_tokens": 91218751.0, + "step": 3576 + }, + { + "epoch": 0.39281792224906653, + "grad_norm": 2.4975831508636475, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6983113884925842, + "num_tokens": 91240901.0, + "step": 3577 + }, + { + "epoch": 0.39292773995168023, + "grad_norm": 2.3296923637390137, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6940855979919434, + "num_tokens": 91267249.0, + "step": 3578 + }, + { + "epoch": 0.3930375576542939, + "grad_norm": 1.9712703227996826, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.6953628063201904, + "num_tokens": 91297659.0, + "step": 3579 + }, + { + "epoch": 0.3931473753569075, + "grad_norm": 2.1101977825164795, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.697426438331604, + "num_tokens": 91326748.0, + "step": 3580 + }, + { + "epoch": 0.39325719305952117, + "grad_norm": 2.2277328968048096, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7094507217407227, + "num_tokens": 91352504.0, + "step": 3581 + }, + { + "epoch": 0.39336701076213487, + "grad_norm": 2.3250932693481445, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.6993404626846313, + "num_tokens": 91376313.0, + "step": 3582 + }, + { + "epoch": 0.3934768284647485, + "grad_norm": 2.200014591217041, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7033708095550537, + "num_tokens": 91400564.0, + "step": 3583 + }, + { + "epoch": 0.39358664616736216, + "grad_norm": 2.0625534057617188, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6913062930107117, + "num_tokens": 91428562.0, + "step": 3584 + }, + { + "epoch": 0.39369646386997587, + "grad_norm": 2.1858630180358887, + "learning_rate": 1e-06, + "loss": 1.0537, + "mean_token_accuracy": 0.681995153427124, + "num_tokens": 91455638.0, + "step": 3585 + }, + { + "epoch": 0.3938062815725895, + "grad_norm": 2.3456408977508545, + "learning_rate": 1e-06, + "loss": 1.1016, + "mean_token_accuracy": 0.6709662079811096, + "num_tokens": 91480271.0, + "step": 3586 + }, + { + "epoch": 0.39391609927520316, + "grad_norm": 2.0527713298797607, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.7038478851318359, + "num_tokens": 91508638.0, + "step": 3587 + }, + { + "epoch": 0.3940259169778168, + "grad_norm": 2.0399739742279053, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.693662166595459, + "num_tokens": 91540287.0, + "step": 3588 + }, + { + "epoch": 0.3941357346804305, + "grad_norm": 2.072415828704834, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7083224058151245, + "num_tokens": 91568110.0, + "step": 3589 + }, + { + "epoch": 0.39424555238304415, + "grad_norm": 2.158742904663086, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6917717456817627, + "num_tokens": 91594948.0, + "step": 3590 + }, + { + "epoch": 0.3943553700856578, + "grad_norm": 2.216942071914673, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7038397192955017, + "num_tokens": 91620330.0, + "step": 3591 + }, + { + "epoch": 0.39446518778827144, + "grad_norm": 2.0447494983673096, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.7024962306022644, + "num_tokens": 91650407.0, + "step": 3592 + }, + { + "epoch": 0.39457500549088514, + "grad_norm": 2.274968385696411, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.6900632381439209, + "num_tokens": 91672960.0, + "step": 3593 + }, + { + "epoch": 0.3946848231934988, + "grad_norm": 2.2531814575195312, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6985908150672913, + "num_tokens": 91699755.0, + "step": 3594 + }, + { + "epoch": 0.39479464089611244, + "grad_norm": 2.1062839031219482, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6968260407447815, + "num_tokens": 91728283.0, + "step": 3595 + }, + { + "epoch": 0.39490445859872614, + "grad_norm": 2.0423104763031006, + "learning_rate": 1e-06, + "loss": 1.1088, + "mean_token_accuracy": 0.6664987206459045, + "num_tokens": 91761233.0, + "step": 3596 + }, + { + "epoch": 0.3950142763013398, + "grad_norm": 2.0154662132263184, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6952750086784363, + "num_tokens": 91791619.0, + "step": 3597 + }, + { + "epoch": 0.39512409400395343, + "grad_norm": 1.9488619565963745, + "learning_rate": 1e-06, + "loss": 1.0366, + "mean_token_accuracy": 0.6871356964111328, + "num_tokens": 91822125.0, + "step": 3598 + }, + { + "epoch": 0.3952339117065671, + "grad_norm": 2.2362661361694336, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.6989921927452087, + "num_tokens": 91846502.0, + "step": 3599 + }, + { + "epoch": 0.3953437294091808, + "grad_norm": 2.174661636352539, + "learning_rate": 1e-06, + "loss": 1.0691, + "mean_token_accuracy": 0.6806957125663757, + "num_tokens": 91874591.0, + "step": 3600 + }, + { + "epoch": 0.3954535471117944, + "grad_norm": 2.291029453277588, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.711144208908081, + "num_tokens": 91900618.0, + "step": 3601 + }, + { + "epoch": 0.39556336481440807, + "grad_norm": 2.266906499862671, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6879333257675171, + "num_tokens": 91928100.0, + "step": 3602 + }, + { + "epoch": 0.39567318251702177, + "grad_norm": 2.1287882328033447, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.695494532585144, + "num_tokens": 91955308.0, + "step": 3603 + }, + { + "epoch": 0.3957830002196354, + "grad_norm": 2.405365467071533, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.688735842704773, + "num_tokens": 91978410.0, + "step": 3604 + }, + { + "epoch": 0.39589281792224906, + "grad_norm": 2.1564717292785645, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6888901591300964, + "num_tokens": 92006705.0, + "step": 3605 + }, + { + "epoch": 0.3960026356248627, + "grad_norm": 2.2709832191467285, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6937723159790039, + "num_tokens": 92030898.0, + "step": 3606 + }, + { + "epoch": 0.3961124533274764, + "grad_norm": 2.45890736579895, + "learning_rate": 1e-06, + "loss": 1.0739, + "mean_token_accuracy": 0.6787805557250977, + "num_tokens": 92052295.0, + "step": 3607 + }, + { + "epoch": 0.39622227103009006, + "grad_norm": 2.3928093910217285, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.6967865228652954, + "num_tokens": 92074045.0, + "step": 3608 + }, + { + "epoch": 0.3963320887327037, + "grad_norm": 2.2743542194366455, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7139902114868164, + "num_tokens": 92098986.0, + "step": 3609 + }, + { + "epoch": 0.39644190643531735, + "grad_norm": 2.2324860095977783, + "learning_rate": 1e-06, + "loss": 1.0414, + "mean_token_accuracy": 0.6836711764335632, + "num_tokens": 92124549.0, + "step": 3610 + }, + { + "epoch": 0.39655172413793105, + "grad_norm": 2.4475975036621094, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.701258659362793, + "num_tokens": 92145194.0, + "step": 3611 + }, + { + "epoch": 0.3966615418405447, + "grad_norm": 2.102686643600464, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7074353694915771, + "num_tokens": 92170944.0, + "step": 3612 + }, + { + "epoch": 0.39677135954315834, + "grad_norm": 2.080611228942871, + "learning_rate": 1e-06, + "loss": 1.0527, + "mean_token_accuracy": 0.6924395561218262, + "num_tokens": 92199669.0, + "step": 3613 + }, + { + "epoch": 0.39688117724577204, + "grad_norm": 2.2400102615356445, + "learning_rate": 1e-06, + "loss": 1.0673, + "mean_token_accuracy": 0.6846291422843933, + "num_tokens": 92228380.0, + "step": 3614 + }, + { + "epoch": 0.3969909949483857, + "grad_norm": 1.9631330966949463, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6826133728027344, + "num_tokens": 92261189.0, + "step": 3615 + }, + { + "epoch": 0.39710081265099934, + "grad_norm": 2.241755962371826, + "learning_rate": 1e-06, + "loss": 1.0511, + "mean_token_accuracy": 0.6851685047149658, + "num_tokens": 92287189.0, + "step": 3616 + }, + { + "epoch": 0.397210630353613, + "grad_norm": 2.0265004634857178, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7046297788619995, + "num_tokens": 92318203.0, + "step": 3617 + }, + { + "epoch": 0.3973204480562267, + "grad_norm": 2.204972743988037, + "learning_rate": 1e-06, + "loss": 1.0791, + "mean_token_accuracy": 0.6835440993309021, + "num_tokens": 92345228.0, + "step": 3618 + }, + { + "epoch": 0.39743026575884033, + "grad_norm": 2.2120823860168457, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7153072357177734, + "num_tokens": 92370850.0, + "step": 3619 + }, + { + "epoch": 0.397540083461454, + "grad_norm": 2.0700483322143555, + "learning_rate": 1e-06, + "loss": 1.0974, + "mean_token_accuracy": 0.6788073182106018, + "num_tokens": 92401207.0, + "step": 3620 + }, + { + "epoch": 0.3976499011640676, + "grad_norm": 2.2763257026672363, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7219236493110657, + "num_tokens": 92421644.0, + "step": 3621 + }, + { + "epoch": 0.3977597188666813, + "grad_norm": 2.3797597885131836, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7108675837516785, + "num_tokens": 92443368.0, + "step": 3622 + }, + { + "epoch": 0.39786953656929497, + "grad_norm": 2.3243660926818848, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.707906186580658, + "num_tokens": 92466719.0, + "step": 3623 + }, + { + "epoch": 0.3979793542719086, + "grad_norm": 2.4058992862701416, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7012944221496582, + "num_tokens": 92489135.0, + "step": 3624 + }, + { + "epoch": 0.3980891719745223, + "grad_norm": 2.0130529403686523, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.6982302665710449, + "num_tokens": 92520779.0, + "step": 3625 + }, + { + "epoch": 0.39819898967713596, + "grad_norm": 1.9744973182678223, + "learning_rate": 1e-06, + "loss": 1.0528, + "mean_token_accuracy": 0.6906442642211914, + "num_tokens": 92552920.0, + "step": 3626 + }, + { + "epoch": 0.3983088073797496, + "grad_norm": 2.0856728553771973, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7162607908248901, + "num_tokens": 92579833.0, + "step": 3627 + }, + { + "epoch": 0.39841862508236325, + "grad_norm": 2.2033355236053467, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7377742528915405, + "num_tokens": 92603705.0, + "step": 3628 + }, + { + "epoch": 0.39852844278497696, + "grad_norm": 2.128614902496338, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6979155540466309, + "num_tokens": 92629860.0, + "step": 3629 + }, + { + "epoch": 0.3986382604875906, + "grad_norm": 2.0466272830963135, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7131710052490234, + "num_tokens": 92659411.0, + "step": 3630 + }, + { + "epoch": 0.39874807819020425, + "grad_norm": 2.2196972370147705, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6998916864395142, + "num_tokens": 92685908.0, + "step": 3631 + }, + { + "epoch": 0.39885789589281795, + "grad_norm": 2.1610448360443115, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6972970962524414, + "num_tokens": 92712466.0, + "step": 3632 + }, + { + "epoch": 0.3989677135954316, + "grad_norm": 2.628340721130371, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.695273756980896, + "num_tokens": 92731740.0, + "step": 3633 + }, + { + "epoch": 0.39907753129804524, + "grad_norm": 2.4535725116729736, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7066028714179993, + "num_tokens": 92756089.0, + "step": 3634 + }, + { + "epoch": 0.3991873490006589, + "grad_norm": 1.9713926315307617, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.7000094056129456, + "num_tokens": 92785688.0, + "step": 3635 + }, + { + "epoch": 0.3992971667032726, + "grad_norm": 2.574228525161743, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7212091684341431, + "num_tokens": 92804900.0, + "step": 3636 + }, + { + "epoch": 0.39940698440588623, + "grad_norm": 1.962624192237854, + "learning_rate": 1e-06, + "loss": 1.0509, + "mean_token_accuracy": 0.6915836334228516, + "num_tokens": 92839048.0, + "step": 3637 + }, + { + "epoch": 0.3995168021084999, + "grad_norm": 2.0986883640289307, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.690773069858551, + "num_tokens": 92869364.0, + "step": 3638 + }, + { + "epoch": 0.3996266198111135, + "grad_norm": 2.31593656539917, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7089922428131104, + "num_tokens": 92891811.0, + "step": 3639 + }, + { + "epoch": 0.3997364375137272, + "grad_norm": 1.9878147840499878, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6938656568527222, + "num_tokens": 92922464.0, + "step": 3640 + }, + { + "epoch": 0.3998462552163409, + "grad_norm": 2.068897008895874, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7132229208946228, + "num_tokens": 92950331.0, + "step": 3641 + }, + { + "epoch": 0.3999560729189545, + "grad_norm": 2.244393825531006, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7177654504776001, + "num_tokens": 92974350.0, + "step": 3642 + }, + { + "epoch": 0.4000658906215682, + "grad_norm": 2.0018997192382812, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7167621850967407, + "num_tokens": 93004276.0, + "step": 3643 + }, + { + "epoch": 0.40017570832418187, + "grad_norm": 2.0877139568328857, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6990938782691956, + "num_tokens": 93032897.0, + "step": 3644 + }, + { + "epoch": 0.4002855260267955, + "grad_norm": 1.919227123260498, + "learning_rate": 1e-06, + "loss": 1.128, + "mean_token_accuracy": 0.6707637906074524, + "num_tokens": 93067905.0, + "step": 3645 + }, + { + "epoch": 0.40039534372940916, + "grad_norm": 2.3633127212524414, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7108433246612549, + "num_tokens": 93089882.0, + "step": 3646 + }, + { + "epoch": 0.40050516143202286, + "grad_norm": 2.2306554317474365, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7115359306335449, + "num_tokens": 93114993.0, + "step": 3647 + }, + { + "epoch": 0.4006149791346365, + "grad_norm": 2.0973379611968994, + "learning_rate": 1e-06, + "loss": 1.0399, + "mean_token_accuracy": 0.6872222423553467, + "num_tokens": 93142948.0, + "step": 3648 + }, + { + "epoch": 0.40072479683725015, + "grad_norm": 2.2144863605499268, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6832141876220703, + "num_tokens": 93168656.0, + "step": 3649 + }, + { + "epoch": 0.4008346145398638, + "grad_norm": 2.558835506439209, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6898523569107056, + "num_tokens": 93188324.0, + "step": 3650 + }, + { + "epoch": 0.4009444322424775, + "grad_norm": 2.1567907333374023, + "learning_rate": 1e-06, + "loss": 1.0556, + "mean_token_accuracy": 0.6817781329154968, + "num_tokens": 93215422.0, + "step": 3651 + }, + { + "epoch": 0.40105424994509115, + "grad_norm": 2.081501007080078, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.7008433938026428, + "num_tokens": 93242787.0, + "step": 3652 + }, + { + "epoch": 0.4011640676477048, + "grad_norm": 2.129995584487915, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7087140679359436, + "num_tokens": 93270846.0, + "step": 3653 + }, + { + "epoch": 0.4012738853503185, + "grad_norm": 2.1034321784973145, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7054730653762817, + "num_tokens": 93300933.0, + "step": 3654 + }, + { + "epoch": 0.40138370305293214, + "grad_norm": 2.2615160942077637, + "learning_rate": 1e-06, + "loss": 1.0423, + "mean_token_accuracy": 0.694448709487915, + "num_tokens": 93325043.0, + "step": 3655 + }, + { + "epoch": 0.4014935207555458, + "grad_norm": 2.2296202182769775, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7089343667030334, + "num_tokens": 93348912.0, + "step": 3656 + }, + { + "epoch": 0.40160333845815943, + "grad_norm": 2.2443175315856934, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.71100914478302, + "num_tokens": 93373312.0, + "step": 3657 + }, + { + "epoch": 0.40171315616077313, + "grad_norm": 2.1110119819641113, + "learning_rate": 1e-06, + "loss": 1.0576, + "mean_token_accuracy": 0.683300256729126, + "num_tokens": 93401676.0, + "step": 3658 + }, + { + "epoch": 0.4018229738633868, + "grad_norm": 2.2969253063201904, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7161346673965454, + "num_tokens": 93426958.0, + "step": 3659 + }, + { + "epoch": 0.4019327915660004, + "grad_norm": 2.3417303562164307, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7082049250602722, + "num_tokens": 93452511.0, + "step": 3660 + }, + { + "epoch": 0.4020426092686141, + "grad_norm": 1.9484013319015503, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6989443302154541, + "num_tokens": 93483821.0, + "step": 3661 + }, + { + "epoch": 0.40215242697122777, + "grad_norm": 2.3652408123016357, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7138139605522156, + "num_tokens": 93506671.0, + "step": 3662 + }, + { + "epoch": 0.4022622446738414, + "grad_norm": 2.5756800174713135, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7241783738136292, + "num_tokens": 93525727.0, + "step": 3663 + }, + { + "epoch": 0.40237206237645506, + "grad_norm": 2.4023983478546143, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.7011503577232361, + "num_tokens": 93547396.0, + "step": 3664 + }, + { + "epoch": 0.40248188007906877, + "grad_norm": 2.597815752029419, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7115756869316101, + "num_tokens": 93566607.0, + "step": 3665 + }, + { + "epoch": 0.4025916977816824, + "grad_norm": 2.004391670227051, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6927582621574402, + "num_tokens": 93602228.0, + "step": 3666 + }, + { + "epoch": 0.40270151548429606, + "grad_norm": 2.725318670272827, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.6948603987693787, + "num_tokens": 93621041.0, + "step": 3667 + }, + { + "epoch": 0.4028113331869097, + "grad_norm": 2.080660104751587, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6967326402664185, + "num_tokens": 93650723.0, + "step": 3668 + }, + { + "epoch": 0.4029211508895234, + "grad_norm": 2.1234867572784424, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6889150142669678, + "num_tokens": 93676722.0, + "step": 3669 + }, + { + "epoch": 0.40303096859213705, + "grad_norm": 1.8790628910064697, + "learning_rate": 1e-06, + "loss": 1.0605, + "mean_token_accuracy": 0.6724691987037659, + "num_tokens": 93709746.0, + "step": 3670 + }, + { + "epoch": 0.4031407862947507, + "grad_norm": 2.252506971359253, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7220752239227295, + "num_tokens": 93733139.0, + "step": 3671 + }, + { + "epoch": 0.4032506039973644, + "grad_norm": 2.0269880294799805, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7052866220474243, + "num_tokens": 93763280.0, + "step": 3672 + }, + { + "epoch": 0.40336042169997804, + "grad_norm": 2.1997599601745605, + "learning_rate": 1e-06, + "loss": 1.0484, + "mean_token_accuracy": 0.6859158277511597, + "num_tokens": 93791815.0, + "step": 3673 + }, + { + "epoch": 0.4034702394025917, + "grad_norm": 2.0840489864349365, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.709801197052002, + "num_tokens": 93818122.0, + "step": 3674 + }, + { + "epoch": 0.40358005710520534, + "grad_norm": 2.5259578227996826, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6960363388061523, + "num_tokens": 93839509.0, + "step": 3675 + }, + { + "epoch": 0.40368987480781904, + "grad_norm": 2.486388921737671, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.701261579990387, + "num_tokens": 93863248.0, + "step": 3676 + }, + { + "epoch": 0.4037996925104327, + "grad_norm": 2.3154513835906982, + "learning_rate": 1e-06, + "loss": 1.1135, + "mean_token_accuracy": 0.6727589964866638, + "num_tokens": 93886693.0, + "step": 3677 + }, + { + "epoch": 0.40390951021304633, + "grad_norm": 2.473903179168701, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7209059000015259, + "num_tokens": 93907645.0, + "step": 3678 + }, + { + "epoch": 0.40401932791566003, + "grad_norm": 1.8398791551589966, + "learning_rate": 1e-06, + "loss": 1.0583, + "mean_token_accuracy": 0.683044970035553, + "num_tokens": 93944715.0, + "step": 3679 + }, + { + "epoch": 0.4041291456182737, + "grad_norm": 2.0340564250946045, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.6909263134002686, + "num_tokens": 93977305.0, + "step": 3680 + }, + { + "epoch": 0.4042389633208873, + "grad_norm": 2.2670652866363525, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7043983936309814, + "num_tokens": 94000386.0, + "step": 3681 + }, + { + "epoch": 0.40434878102350097, + "grad_norm": 2.2046265602111816, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7076238393783569, + "num_tokens": 94024853.0, + "step": 3682 + }, + { + "epoch": 0.40445859872611467, + "grad_norm": 2.0562548637390137, + "learning_rate": 1e-06, + "loss": 1.0508, + "mean_token_accuracy": 0.6854718327522278, + "num_tokens": 94053950.0, + "step": 3683 + }, + { + "epoch": 0.4045684164287283, + "grad_norm": 2.3530313968658447, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7066818475723267, + "num_tokens": 94076414.0, + "step": 3684 + }, + { + "epoch": 0.40467823413134196, + "grad_norm": 1.9528508186340332, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6961308121681213, + "num_tokens": 94109022.0, + "step": 3685 + }, + { + "epoch": 0.4047880518339556, + "grad_norm": 2.3702051639556885, + "learning_rate": 1e-06, + "loss": 1.0857, + "mean_token_accuracy": 0.6815313100814819, + "num_tokens": 94134614.0, + "step": 3686 + }, + { + "epoch": 0.4048978695365693, + "grad_norm": 2.460341453552246, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6960596442222595, + "num_tokens": 94155915.0, + "step": 3687 + }, + { + "epoch": 0.40500768723918296, + "grad_norm": 2.1489531993865967, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.6931400895118713, + "num_tokens": 94182273.0, + "step": 3688 + }, + { + "epoch": 0.4051175049417966, + "grad_norm": 2.1111457347869873, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7353314757347107, + "num_tokens": 94208397.0, + "step": 3689 + }, + { + "epoch": 0.4052273226444103, + "grad_norm": 2.335756540298462, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.701352596282959, + "num_tokens": 94231378.0, + "step": 3690 + }, + { + "epoch": 0.40533714034702395, + "grad_norm": 2.2735965251922607, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7213233709335327, + "num_tokens": 94255536.0, + "step": 3691 + }, + { + "epoch": 0.4054469580496376, + "grad_norm": 2.8122591972351074, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7225003242492676, + "num_tokens": 94271684.0, + "step": 3692 + }, + { + "epoch": 0.40555677575225124, + "grad_norm": 2.0454742908477783, + "learning_rate": 1e-06, + "loss": 1.0306, + "mean_token_accuracy": 0.6905036568641663, + "num_tokens": 94301328.0, + "step": 3693 + }, + { + "epoch": 0.40566659345486494, + "grad_norm": 2.4603090286254883, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7187197208404541, + "num_tokens": 94322304.0, + "step": 3694 + }, + { + "epoch": 0.4057764111574786, + "grad_norm": 2.1620230674743652, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.6915323138237, + "num_tokens": 94347961.0, + "step": 3695 + }, + { + "epoch": 0.40588622886009224, + "grad_norm": 2.2089931964874268, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7070654630661011, + "num_tokens": 94371687.0, + "step": 3696 + }, + { + "epoch": 0.4059960465627059, + "grad_norm": 2.2708852291107178, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6981786489486694, + "num_tokens": 94395770.0, + "step": 3697 + }, + { + "epoch": 0.4061058642653196, + "grad_norm": 2.3101165294647217, + "learning_rate": 1e-06, + "loss": 1.0238, + "mean_token_accuracy": 0.6959893107414246, + "num_tokens": 94420602.0, + "step": 3698 + }, + { + "epoch": 0.40621568196793323, + "grad_norm": 2.2852251529693604, + "learning_rate": 1e-06, + "loss": 1.0908, + "mean_token_accuracy": 0.6829042434692383, + "num_tokens": 94448589.0, + "step": 3699 + }, + { + "epoch": 0.4063254996705469, + "grad_norm": 2.206765651702881, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.707962691783905, + "num_tokens": 94473993.0, + "step": 3700 + }, + { + "epoch": 0.4064353173731606, + "grad_norm": 2.284844160079956, + "learning_rate": 1e-06, + "loss": 1.0582, + "mean_token_accuracy": 0.6787744760513306, + "num_tokens": 94499499.0, + "step": 3701 + }, + { + "epoch": 0.4065451350757742, + "grad_norm": 2.298804521560669, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.726544976234436, + "num_tokens": 94522241.0, + "step": 3702 + }, + { + "epoch": 0.40665495277838787, + "grad_norm": 2.410019874572754, + "learning_rate": 1e-06, + "loss": 1.0455, + "mean_token_accuracy": 0.6923513412475586, + "num_tokens": 94545225.0, + "step": 3703 + }, + { + "epoch": 0.4067647704810015, + "grad_norm": 2.054285764694214, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7088900804519653, + "num_tokens": 94573998.0, + "step": 3704 + }, + { + "epoch": 0.4068745881836152, + "grad_norm": 2.3987951278686523, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.6957306861877441, + "num_tokens": 94597517.0, + "step": 3705 + }, + { + "epoch": 0.40698440588622886, + "grad_norm": 2.6015450954437256, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7256643176078796, + "num_tokens": 94616902.0, + "step": 3706 + }, + { + "epoch": 0.4070942235888425, + "grad_norm": 2.2934038639068604, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7116504907608032, + "num_tokens": 94639717.0, + "step": 3707 + }, + { + "epoch": 0.4072040412914562, + "grad_norm": 1.9655499458312988, + "learning_rate": 1e-06, + "loss": 1.0789, + "mean_token_accuracy": 0.6888222098350525, + "num_tokens": 94670997.0, + "step": 3708 + }, + { + "epoch": 0.40731385899406986, + "grad_norm": 2.3057515621185303, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6984789967536926, + "num_tokens": 94695407.0, + "step": 3709 + }, + { + "epoch": 0.4074236766966835, + "grad_norm": 2.354184865951538, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7071325778961182, + "num_tokens": 94718015.0, + "step": 3710 + }, + { + "epoch": 0.40753349439929715, + "grad_norm": 2.287559747695923, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.703657329082489, + "num_tokens": 94742999.0, + "step": 3711 + }, + { + "epoch": 0.40764331210191085, + "grad_norm": 2.3855433464050293, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7151397466659546, + "num_tokens": 94765028.0, + "step": 3712 + }, + { + "epoch": 0.4077531298045245, + "grad_norm": 2.4869887828826904, + "learning_rate": 1e-06, + "loss": 1.0704, + "mean_token_accuracy": 0.681825578212738, + "num_tokens": 94787295.0, + "step": 3713 + }, + { + "epoch": 0.40786294750713814, + "grad_norm": 2.5584609508514404, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7093331217765808, + "num_tokens": 94806899.0, + "step": 3714 + }, + { + "epoch": 0.4079727652097518, + "grad_norm": 2.186821699142456, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7111083269119263, + "num_tokens": 94832984.0, + "step": 3715 + }, + { + "epoch": 0.4080825829123655, + "grad_norm": 2.326138496398926, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.7012150287628174, + "num_tokens": 94854837.0, + "step": 3716 + }, + { + "epoch": 0.40819240061497913, + "grad_norm": 2.5000572204589844, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7236847877502441, + "num_tokens": 94874719.0, + "step": 3717 + }, + { + "epoch": 0.4083022183175928, + "grad_norm": 2.0842247009277344, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6948410272598267, + "num_tokens": 94901765.0, + "step": 3718 + }, + { + "epoch": 0.4084120360202065, + "grad_norm": 2.2334702014923096, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.692196786403656, + "num_tokens": 94926934.0, + "step": 3719 + }, + { + "epoch": 0.4085218537228201, + "grad_norm": 2.186452865600586, + "learning_rate": 1e-06, + "loss": 1.0565, + "mean_token_accuracy": 0.6794766187667847, + "num_tokens": 94951999.0, + "step": 3720 + }, + { + "epoch": 0.4086316714254338, + "grad_norm": 2.1483700275421143, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7170479893684387, + "num_tokens": 94977392.0, + "step": 3721 + }, + { + "epoch": 0.4087414891280474, + "grad_norm": 2.0565125942230225, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.6999996900558472, + "num_tokens": 95007815.0, + "step": 3722 + }, + { + "epoch": 0.4088513068306611, + "grad_norm": 2.2669732570648193, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.6985794305801392, + "num_tokens": 95031196.0, + "step": 3723 + }, + { + "epoch": 0.40896112453327477, + "grad_norm": 2.0283186435699463, + "learning_rate": 1e-06, + "loss": 1.0727, + "mean_token_accuracy": 0.6824303269386292, + "num_tokens": 95062053.0, + "step": 3724 + }, + { + "epoch": 0.4090709422358884, + "grad_norm": 2.112321376800537, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7097627520561218, + "num_tokens": 95089625.0, + "step": 3725 + }, + { + "epoch": 0.40918075993850206, + "grad_norm": 2.3219974040985107, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6851910352706909, + "num_tokens": 95116121.0, + "step": 3726 + }, + { + "epoch": 0.40929057764111576, + "grad_norm": 2.3662052154541016, + "learning_rate": 1e-06, + "loss": 1.065, + "mean_token_accuracy": 0.6826670169830322, + "num_tokens": 95141225.0, + "step": 3727 + }, + { + "epoch": 0.4094003953437294, + "grad_norm": 2.32497239112854, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.7023307681083679, + "num_tokens": 95164198.0, + "step": 3728 + }, + { + "epoch": 0.40951021304634305, + "grad_norm": 2.0525455474853516, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6906152367591858, + "num_tokens": 95192030.0, + "step": 3729 + }, + { + "epoch": 0.40962003074895675, + "grad_norm": 2.1897873878479004, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.694351077079773, + "num_tokens": 95217910.0, + "step": 3730 + }, + { + "epoch": 0.4097298484515704, + "grad_norm": 2.288457155227661, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6940215826034546, + "num_tokens": 95242295.0, + "step": 3731 + }, + { + "epoch": 0.40983966615418405, + "grad_norm": 2.4046823978424072, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7033568620681763, + "num_tokens": 95266140.0, + "step": 3732 + }, + { + "epoch": 0.4099494838567977, + "grad_norm": 2.293285846710205, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7338240146636963, + "num_tokens": 95287599.0, + "step": 3733 + }, + { + "epoch": 0.4100593015594114, + "grad_norm": 2.4795963764190674, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.732101559638977, + "num_tokens": 95307460.0, + "step": 3734 + }, + { + "epoch": 0.41016911926202504, + "grad_norm": 2.505601406097412, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7203471064567566, + "num_tokens": 95328741.0, + "step": 3735 + }, + { + "epoch": 0.4102789369646387, + "grad_norm": 2.4062819480895996, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6931265592575073, + "num_tokens": 95350828.0, + "step": 3736 + }, + { + "epoch": 0.4103887546672524, + "grad_norm": 2.3855390548706055, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7167582511901855, + "num_tokens": 95371405.0, + "step": 3737 + }, + { + "epoch": 0.41049857236986603, + "grad_norm": 2.1370584964752197, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7296678423881531, + "num_tokens": 95396884.0, + "step": 3738 + }, + { + "epoch": 0.4106083900724797, + "grad_norm": 2.0955793857574463, + "learning_rate": 1e-06, + "loss": 1.0427, + "mean_token_accuracy": 0.6840167045593262, + "num_tokens": 95425331.0, + "step": 3739 + }, + { + "epoch": 0.4107182077750933, + "grad_norm": 2.291663646697998, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.6993625164031982, + "num_tokens": 95449000.0, + "step": 3740 + }, + { + "epoch": 0.410828025477707, + "grad_norm": 2.418757438659668, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.6926902532577515, + "num_tokens": 95470193.0, + "step": 3741 + }, + { + "epoch": 0.41093784318032067, + "grad_norm": 2.167189359664917, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7037292718887329, + "num_tokens": 95495916.0, + "step": 3742 + }, + { + "epoch": 0.4110476608829343, + "grad_norm": 2.104942798614502, + "learning_rate": 1e-06, + "loss": 1.051, + "mean_token_accuracy": 0.6871089935302734, + "num_tokens": 95526696.0, + "step": 3743 + }, + { + "epoch": 0.41115747858554796, + "grad_norm": 2.189436912536621, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.6874792575836182, + "num_tokens": 95554891.0, + "step": 3744 + }, + { + "epoch": 0.41126729628816167, + "grad_norm": 2.3533506393432617, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.701612114906311, + "num_tokens": 95576916.0, + "step": 3745 + }, + { + "epoch": 0.4113771139907753, + "grad_norm": 2.336632490158081, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7084874510765076, + "num_tokens": 95602740.0, + "step": 3746 + }, + { + "epoch": 0.41148693169338896, + "grad_norm": 2.081622362136841, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6974972486495972, + "num_tokens": 95631395.0, + "step": 3747 + }, + { + "epoch": 0.41159674939600266, + "grad_norm": 2.073625087738037, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6913846731185913, + "num_tokens": 95658757.0, + "step": 3748 + }, + { + "epoch": 0.4117065670986163, + "grad_norm": 1.994935393333435, + "learning_rate": 1e-06, + "loss": 1.0826, + "mean_token_accuracy": 0.6728326082229614, + "num_tokens": 95689966.0, + "step": 3749 + }, + { + "epoch": 0.41181638480122995, + "grad_norm": 2.355360746383667, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.7001636624336243, + "num_tokens": 95713657.0, + "step": 3750 + }, + { + "epoch": 0.4119262025038436, + "grad_norm": 2.0855181217193604, + "learning_rate": 1e-06, + "loss": 1.0426, + "mean_token_accuracy": 0.6837965846061707, + "num_tokens": 95742775.0, + "step": 3751 + }, + { + "epoch": 0.4120360202064573, + "grad_norm": 1.865418553352356, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7112741470336914, + "num_tokens": 95776591.0, + "step": 3752 + }, + { + "epoch": 0.41214583790907094, + "grad_norm": 2.178449869155884, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.7048964500427246, + "num_tokens": 95802498.0, + "step": 3753 + }, + { + "epoch": 0.4122556556116846, + "grad_norm": 2.2301788330078125, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.7036405801773071, + "num_tokens": 95827782.0, + "step": 3754 + }, + { + "epoch": 0.4123654733142983, + "grad_norm": 2.321739912033081, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.703753650188446, + "num_tokens": 95851796.0, + "step": 3755 + }, + { + "epoch": 0.41247529101691194, + "grad_norm": 2.0324320793151855, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.727953314781189, + "num_tokens": 95877867.0, + "step": 3756 + }, + { + "epoch": 0.4125851087195256, + "grad_norm": 2.4020986557006836, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.706107497215271, + "num_tokens": 95901289.0, + "step": 3757 + }, + { + "epoch": 0.41269492642213923, + "grad_norm": 2.2738571166992188, + "learning_rate": 1e-06, + "loss": 1.047, + "mean_token_accuracy": 0.6852188110351562, + "num_tokens": 95929923.0, + "step": 3758 + }, + { + "epoch": 0.41280474412475293, + "grad_norm": 1.981680154800415, + "learning_rate": 1e-06, + "loss": 1.0454, + "mean_token_accuracy": 0.6853484511375427, + "num_tokens": 95961480.0, + "step": 3759 + }, + { + "epoch": 0.4129145618273666, + "grad_norm": 2.2746644020080566, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7047044038772583, + "num_tokens": 95986701.0, + "step": 3760 + }, + { + "epoch": 0.4130243795299802, + "grad_norm": 2.3082616329193115, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.7077587842941284, + "num_tokens": 96011306.0, + "step": 3761 + }, + { + "epoch": 0.41313419723259387, + "grad_norm": 2.3590970039367676, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6948376893997192, + "num_tokens": 96038401.0, + "step": 3762 + }, + { + "epoch": 0.41324401493520757, + "grad_norm": 2.0806961059570312, + "learning_rate": 1e-06, + "loss": 1.0689, + "mean_token_accuracy": 0.6823259592056274, + "num_tokens": 96068663.0, + "step": 3763 + }, + { + "epoch": 0.4133538326378212, + "grad_norm": 2.1851258277893066, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7002344727516174, + "num_tokens": 96096839.0, + "step": 3764 + }, + { + "epoch": 0.41346365034043486, + "grad_norm": 2.2180633544921875, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7094981670379639, + "num_tokens": 96122061.0, + "step": 3765 + }, + { + "epoch": 0.41357346804304856, + "grad_norm": 2.336503744125366, + "learning_rate": 1e-06, + "loss": 1.0416, + "mean_token_accuracy": 0.6792629361152649, + "num_tokens": 96145322.0, + "step": 3766 + }, + { + "epoch": 0.4136832857456622, + "grad_norm": 2.160148859024048, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6923327445983887, + "num_tokens": 96172623.0, + "step": 3767 + }, + { + "epoch": 0.41379310344827586, + "grad_norm": 2.364264726638794, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7074454426765442, + "num_tokens": 96198445.0, + "step": 3768 + }, + { + "epoch": 0.4139029211508895, + "grad_norm": 2.245997905731201, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6988915205001831, + "num_tokens": 96222556.0, + "step": 3769 + }, + { + "epoch": 0.4140127388535032, + "grad_norm": 2.3093514442443848, + "learning_rate": 1e-06, + "loss": 1.0429, + "mean_token_accuracy": 0.6907898187637329, + "num_tokens": 96247984.0, + "step": 3770 + }, + { + "epoch": 0.41412255655611685, + "grad_norm": 2.0997443199157715, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6971591711044312, + "num_tokens": 96275043.0, + "step": 3771 + }, + { + "epoch": 0.4142323742587305, + "grad_norm": 2.1166346073150635, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7209113240242004, + "num_tokens": 96302127.0, + "step": 3772 + }, + { + "epoch": 0.41434219196134414, + "grad_norm": 2.410996675491333, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6936739087104797, + "num_tokens": 96323646.0, + "step": 3773 + }, + { + "epoch": 0.41445200966395784, + "grad_norm": 2.0761525630950928, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7053519487380981, + "num_tokens": 96353030.0, + "step": 3774 + }, + { + "epoch": 0.4145618273665715, + "grad_norm": 2.3657987117767334, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7252638339996338, + "num_tokens": 96376502.0, + "step": 3775 + }, + { + "epoch": 0.41467164506918514, + "grad_norm": 2.2147715091705322, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6929134726524353, + "num_tokens": 96404781.0, + "step": 3776 + }, + { + "epoch": 0.41478146277179884, + "grad_norm": 2.47430682182312, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7105616331100464, + "num_tokens": 96425543.0, + "step": 3777 + }, + { + "epoch": 0.4148912804744125, + "grad_norm": 2.0379745960235596, + "learning_rate": 1e-06, + "loss": 1.1023, + "mean_token_accuracy": 0.67002934217453, + "num_tokens": 96456151.0, + "step": 3778 + }, + { + "epoch": 0.41500109817702613, + "grad_norm": 2.1459567546844482, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.6983593702316284, + "num_tokens": 96483141.0, + "step": 3779 + }, + { + "epoch": 0.4151109158796398, + "grad_norm": 2.6058568954467773, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6988843679428101, + "num_tokens": 96503323.0, + "step": 3780 + }, + { + "epoch": 0.4152207335822535, + "grad_norm": 2.6404244899749756, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7106570601463318, + "num_tokens": 96522229.0, + "step": 3781 + }, + { + "epoch": 0.4153305512848671, + "grad_norm": 2.095247268676758, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6904928088188171, + "num_tokens": 96554232.0, + "step": 3782 + }, + { + "epoch": 0.41544036898748077, + "grad_norm": 2.3964738845825195, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7207667231559753, + "num_tokens": 96574542.0, + "step": 3783 + }, + { + "epoch": 0.41555018669009447, + "grad_norm": 2.2224068641662598, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.6917222738265991, + "num_tokens": 96599001.0, + "step": 3784 + }, + { + "epoch": 0.4156600043927081, + "grad_norm": 2.452301263809204, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.6973434686660767, + "num_tokens": 96620919.0, + "step": 3785 + }, + { + "epoch": 0.41576982209532176, + "grad_norm": 2.1289873123168945, + "learning_rate": 1e-06, + "loss": 1.0718, + "mean_token_accuracy": 0.6858287453651428, + "num_tokens": 96646398.0, + "step": 3786 + }, + { + "epoch": 0.4158796397979354, + "grad_norm": 1.992493987083435, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7244081497192383, + "num_tokens": 96673785.0, + "step": 3787 + }, + { + "epoch": 0.4159894575005491, + "grad_norm": 2.094172954559326, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.700806736946106, + "num_tokens": 96701762.0, + "step": 3788 + }, + { + "epoch": 0.41609927520316276, + "grad_norm": 2.3781983852386475, + "learning_rate": 1e-06, + "loss": 1.0303, + "mean_token_accuracy": 0.6867462992668152, + "num_tokens": 96725792.0, + "step": 3789 + }, + { + "epoch": 0.4162090929057764, + "grad_norm": 2.3876092433929443, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7248799204826355, + "num_tokens": 96746549.0, + "step": 3790 + }, + { + "epoch": 0.41631891060839005, + "grad_norm": 2.2932045459747314, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7138233780860901, + "num_tokens": 96767249.0, + "step": 3791 + }, + { + "epoch": 0.41642872831100375, + "grad_norm": 2.026848316192627, + "learning_rate": 1e-06, + "loss": 1.0827, + "mean_token_accuracy": 0.6725767850875854, + "num_tokens": 96797912.0, + "step": 3792 + }, + { + "epoch": 0.4165385460136174, + "grad_norm": 2.0396249294281006, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7084804773330688, + "num_tokens": 96826012.0, + "step": 3793 + }, + { + "epoch": 0.41664836371623104, + "grad_norm": 1.8855215311050415, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7111958861351013, + "num_tokens": 96858791.0, + "step": 3794 + }, + { + "epoch": 0.41675818141884474, + "grad_norm": 2.1138789653778076, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.6865643858909607, + "num_tokens": 96886000.0, + "step": 3795 + }, + { + "epoch": 0.4168679991214584, + "grad_norm": 2.0300211906433105, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.7014511823654175, + "num_tokens": 96914517.0, + "step": 3796 + }, + { + "epoch": 0.41697781682407203, + "grad_norm": 2.1937472820281982, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6942410469055176, + "num_tokens": 96941013.0, + "step": 3797 + }, + { + "epoch": 0.4170876345266857, + "grad_norm": 2.427881956100464, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7079914808273315, + "num_tokens": 96962235.0, + "step": 3798 + }, + { + "epoch": 0.4171974522292994, + "grad_norm": 2.7011353969573975, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7171753644943237, + "num_tokens": 96979641.0, + "step": 3799 + }, + { + "epoch": 0.417307269931913, + "grad_norm": 2.1736555099487305, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7065833806991577, + "num_tokens": 97007333.0, + "step": 3800 + }, + { + "epoch": 0.4174170876345267, + "grad_norm": 2.421248435974121, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.6961865425109863, + "num_tokens": 97029468.0, + "step": 3801 + }, + { + "epoch": 0.4175269053371403, + "grad_norm": 2.1068003177642822, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7053951025009155, + "num_tokens": 97057145.0, + "step": 3802 + }, + { + "epoch": 0.417636723039754, + "grad_norm": 2.325316905975342, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7131036520004272, + "num_tokens": 97080899.0, + "step": 3803 + }, + { + "epoch": 0.41774654074236767, + "grad_norm": 2.2248198986053467, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.746688961982727, + "num_tokens": 97105439.0, + "step": 3804 + }, + { + "epoch": 0.4178563584449813, + "grad_norm": 2.2118980884552, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.7022534608840942, + "num_tokens": 97131038.0, + "step": 3805 + }, + { + "epoch": 0.417966176147595, + "grad_norm": 2.3717904090881348, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.706375241279602, + "num_tokens": 97153654.0, + "step": 3806 + }, + { + "epoch": 0.41807599385020866, + "grad_norm": 1.98939049243927, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.697123110294342, + "num_tokens": 97184188.0, + "step": 3807 + }, + { + "epoch": 0.4181858115528223, + "grad_norm": 2.20607328414917, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7321096062660217, + "num_tokens": 97208036.0, + "step": 3808 + }, + { + "epoch": 0.41829562925543595, + "grad_norm": 2.421595335006714, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.6950176358222961, + "num_tokens": 97229820.0, + "step": 3809 + }, + { + "epoch": 0.41840544695804965, + "grad_norm": 2.303070068359375, + "learning_rate": 1e-06, + "loss": 1.012, + "mean_token_accuracy": 0.688878059387207, + "num_tokens": 97254554.0, + "step": 3810 + }, + { + "epoch": 0.4185152646606633, + "grad_norm": 1.8438799381256104, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7272220849990845, + "num_tokens": 97289675.0, + "step": 3811 + }, + { + "epoch": 0.41862508236327695, + "grad_norm": 2.246925115585327, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.7011314630508423, + "num_tokens": 97314425.0, + "step": 3812 + }, + { + "epoch": 0.41873490006589065, + "grad_norm": 2.186588764190674, + "learning_rate": 1e-06, + "loss": 1.0365, + "mean_token_accuracy": 0.6831231117248535, + "num_tokens": 97341620.0, + "step": 3813 + }, + { + "epoch": 0.4188447177685043, + "grad_norm": 2.415100574493408, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6940879821777344, + "num_tokens": 97363936.0, + "step": 3814 + }, + { + "epoch": 0.41895453547111794, + "grad_norm": 1.910162091255188, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6879183053970337, + "num_tokens": 97396983.0, + "step": 3815 + }, + { + "epoch": 0.4190643531737316, + "grad_norm": 2.2228713035583496, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6969497203826904, + "num_tokens": 97421040.0, + "step": 3816 + }, + { + "epoch": 0.4191741708763453, + "grad_norm": 2.2040679454803467, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7135618329048157, + "num_tokens": 97446607.0, + "step": 3817 + }, + { + "epoch": 0.41928398857895893, + "grad_norm": 2.1044762134552, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7226665019989014, + "num_tokens": 97473818.0, + "step": 3818 + }, + { + "epoch": 0.4193938062815726, + "grad_norm": 2.5093019008636475, + "learning_rate": 1e-06, + "loss": 1.0246, + "mean_token_accuracy": 0.6920753717422485, + "num_tokens": 97495604.0, + "step": 3819 + }, + { + "epoch": 0.4195036239841862, + "grad_norm": 2.326209783554077, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.7039135098457336, + "num_tokens": 97520997.0, + "step": 3820 + }, + { + "epoch": 0.4196134416867999, + "grad_norm": 2.3032543659210205, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7235320210456848, + "num_tokens": 97544760.0, + "step": 3821 + }, + { + "epoch": 0.41972325938941357, + "grad_norm": 2.422297477722168, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7269959449768066, + "num_tokens": 97565120.0, + "step": 3822 + }, + { + "epoch": 0.4198330770920272, + "grad_norm": 2.3182451725006104, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.7064775824546814, + "num_tokens": 97588726.0, + "step": 3823 + }, + { + "epoch": 0.4199428947946409, + "grad_norm": 2.252391815185547, + "learning_rate": 1e-06, + "loss": 1.0929, + "mean_token_accuracy": 0.6752947568893433, + "num_tokens": 97614485.0, + "step": 3824 + }, + { + "epoch": 0.42005271249725457, + "grad_norm": 2.431431770324707, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7268673181533813, + "num_tokens": 97635788.0, + "step": 3825 + }, + { + "epoch": 0.4201625301998682, + "grad_norm": 2.395369529724121, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.730810284614563, + "num_tokens": 97657521.0, + "step": 3826 + }, + { + "epoch": 0.42027234790248186, + "grad_norm": 2.3536174297332764, + "learning_rate": 1e-06, + "loss": 1.0424, + "mean_token_accuracy": 0.6903249025344849, + "num_tokens": 97682976.0, + "step": 3827 + }, + { + "epoch": 0.42038216560509556, + "grad_norm": 2.5444698333740234, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7055497169494629, + "num_tokens": 97703386.0, + "step": 3828 + }, + { + "epoch": 0.4204919833077092, + "grad_norm": 2.28617787361145, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.6978609561920166, + "num_tokens": 97730542.0, + "step": 3829 + }, + { + "epoch": 0.42060180101032285, + "grad_norm": 2.322126865386963, + "learning_rate": 1e-06, + "loss": 1.0431, + "mean_token_accuracy": 0.6831732988357544, + "num_tokens": 97754348.0, + "step": 3830 + }, + { + "epoch": 0.42071161871293655, + "grad_norm": 2.5314483642578125, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7167538404464722, + "num_tokens": 97775307.0, + "step": 3831 + }, + { + "epoch": 0.4208214364155502, + "grad_norm": 2.127354383468628, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6981687545776367, + "num_tokens": 97800459.0, + "step": 3832 + }, + { + "epoch": 0.42093125411816384, + "grad_norm": 2.3121602535247803, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7084186673164368, + "num_tokens": 97824761.0, + "step": 3833 + }, + { + "epoch": 0.4210410718207775, + "grad_norm": 2.372021436691284, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.6907181739807129, + "num_tokens": 97848625.0, + "step": 3834 + }, + { + "epoch": 0.4211508895233912, + "grad_norm": 2.123802661895752, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.7009857296943665, + "num_tokens": 97877018.0, + "step": 3835 + }, + { + "epoch": 0.42126070722600484, + "grad_norm": 2.1754441261291504, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7146656513214111, + "num_tokens": 97903352.0, + "step": 3836 + }, + { + "epoch": 0.4213705249286185, + "grad_norm": 2.186005115509033, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6928513050079346, + "num_tokens": 97930748.0, + "step": 3837 + }, + { + "epoch": 0.42148034263123213, + "grad_norm": 2.121971368789673, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7163446545600891, + "num_tokens": 97958599.0, + "step": 3838 + }, + { + "epoch": 0.42159016033384583, + "grad_norm": 1.9462008476257324, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7104896306991577, + "num_tokens": 97987943.0, + "step": 3839 + }, + { + "epoch": 0.4216999780364595, + "grad_norm": 2.3623485565185547, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7030878663063049, + "num_tokens": 98011353.0, + "step": 3840 + }, + { + "epoch": 0.4218097957390731, + "grad_norm": 2.3075335025787354, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7067347764968872, + "num_tokens": 98033803.0, + "step": 3841 + }, + { + "epoch": 0.4219196134416868, + "grad_norm": 1.93156099319458, + "learning_rate": 1e-06, + "loss": 1.0747, + "mean_token_accuracy": 0.6825244426727295, + "num_tokens": 98068865.0, + "step": 3842 + }, + { + "epoch": 0.42202943114430047, + "grad_norm": 2.0179853439331055, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7208755016326904, + "num_tokens": 98098033.0, + "step": 3843 + }, + { + "epoch": 0.4221392488469141, + "grad_norm": 2.2043182849884033, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.701806366443634, + "num_tokens": 98122767.0, + "step": 3844 + }, + { + "epoch": 0.42224906654952776, + "grad_norm": 2.014185667037964, + "learning_rate": 1e-06, + "loss": 1.0654, + "mean_token_accuracy": 0.6819778084754944, + "num_tokens": 98152272.0, + "step": 3845 + }, + { + "epoch": 0.42235888425214146, + "grad_norm": 2.1994428634643555, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7009063959121704, + "num_tokens": 98179494.0, + "step": 3846 + }, + { + "epoch": 0.4224687019547551, + "grad_norm": 2.2090694904327393, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7147611379623413, + "num_tokens": 98202769.0, + "step": 3847 + }, + { + "epoch": 0.42257851965736876, + "grad_norm": 2.145962715148926, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.7093225717544556, + "num_tokens": 98229880.0, + "step": 3848 + }, + { + "epoch": 0.4226883373599824, + "grad_norm": 2.195457696914673, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6972644329071045, + "num_tokens": 98256921.0, + "step": 3849 + }, + { + "epoch": 0.4227981550625961, + "grad_norm": 2.126896858215332, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6990343332290649, + "num_tokens": 98284637.0, + "step": 3850 + }, + { + "epoch": 0.42290797276520975, + "grad_norm": 2.367408275604248, + "learning_rate": 1e-06, + "loss": 1.0901, + "mean_token_accuracy": 0.6857497692108154, + "num_tokens": 98308989.0, + "step": 3851 + }, + { + "epoch": 0.4230177904678234, + "grad_norm": 2.300356388092041, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7238364219665527, + "num_tokens": 98331200.0, + "step": 3852 + }, + { + "epoch": 0.4231276081704371, + "grad_norm": 2.244748115539551, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.6899807453155518, + "num_tokens": 98356275.0, + "step": 3853 + }, + { + "epoch": 0.42323742587305074, + "grad_norm": 2.242626428604126, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.705399751663208, + "num_tokens": 98381760.0, + "step": 3854 + }, + { + "epoch": 0.4233472435756644, + "grad_norm": 2.268585681915283, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7078238129615784, + "num_tokens": 98405183.0, + "step": 3855 + }, + { + "epoch": 0.42345706127827804, + "grad_norm": 2.3776469230651855, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7044410109519958, + "num_tokens": 98428370.0, + "step": 3856 + }, + { + "epoch": 0.42356687898089174, + "grad_norm": 2.2601287364959717, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7076104879379272, + "num_tokens": 98452075.0, + "step": 3857 + }, + { + "epoch": 0.4236766966835054, + "grad_norm": 2.2357053756713867, + "learning_rate": 1e-06, + "loss": 1.0376, + "mean_token_accuracy": 0.6929848790168762, + "num_tokens": 98478139.0, + "step": 3858 + }, + { + "epoch": 0.42378651438611903, + "grad_norm": 2.2513487339019775, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7198485136032104, + "num_tokens": 98502897.0, + "step": 3859 + }, + { + "epoch": 0.42389633208873273, + "grad_norm": 1.9753377437591553, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6949198842048645, + "num_tokens": 98533954.0, + "step": 3860 + }, + { + "epoch": 0.4240061497913464, + "grad_norm": 2.1022119522094727, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7041457295417786, + "num_tokens": 98559924.0, + "step": 3861 + }, + { + "epoch": 0.42411596749396, + "grad_norm": 2.111417770385742, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.7054202556610107, + "num_tokens": 98589547.0, + "step": 3862 + }, + { + "epoch": 0.42422578519657367, + "grad_norm": 2.6786463260650635, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7126818299293518, + "num_tokens": 98608519.0, + "step": 3863 + }, + { + "epoch": 0.42433560289918737, + "grad_norm": 2.014772415161133, + "learning_rate": 1e-06, + "loss": 1.0542, + "mean_token_accuracy": 0.678663969039917, + "num_tokens": 98639303.0, + "step": 3864 + }, + { + "epoch": 0.424445420601801, + "grad_norm": 2.347771644592285, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7045179605484009, + "num_tokens": 98662280.0, + "step": 3865 + }, + { + "epoch": 0.42455523830441466, + "grad_norm": 2.0985264778137207, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6927411556243896, + "num_tokens": 98692198.0, + "step": 3866 + }, + { + "epoch": 0.4246650560070283, + "grad_norm": 2.082422971725464, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6927849054336548, + "num_tokens": 98721674.0, + "step": 3867 + }, + { + "epoch": 0.424774873709642, + "grad_norm": 2.057567596435547, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7253385782241821, + "num_tokens": 98749520.0, + "step": 3868 + }, + { + "epoch": 0.42488469141225566, + "grad_norm": 2.3787660598754883, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7278538942337036, + "num_tokens": 98770375.0, + "step": 3869 + }, + { + "epoch": 0.4249945091148693, + "grad_norm": 2.0938215255737305, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.6958305835723877, + "num_tokens": 98797989.0, + "step": 3870 + }, + { + "epoch": 0.425104326817483, + "grad_norm": 2.40338397026062, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7138822078704834, + "num_tokens": 98821522.0, + "step": 3871 + }, + { + "epoch": 0.42521414452009665, + "grad_norm": 2.092926502227783, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.6997407078742981, + "num_tokens": 98849102.0, + "step": 3872 + }, + { + "epoch": 0.4253239622227103, + "grad_norm": 2.1824023723602295, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7055863738059998, + "num_tokens": 98875007.0, + "step": 3873 + }, + { + "epoch": 0.42543377992532394, + "grad_norm": 2.4640872478485107, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.7050254344940186, + "num_tokens": 98896235.0, + "step": 3874 + }, + { + "epoch": 0.42554359762793764, + "grad_norm": 2.1018903255462646, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6918612122535706, + "num_tokens": 98926085.0, + "step": 3875 + }, + { + "epoch": 0.4256534153305513, + "grad_norm": 1.9488400220870972, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6940597891807556, + "num_tokens": 98961104.0, + "step": 3876 + }, + { + "epoch": 0.42576323303316493, + "grad_norm": 2.1570568084716797, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.6997861862182617, + "num_tokens": 98986227.0, + "step": 3877 + }, + { + "epoch": 0.4258730507357786, + "grad_norm": 2.5407137870788574, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.717170238494873, + "num_tokens": 99005445.0, + "step": 3878 + }, + { + "epoch": 0.4259828684383923, + "grad_norm": 2.3308355808258057, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6860426664352417, + "num_tokens": 99029937.0, + "step": 3879 + }, + { + "epoch": 0.4260926861410059, + "grad_norm": 2.503614664077759, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.6878302693367004, + "num_tokens": 99052490.0, + "step": 3880 + }, + { + "epoch": 0.4262025038436196, + "grad_norm": 2.143606185913086, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7130440473556519, + "num_tokens": 99078443.0, + "step": 3881 + }, + { + "epoch": 0.4263123215462333, + "grad_norm": 2.1071274280548096, + "learning_rate": 1e-06, + "loss": 1.0137, + "mean_token_accuracy": 0.6951308250427246, + "num_tokens": 99104823.0, + "step": 3882 + }, + { + "epoch": 0.4264221392488469, + "grad_norm": 2.179865837097168, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6920367479324341, + "num_tokens": 99132812.0, + "step": 3883 + }, + { + "epoch": 0.42653195695146057, + "grad_norm": 2.4287476539611816, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7234655022621155, + "num_tokens": 99154294.0, + "step": 3884 + }, + { + "epoch": 0.4266417746540742, + "grad_norm": 2.266749858856201, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.6895478367805481, + "num_tokens": 99181707.0, + "step": 3885 + }, + { + "epoch": 0.4267515923566879, + "grad_norm": 2.04475474357605, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.715655505657196, + "num_tokens": 99208183.0, + "step": 3886 + }, + { + "epoch": 0.42686141005930156, + "grad_norm": 2.1526224613189697, + "learning_rate": 1e-06, + "loss": 1.1288, + "mean_token_accuracy": 0.6656522750854492, + "num_tokens": 99235963.0, + "step": 3887 + }, + { + "epoch": 0.4269712277619152, + "grad_norm": 2.2526655197143555, + "learning_rate": 1e-06, + "loss": 1.0399, + "mean_token_accuracy": 0.6846604943275452, + "num_tokens": 99261326.0, + "step": 3888 + }, + { + "epoch": 0.4270810454645289, + "grad_norm": 2.3226354122161865, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6959558725357056, + "num_tokens": 99285039.0, + "step": 3889 + }, + { + "epoch": 0.42719086316714255, + "grad_norm": 2.2457168102264404, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6909666061401367, + "num_tokens": 99310910.0, + "step": 3890 + }, + { + "epoch": 0.4273006808697562, + "grad_norm": 2.376600980758667, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7071793079376221, + "num_tokens": 99332777.0, + "step": 3891 + }, + { + "epoch": 0.42741049857236985, + "grad_norm": 2.4070236682891846, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7303431034088135, + "num_tokens": 99353708.0, + "step": 3892 + }, + { + "epoch": 0.42752031627498355, + "grad_norm": 2.40370774269104, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7108442783355713, + "num_tokens": 99376961.0, + "step": 3893 + }, + { + "epoch": 0.4276301339775972, + "grad_norm": 1.9581098556518555, + "learning_rate": 1e-06, + "loss": 1.097, + "mean_token_accuracy": 0.6763458847999573, + "num_tokens": 99410777.0, + "step": 3894 + }, + { + "epoch": 0.42773995168021084, + "grad_norm": 2.009025812149048, + "learning_rate": 1e-06, + "loss": 1.0196, + "mean_token_accuracy": 0.6977394819259644, + "num_tokens": 99442985.0, + "step": 3895 + }, + { + "epoch": 0.4278497693828245, + "grad_norm": 2.525559663772583, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7126258611679077, + "num_tokens": 99462621.0, + "step": 3896 + }, + { + "epoch": 0.4279595870854382, + "grad_norm": 2.579965829849243, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7017806172370911, + "num_tokens": 99482583.0, + "step": 3897 + }, + { + "epoch": 0.42806940478805183, + "grad_norm": 1.9481252431869507, + "learning_rate": 1e-06, + "loss": 1.0461, + "mean_token_accuracy": 0.6832141876220703, + "num_tokens": 99517437.0, + "step": 3898 + }, + { + "epoch": 0.4281792224906655, + "grad_norm": 2.3701517581939697, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7076502442359924, + "num_tokens": 99540434.0, + "step": 3899 + }, + { + "epoch": 0.4282890401932792, + "grad_norm": 2.0616235733032227, + "learning_rate": 1e-06, + "loss": 1.1151, + "mean_token_accuracy": 0.6717287302017212, + "num_tokens": 99573106.0, + "step": 3900 + }, + { + "epoch": 0.4283988578958928, + "grad_norm": 2.721621036529541, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7203371524810791, + "num_tokens": 99590578.0, + "step": 3901 + }, + { + "epoch": 0.42850867559850647, + "grad_norm": 1.8597837686538696, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7132161855697632, + "num_tokens": 99624302.0, + "step": 3902 + }, + { + "epoch": 0.4286184933011201, + "grad_norm": 2.298097610473633, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6976983547210693, + "num_tokens": 99648489.0, + "step": 3903 + }, + { + "epoch": 0.4287283110037338, + "grad_norm": 2.206705331802368, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7210960388183594, + "num_tokens": 99674893.0, + "step": 3904 + }, + { + "epoch": 0.42883812870634747, + "grad_norm": 2.226271629333496, + "learning_rate": 1e-06, + "loss": 1.0507, + "mean_token_accuracy": 0.6820395588874817, + "num_tokens": 99702147.0, + "step": 3905 + }, + { + "epoch": 0.4289479464089611, + "grad_norm": 2.0771052837371826, + "learning_rate": 1e-06, + "loss": 1.0669, + "mean_token_accuracy": 0.6828333139419556, + "num_tokens": 99731284.0, + "step": 3906 + }, + { + "epoch": 0.4290577641115748, + "grad_norm": 2.0099902153015137, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.6995246410369873, + "num_tokens": 99762676.0, + "step": 3907 + }, + { + "epoch": 0.42916758181418846, + "grad_norm": 2.4253807067871094, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7081935405731201, + "num_tokens": 99785173.0, + "step": 3908 + }, + { + "epoch": 0.4292773995168021, + "grad_norm": 2.1394450664520264, + "learning_rate": 1e-06, + "loss": 1.0561, + "mean_token_accuracy": 0.6818372011184692, + "num_tokens": 99812763.0, + "step": 3909 + }, + { + "epoch": 0.42938721721941575, + "grad_norm": 2.231102228164673, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7309715747833252, + "num_tokens": 99835778.0, + "step": 3910 + }, + { + "epoch": 0.42949703492202945, + "grad_norm": 2.455115795135498, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7212285399436951, + "num_tokens": 99857840.0, + "step": 3911 + }, + { + "epoch": 0.4296068526246431, + "grad_norm": 2.357673168182373, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.712890625, + "num_tokens": 99880935.0, + "step": 3912 + }, + { + "epoch": 0.42971667032725674, + "grad_norm": 2.0535526275634766, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6928280591964722, + "num_tokens": 99907910.0, + "step": 3913 + }, + { + "epoch": 0.4298264880298704, + "grad_norm": 2.4959287643432617, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7090733051300049, + "num_tokens": 99928822.0, + "step": 3914 + }, + { + "epoch": 0.4299363057324841, + "grad_norm": 1.9416013956069946, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6929539442062378, + "num_tokens": 99961812.0, + "step": 3915 + }, + { + "epoch": 0.43004612343509774, + "grad_norm": 2.2823402881622314, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7207550406455994, + "num_tokens": 99985790.0, + "step": 3916 + }, + { + "epoch": 0.4301559411377114, + "grad_norm": 2.067789077758789, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7146493196487427, + "num_tokens": 100012593.0, + "step": 3917 + }, + { + "epoch": 0.4302657588403251, + "grad_norm": 2.007143259048462, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6981939077377319, + "num_tokens": 100044155.0, + "step": 3918 + }, + { + "epoch": 0.43037557654293873, + "grad_norm": 2.6280391216278076, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.700523316860199, + "num_tokens": 100063681.0, + "step": 3919 + }, + { + "epoch": 0.4304853942455524, + "grad_norm": 2.327781915664673, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6992851495742798, + "num_tokens": 100088784.0, + "step": 3920 + }, + { + "epoch": 0.430595211948166, + "grad_norm": 2.3955445289611816, + "learning_rate": 1e-06, + "loss": 1.0613, + "mean_token_accuracy": 0.6769982576370239, + "num_tokens": 100113533.0, + "step": 3921 + }, + { + "epoch": 0.4307050296507797, + "grad_norm": 2.1936511993408203, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.6994061470031738, + "num_tokens": 100141001.0, + "step": 3922 + }, + { + "epoch": 0.43081484735339337, + "grad_norm": 2.0232701301574707, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6947110891342163, + "num_tokens": 100170746.0, + "step": 3923 + }, + { + "epoch": 0.430924665056007, + "grad_norm": 2.1211180686950684, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6985607147216797, + "num_tokens": 100198057.0, + "step": 3924 + }, + { + "epoch": 0.43103448275862066, + "grad_norm": 2.2176706790924072, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7148234248161316, + "num_tokens": 100222312.0, + "step": 3925 + }, + { + "epoch": 0.43114430046123436, + "grad_norm": 2.0085902214050293, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7076802253723145, + "num_tokens": 100253400.0, + "step": 3926 + }, + { + "epoch": 0.431254118163848, + "grad_norm": 2.2484781742095947, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.6966837644577026, + "num_tokens": 100278997.0, + "step": 3927 + }, + { + "epoch": 0.43136393586646166, + "grad_norm": 2.5263872146606445, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6931349039077759, + "num_tokens": 100301544.0, + "step": 3928 + }, + { + "epoch": 0.43147375356907536, + "grad_norm": 2.3621840476989746, + "learning_rate": 1e-06, + "loss": 1.0358, + "mean_token_accuracy": 0.6874022483825684, + "num_tokens": 100324795.0, + "step": 3929 + }, + { + "epoch": 0.431583571271689, + "grad_norm": 2.0431711673736572, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6988502144813538, + "num_tokens": 100355537.0, + "step": 3930 + }, + { + "epoch": 0.43169338897430265, + "grad_norm": 2.6251893043518066, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7167716026306152, + "num_tokens": 100374318.0, + "step": 3931 + }, + { + "epoch": 0.4318032066769163, + "grad_norm": 2.058345079421997, + "learning_rate": 1e-06, + "loss": 1.0758, + "mean_token_accuracy": 0.678369402885437, + "num_tokens": 100405836.0, + "step": 3932 + }, + { + "epoch": 0.43191302437953, + "grad_norm": 2.546093225479126, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7026157379150391, + "num_tokens": 100425500.0, + "step": 3933 + }, + { + "epoch": 0.43202284208214364, + "grad_norm": 2.335297107696533, + "learning_rate": 1e-06, + "loss": 1.0221, + "mean_token_accuracy": 0.7004136443138123, + "num_tokens": 100451248.0, + "step": 3934 + }, + { + "epoch": 0.4321326597847573, + "grad_norm": 2.091458559036255, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.725040078163147, + "num_tokens": 100477915.0, + "step": 3935 + }, + { + "epoch": 0.432242477487371, + "grad_norm": 2.468079090118408, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7031362056732178, + "num_tokens": 100496436.0, + "step": 3936 + }, + { + "epoch": 0.43235229518998464, + "grad_norm": 2.4676995277404785, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.6963714361190796, + "num_tokens": 100518088.0, + "step": 3937 + }, + { + "epoch": 0.4324621128925983, + "grad_norm": 2.300722599029541, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7138563990592957, + "num_tokens": 100541102.0, + "step": 3938 + }, + { + "epoch": 0.43257193059521193, + "grad_norm": 2.419306516647339, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6939529180526733, + "num_tokens": 100562670.0, + "step": 3939 + }, + { + "epoch": 0.43268174829782563, + "grad_norm": 2.060609817504883, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7191540598869324, + "num_tokens": 100589553.0, + "step": 3940 + }, + { + "epoch": 0.4327915660004393, + "grad_norm": 2.4588093757629395, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.6953446865081787, + "num_tokens": 100611388.0, + "step": 3941 + }, + { + "epoch": 0.4329013837030529, + "grad_norm": 2.498145818710327, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7085967063903809, + "num_tokens": 100630475.0, + "step": 3942 + }, + { + "epoch": 0.43301120140566657, + "grad_norm": 2.3511221408843994, + "learning_rate": 1e-06, + "loss": 1.0236, + "mean_token_accuracy": 0.6956071853637695, + "num_tokens": 100653956.0, + "step": 3943 + }, + { + "epoch": 0.43312101910828027, + "grad_norm": 2.3303065299987793, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7089213132858276, + "num_tokens": 100676971.0, + "step": 3944 + }, + { + "epoch": 0.4332308368108939, + "grad_norm": 2.413205862045288, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6959163546562195, + "num_tokens": 100700722.0, + "step": 3945 + }, + { + "epoch": 0.43334065451350756, + "grad_norm": 2.5571072101593018, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7184929847717285, + "num_tokens": 100719607.0, + "step": 3946 + }, + { + "epoch": 0.43345047221612126, + "grad_norm": 2.202410936355591, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.6980150938034058, + "num_tokens": 100745262.0, + "step": 3947 + }, + { + "epoch": 0.4335602899187349, + "grad_norm": 2.210556983947754, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7145982980728149, + "num_tokens": 100771109.0, + "step": 3948 + }, + { + "epoch": 0.43367010762134856, + "grad_norm": 2.4189090728759766, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7096385359764099, + "num_tokens": 100793445.0, + "step": 3949 + }, + { + "epoch": 0.4337799253239622, + "grad_norm": 2.336266279220581, + "learning_rate": 1e-06, + "loss": 1.0236, + "mean_token_accuracy": 0.6894848942756653, + "num_tokens": 100817016.0, + "step": 3950 + }, + { + "epoch": 0.4338897430265759, + "grad_norm": 2.3465688228607178, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7206819653511047, + "num_tokens": 100838490.0, + "step": 3951 + }, + { + "epoch": 0.43399956072918955, + "grad_norm": 2.2832648754119873, + "learning_rate": 1e-06, + "loss": 1.1398, + "mean_token_accuracy": 0.6582134366035461, + "num_tokens": 100869830.0, + "step": 3952 + }, + { + "epoch": 0.4341093784318032, + "grad_norm": 2.7309277057647705, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.717584490776062, + "num_tokens": 100887776.0, + "step": 3953 + }, + { + "epoch": 0.43421919613441684, + "grad_norm": 2.179213285446167, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6935377717018127, + "num_tokens": 100914046.0, + "step": 3954 + }, + { + "epoch": 0.43432901383703054, + "grad_norm": 2.144092321395874, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7239903211593628, + "num_tokens": 100938419.0, + "step": 3955 + }, + { + "epoch": 0.4344388315396442, + "grad_norm": 2.5399246215820312, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7145854830741882, + "num_tokens": 100957483.0, + "step": 3956 + }, + { + "epoch": 0.43454864924225783, + "grad_norm": 2.1930694580078125, + "learning_rate": 1e-06, + "loss": 1.0506, + "mean_token_accuracy": 0.6838817596435547, + "num_tokens": 100985709.0, + "step": 3957 + }, + { + "epoch": 0.43465846694487154, + "grad_norm": 2.354142427444458, + "learning_rate": 1e-06, + "loss": 1.0549, + "mean_token_accuracy": 0.683951199054718, + "num_tokens": 101008811.0, + "step": 3958 + }, + { + "epoch": 0.4347682846474852, + "grad_norm": 2.0485785007476807, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7089450359344482, + "num_tokens": 101037234.0, + "step": 3959 + }, + { + "epoch": 0.4348781023500988, + "grad_norm": 2.223135471343994, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7027643322944641, + "num_tokens": 101064534.0, + "step": 3960 + }, + { + "epoch": 0.4349879200527125, + "grad_norm": 2.2848761081695557, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6852781772613525, + "num_tokens": 101088879.0, + "step": 3961 + }, + { + "epoch": 0.4350977377553262, + "grad_norm": 2.589918613433838, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7038427591323853, + "num_tokens": 101108779.0, + "step": 3962 + }, + { + "epoch": 0.4352075554579398, + "grad_norm": 2.4777798652648926, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7316568493843079, + "num_tokens": 101128619.0, + "step": 3963 + }, + { + "epoch": 0.43531737316055347, + "grad_norm": 2.19728946685791, + "learning_rate": 1e-06, + "loss": 1.0125, + "mean_token_accuracy": 0.6966262459754944, + "num_tokens": 101154016.0, + "step": 3964 + }, + { + "epoch": 0.43542719086316717, + "grad_norm": 2.2098493576049805, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7028675079345703, + "num_tokens": 101179833.0, + "step": 3965 + }, + { + "epoch": 0.4355370085657808, + "grad_norm": 2.2135417461395264, + "learning_rate": 1e-06, + "loss": 1.042, + "mean_token_accuracy": 0.6858842372894287, + "num_tokens": 101206779.0, + "step": 3966 + }, + { + "epoch": 0.43564682626839446, + "grad_norm": 2.109117031097412, + "learning_rate": 1e-06, + "loss": 1.0769, + "mean_token_accuracy": 0.6779645681381226, + "num_tokens": 101236355.0, + "step": 3967 + }, + { + "epoch": 0.4357566439710081, + "grad_norm": 2.1379823684692383, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.715766191482544, + "num_tokens": 101263147.0, + "step": 3968 + }, + { + "epoch": 0.4358664616736218, + "grad_norm": 2.1548988819122314, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.690022885799408, + "num_tokens": 101290975.0, + "step": 3969 + }, + { + "epoch": 0.43597627937623545, + "grad_norm": 2.293503999710083, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.688854455947876, + "num_tokens": 101316623.0, + "step": 3970 + }, + { + "epoch": 0.4360860970788491, + "grad_norm": 2.6321845054626465, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7223540544509888, + "num_tokens": 101334658.0, + "step": 3971 + }, + { + "epoch": 0.43619591478146275, + "grad_norm": 2.1309778690338135, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7047187089920044, + "num_tokens": 101359597.0, + "step": 3972 + }, + { + "epoch": 0.43630573248407645, + "grad_norm": 1.9642302989959717, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.7018769979476929, + "num_tokens": 101390940.0, + "step": 3973 + }, + { + "epoch": 0.4364155501866901, + "grad_norm": 2.10758376121521, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7046778798103333, + "num_tokens": 101417619.0, + "step": 3974 + }, + { + "epoch": 0.43652536788930374, + "grad_norm": 2.235246419906616, + "learning_rate": 1e-06, + "loss": 1.1333, + "mean_token_accuracy": 0.6695955395698547, + "num_tokens": 101442805.0, + "step": 3975 + }, + { + "epoch": 0.43663518559191744, + "grad_norm": 2.3964474201202393, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7135182619094849, + "num_tokens": 101464843.0, + "step": 3976 + }, + { + "epoch": 0.4367450032945311, + "grad_norm": 2.4825804233551025, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7226719856262207, + "num_tokens": 101486840.0, + "step": 3977 + }, + { + "epoch": 0.43685482099714473, + "grad_norm": 2.153306722640991, + "learning_rate": 1e-06, + "loss": 1.041, + "mean_token_accuracy": 0.6848410367965698, + "num_tokens": 101516009.0, + "step": 3978 + }, + { + "epoch": 0.4369646386997584, + "grad_norm": 2.4751980304718018, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7159528732299805, + "num_tokens": 101535530.0, + "step": 3979 + }, + { + "epoch": 0.4370744564023721, + "grad_norm": 2.3011903762817383, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7278224229812622, + "num_tokens": 101556942.0, + "step": 3980 + }, + { + "epoch": 0.4371842741049857, + "grad_norm": 2.175583839416504, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7075586318969727, + "num_tokens": 101581922.0, + "step": 3981 + }, + { + "epoch": 0.43729409180759937, + "grad_norm": 2.1503779888153076, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7114322781562805, + "num_tokens": 101609680.0, + "step": 3982 + }, + { + "epoch": 0.4374039095102131, + "grad_norm": 2.2777864933013916, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.701803982257843, + "num_tokens": 101633426.0, + "step": 3983 + }, + { + "epoch": 0.4375137272128267, + "grad_norm": 2.162504196166992, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.706271767616272, + "num_tokens": 101659018.0, + "step": 3984 + }, + { + "epoch": 0.43762354491544037, + "grad_norm": 2.443300247192383, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7044415473937988, + "num_tokens": 101681286.0, + "step": 3985 + }, + { + "epoch": 0.437733362618054, + "grad_norm": 2.1588239669799805, + "learning_rate": 1e-06, + "loss": 1.0586, + "mean_token_accuracy": 0.6884462833404541, + "num_tokens": 101707902.0, + "step": 3986 + }, + { + "epoch": 0.4378431803206677, + "grad_norm": 2.301846742630005, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6864911317825317, + "num_tokens": 101731709.0, + "step": 3987 + }, + { + "epoch": 0.43795299802328136, + "grad_norm": 2.440150022506714, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6987159252166748, + "num_tokens": 101753897.0, + "step": 3988 + }, + { + "epoch": 0.438062815725895, + "grad_norm": 2.2132863998413086, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7284529805183411, + "num_tokens": 101779349.0, + "step": 3989 + }, + { + "epoch": 0.43817263342850865, + "grad_norm": 2.2410550117492676, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6876912713050842, + "num_tokens": 101804044.0, + "step": 3990 + }, + { + "epoch": 0.43828245113112235, + "grad_norm": 2.190478801727295, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7179416418075562, + "num_tokens": 101829524.0, + "step": 3991 + }, + { + "epoch": 0.438392268833736, + "grad_norm": 2.129728078842163, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7073338031768799, + "num_tokens": 101856433.0, + "step": 3992 + }, + { + "epoch": 0.43850208653634964, + "grad_norm": 2.2529819011688232, + "learning_rate": 1e-06, + "loss": 1.0175, + "mean_token_accuracy": 0.6914716958999634, + "num_tokens": 101882041.0, + "step": 3993 + }, + { + "epoch": 0.43861190423896335, + "grad_norm": 1.823614478111267, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7063457369804382, + "num_tokens": 101917508.0, + "step": 3994 + }, + { + "epoch": 0.438721721941577, + "grad_norm": 2.536797523498535, + "learning_rate": 1e-06, + "loss": 1.0528, + "mean_token_accuracy": 0.7040651440620422, + "num_tokens": 101938323.0, + "step": 3995 + }, + { + "epoch": 0.43883153964419064, + "grad_norm": 2.4086267948150635, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7152172327041626, + "num_tokens": 101959654.0, + "step": 3996 + }, + { + "epoch": 0.4389413573468043, + "grad_norm": 2.491605758666992, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7081006169319153, + "num_tokens": 101979496.0, + "step": 3997 + }, + { + "epoch": 0.439051175049418, + "grad_norm": 2.3834457397460938, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6810922026634216, + "num_tokens": 102003486.0, + "step": 3998 + }, + { + "epoch": 0.43916099275203163, + "grad_norm": 2.4533703327178955, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7064505219459534, + "num_tokens": 102024809.0, + "step": 3999 + }, + { + "epoch": 0.4392708104546453, + "grad_norm": 2.191732883453369, + "learning_rate": 1e-06, + "loss": 1.1348, + "mean_token_accuracy": 0.6716524958610535, + "num_tokens": 102053591.0, + "step": 4000 + }, + { + "epoch": 0.4393806281572589, + "grad_norm": 2.245619058609009, + "learning_rate": 1e-06, + "loss": 1.0867, + "mean_token_accuracy": 0.6729896068572998, + "num_tokens": 102080419.0, + "step": 4001 + }, + { + "epoch": 0.4394904458598726, + "grad_norm": 2.4153010845184326, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.6864710450172424, + "num_tokens": 102102106.0, + "step": 4002 + }, + { + "epoch": 0.43960026356248627, + "grad_norm": 2.1621735095977783, + "learning_rate": 1e-06, + "loss": 1.0588, + "mean_token_accuracy": 0.6866258382797241, + "num_tokens": 102131059.0, + "step": 4003 + }, + { + "epoch": 0.4397100812650999, + "grad_norm": 2.4776930809020996, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.700393795967102, + "num_tokens": 102151974.0, + "step": 4004 + }, + { + "epoch": 0.4398198989677136, + "grad_norm": 2.24357008934021, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6898611187934875, + "num_tokens": 102177140.0, + "step": 4005 + }, + { + "epoch": 0.43992971667032726, + "grad_norm": 2.2756760120391846, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.7031317353248596, + "num_tokens": 102202228.0, + "step": 4006 + }, + { + "epoch": 0.4400395343729409, + "grad_norm": 2.0214977264404297, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7104434967041016, + "num_tokens": 102231282.0, + "step": 4007 + }, + { + "epoch": 0.44014935207555456, + "grad_norm": 2.698474884033203, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6855006814002991, + "num_tokens": 102250584.0, + "step": 4008 + }, + { + "epoch": 0.44025916977816826, + "grad_norm": 2.188375234603882, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.701803982257843, + "num_tokens": 102276794.0, + "step": 4009 + }, + { + "epoch": 0.4403689874807819, + "grad_norm": 2.035491943359375, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.6945570707321167, + "num_tokens": 102307867.0, + "step": 4010 + }, + { + "epoch": 0.44047880518339555, + "grad_norm": 2.1800589561462402, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6951892375946045, + "num_tokens": 102336460.0, + "step": 4011 + }, + { + "epoch": 0.44058862288600925, + "grad_norm": 2.134281635284424, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.6998690366744995, + "num_tokens": 102361453.0, + "step": 4012 + }, + { + "epoch": 0.4406984405886229, + "grad_norm": 2.484337329864502, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.7022117972373962, + "num_tokens": 102381183.0, + "step": 4013 + }, + { + "epoch": 0.44080825829123654, + "grad_norm": 2.2160041332244873, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.696570634841919, + "num_tokens": 102405954.0, + "step": 4014 + }, + { + "epoch": 0.4409180759938502, + "grad_norm": 2.0334298610687256, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6955274939537048, + "num_tokens": 102436045.0, + "step": 4015 + }, + { + "epoch": 0.4410278936964639, + "grad_norm": 2.3005940914154053, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7222611308097839, + "num_tokens": 102459418.0, + "step": 4016 + }, + { + "epoch": 0.44113771139907754, + "grad_norm": 2.371920347213745, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7472224235534668, + "num_tokens": 102480888.0, + "step": 4017 + }, + { + "epoch": 0.4412475291016912, + "grad_norm": 1.9146672487258911, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6953807473182678, + "num_tokens": 102512670.0, + "step": 4018 + }, + { + "epoch": 0.44135734680430483, + "grad_norm": 2.2522242069244385, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.6962276697158813, + "num_tokens": 102538632.0, + "step": 4019 + }, + { + "epoch": 0.44146716450691853, + "grad_norm": 2.1766061782836914, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.7465587258338928, + "num_tokens": 102564649.0, + "step": 4020 + }, + { + "epoch": 0.4415769822095322, + "grad_norm": 2.676375389099121, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7215152978897095, + "num_tokens": 102581392.0, + "step": 4021 + }, + { + "epoch": 0.4416867999121458, + "grad_norm": 2.156985282897949, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.7016704678535461, + "num_tokens": 102606944.0, + "step": 4022 + }, + { + "epoch": 0.4417966176147595, + "grad_norm": 2.096208333969116, + "learning_rate": 1e-06, + "loss": 1.0479, + "mean_token_accuracy": 0.6807550191879272, + "num_tokens": 102636060.0, + "step": 4023 + }, + { + "epoch": 0.44190643531737317, + "grad_norm": 2.486398220062256, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.6973938941955566, + "num_tokens": 102657089.0, + "step": 4024 + }, + { + "epoch": 0.4420162530199868, + "grad_norm": 2.2311511039733887, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6954954266548157, + "num_tokens": 102683570.0, + "step": 4025 + }, + { + "epoch": 0.44212607072260046, + "grad_norm": 2.3571078777313232, + "learning_rate": 1e-06, + "loss": 1.0525, + "mean_token_accuracy": 0.6811952590942383, + "num_tokens": 102708628.0, + "step": 4026 + }, + { + "epoch": 0.44223588842521416, + "grad_norm": 2.597168445587158, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7069442272186279, + "num_tokens": 102728375.0, + "step": 4027 + }, + { + "epoch": 0.4423457061278278, + "grad_norm": 2.3157241344451904, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7136989235877991, + "num_tokens": 102750481.0, + "step": 4028 + }, + { + "epoch": 0.44245552383044146, + "grad_norm": 2.344827175140381, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.7054082155227661, + "num_tokens": 102773464.0, + "step": 4029 + }, + { + "epoch": 0.4425653415330551, + "grad_norm": 1.9643806219100952, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7163840532302856, + "num_tokens": 102805083.0, + "step": 4030 + }, + { + "epoch": 0.4426751592356688, + "grad_norm": 2.0902624130249023, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6906906962394714, + "num_tokens": 102835091.0, + "step": 4031 + }, + { + "epoch": 0.44278497693828245, + "grad_norm": 2.0203754901885986, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6987121105194092, + "num_tokens": 102864426.0, + "step": 4032 + }, + { + "epoch": 0.4428947946408961, + "grad_norm": 2.6093130111694336, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.691460132598877, + "num_tokens": 102882883.0, + "step": 4033 + }, + { + "epoch": 0.4430046123435098, + "grad_norm": 2.288590431213379, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.7056066989898682, + "num_tokens": 102906700.0, + "step": 4034 + }, + { + "epoch": 0.44311443004612344, + "grad_norm": 2.2008702754974365, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7007535099983215, + "num_tokens": 102932608.0, + "step": 4035 + }, + { + "epoch": 0.4432242477487371, + "grad_norm": 2.277269124984741, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7086989879608154, + "num_tokens": 102957219.0, + "step": 4036 + }, + { + "epoch": 0.44333406545135073, + "grad_norm": 2.1561996936798096, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6951558589935303, + "num_tokens": 102984573.0, + "step": 4037 + }, + { + "epoch": 0.44344388315396444, + "grad_norm": 2.34110951423645, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6960319876670837, + "num_tokens": 103009336.0, + "step": 4038 + }, + { + "epoch": 0.4435537008565781, + "grad_norm": 2.3977723121643066, + "learning_rate": 1e-06, + "loss": 1.077, + "mean_token_accuracy": 0.6869858503341675, + "num_tokens": 103031241.0, + "step": 4039 + }, + { + "epoch": 0.4436635185591917, + "grad_norm": 1.7672268152236938, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7034521102905273, + "num_tokens": 103066862.0, + "step": 4040 + }, + { + "epoch": 0.44377333626180543, + "grad_norm": 2.427351236343384, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7179531455039978, + "num_tokens": 103090251.0, + "step": 4041 + }, + { + "epoch": 0.4438831539644191, + "grad_norm": 2.336052656173706, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.6997726559638977, + "num_tokens": 103113883.0, + "step": 4042 + }, + { + "epoch": 0.4439929716670327, + "grad_norm": 2.1624112129211426, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6917305588722229, + "num_tokens": 103140130.0, + "step": 4043 + }, + { + "epoch": 0.44410278936964637, + "grad_norm": 2.3368232250213623, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7291392087936401, + "num_tokens": 103160664.0, + "step": 4044 + }, + { + "epoch": 0.44421260707226007, + "grad_norm": 2.3432505130767822, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7098058462142944, + "num_tokens": 103184179.0, + "step": 4045 + }, + { + "epoch": 0.4443224247748737, + "grad_norm": 2.4552688598632812, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.690988302230835, + "num_tokens": 103206967.0, + "step": 4046 + }, + { + "epoch": 0.44443224247748736, + "grad_norm": 2.2229816913604736, + "learning_rate": 1e-06, + "loss": 1.045, + "mean_token_accuracy": 0.6928497552871704, + "num_tokens": 103234193.0, + "step": 4047 + }, + { + "epoch": 0.444542060180101, + "grad_norm": 2.319617986679077, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7042391300201416, + "num_tokens": 103256017.0, + "step": 4048 + }, + { + "epoch": 0.4446518778827147, + "grad_norm": 2.062224864959717, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.704978346824646, + "num_tokens": 103285768.0, + "step": 4049 + }, + { + "epoch": 0.44476169558532835, + "grad_norm": 2.2604708671569824, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6981662511825562, + "num_tokens": 103310181.0, + "step": 4050 + }, + { + "epoch": 0.444871513287942, + "grad_norm": 2.3033039569854736, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.691798746585846, + "num_tokens": 103334458.0, + "step": 4051 + }, + { + "epoch": 0.4449813309905557, + "grad_norm": 2.2721667289733887, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7039883136749268, + "num_tokens": 103358784.0, + "step": 4052 + }, + { + "epoch": 0.44509114869316935, + "grad_norm": 2.096196413040161, + "learning_rate": 1e-06, + "loss": 1.0234, + "mean_token_accuracy": 0.6921359300613403, + "num_tokens": 103387051.0, + "step": 4053 + }, + { + "epoch": 0.445200966395783, + "grad_norm": 2.447749614715576, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.717564582824707, + "num_tokens": 103414243.0, + "step": 4054 + }, + { + "epoch": 0.44531078409839664, + "grad_norm": 2.306795358657837, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.707596480846405, + "num_tokens": 103438867.0, + "step": 4055 + }, + { + "epoch": 0.44542060180101034, + "grad_norm": 2.122474193572998, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7148204445838928, + "num_tokens": 103466420.0, + "step": 4056 + }, + { + "epoch": 0.445530419503624, + "grad_norm": 2.2401373386383057, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.6883102655410767, + "num_tokens": 103490828.0, + "step": 4057 + }, + { + "epoch": 0.44564023720623763, + "grad_norm": 2.41430926322937, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7094902396202087, + "num_tokens": 103513843.0, + "step": 4058 + }, + { + "epoch": 0.44575005490885133, + "grad_norm": 2.5005617141723633, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.683759331703186, + "num_tokens": 103536385.0, + "step": 4059 + }, + { + "epoch": 0.445859872611465, + "grad_norm": 2.168369770050049, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6950567960739136, + "num_tokens": 103564543.0, + "step": 4060 + }, + { + "epoch": 0.4459696903140786, + "grad_norm": 2.2940056324005127, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7017505764961243, + "num_tokens": 103588951.0, + "step": 4061 + }, + { + "epoch": 0.44607950801669227, + "grad_norm": 2.2039523124694824, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7096578478813171, + "num_tokens": 103614937.0, + "step": 4062 + }, + { + "epoch": 0.446189325719306, + "grad_norm": 2.3142082691192627, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7052954435348511, + "num_tokens": 103638589.0, + "step": 4063 + }, + { + "epoch": 0.4462991434219196, + "grad_norm": 2.401367425918579, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6932892799377441, + "num_tokens": 103662844.0, + "step": 4064 + }, + { + "epoch": 0.44640896112453327, + "grad_norm": 2.0977365970611572, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7040789127349854, + "num_tokens": 103690107.0, + "step": 4065 + }, + { + "epoch": 0.4465187788271469, + "grad_norm": 2.068617820739746, + "learning_rate": 1e-06, + "loss": 1.0434, + "mean_token_accuracy": 0.6827445030212402, + "num_tokens": 103718302.0, + "step": 4066 + }, + { + "epoch": 0.4466285965297606, + "grad_norm": 2.083984851837158, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.6843400597572327, + "num_tokens": 103747025.0, + "step": 4067 + }, + { + "epoch": 0.44673841423237426, + "grad_norm": 2.3895199298858643, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7054101824760437, + "num_tokens": 103769601.0, + "step": 4068 + }, + { + "epoch": 0.4468482319349879, + "grad_norm": 2.6830008029937744, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7370631694793701, + "num_tokens": 103789090.0, + "step": 4069 + }, + { + "epoch": 0.4469580496376016, + "grad_norm": 2.151670217514038, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.7026524543762207, + "num_tokens": 103817985.0, + "step": 4070 + }, + { + "epoch": 0.44706786734021525, + "grad_norm": 2.364243268966675, + "learning_rate": 1e-06, + "loss": 1.0716, + "mean_token_accuracy": 0.6807767748832703, + "num_tokens": 103842848.0, + "step": 4071 + }, + { + "epoch": 0.4471776850428289, + "grad_norm": 2.158433675765991, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7034456729888916, + "num_tokens": 103868042.0, + "step": 4072 + }, + { + "epoch": 0.44728750274544254, + "grad_norm": 2.3110482692718506, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7027395963668823, + "num_tokens": 103891522.0, + "step": 4073 + }, + { + "epoch": 0.44739732044805625, + "grad_norm": 2.5162432193756104, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.703487753868103, + "num_tokens": 103913824.0, + "step": 4074 + }, + { + "epoch": 0.4475071381506699, + "grad_norm": 2.415985584259033, + "learning_rate": 1e-06, + "loss": 1.0873, + "mean_token_accuracy": 0.6742101907730103, + "num_tokens": 103938271.0, + "step": 4075 + }, + { + "epoch": 0.44761695585328354, + "grad_norm": 2.401468276977539, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7171013355255127, + "num_tokens": 103961571.0, + "step": 4076 + }, + { + "epoch": 0.4477267735558972, + "grad_norm": 2.5835468769073486, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7165215611457825, + "num_tokens": 103982077.0, + "step": 4077 + }, + { + "epoch": 0.4478365912585109, + "grad_norm": 2.3024282455444336, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7056148052215576, + "num_tokens": 104006938.0, + "step": 4078 + }, + { + "epoch": 0.44794640896112453, + "grad_norm": 2.1000425815582275, + "learning_rate": 1e-06, + "loss": 1.0823, + "mean_token_accuracy": 0.6853656768798828, + "num_tokens": 104037511.0, + "step": 4079 + }, + { + "epoch": 0.4480562266637382, + "grad_norm": 2.329988479614258, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7023745179176331, + "num_tokens": 104062068.0, + "step": 4080 + }, + { + "epoch": 0.4481660443663519, + "grad_norm": 2.4569520950317383, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7110655903816223, + "num_tokens": 104082698.0, + "step": 4081 + }, + { + "epoch": 0.4482758620689655, + "grad_norm": 2.2264654636383057, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.6876527070999146, + "num_tokens": 104107297.0, + "step": 4082 + }, + { + "epoch": 0.44838567977157917, + "grad_norm": 2.233654260635376, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7062340378761292, + "num_tokens": 104133126.0, + "step": 4083 + }, + { + "epoch": 0.4484954974741928, + "grad_norm": 2.176046848297119, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.708897590637207, + "num_tokens": 104158396.0, + "step": 4084 + }, + { + "epoch": 0.4486053151768065, + "grad_norm": 2.136949062347412, + "learning_rate": 1e-06, + "loss": 1.0285, + "mean_token_accuracy": 0.6944440007209778, + "num_tokens": 104185663.0, + "step": 4085 + }, + { + "epoch": 0.44871513287942016, + "grad_norm": 2.193352699279785, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.6961573362350464, + "num_tokens": 104211671.0, + "step": 4086 + }, + { + "epoch": 0.4488249505820338, + "grad_norm": 2.2424018383026123, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7162952423095703, + "num_tokens": 104236574.0, + "step": 4087 + }, + { + "epoch": 0.4489347682846475, + "grad_norm": 2.2571706771850586, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7038172483444214, + "num_tokens": 104259450.0, + "step": 4088 + }, + { + "epoch": 0.44904458598726116, + "grad_norm": 2.5106582641601562, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7207778692245483, + "num_tokens": 104279512.0, + "step": 4089 + }, + { + "epoch": 0.4491544036898748, + "grad_norm": 2.5318713188171387, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6985360980033875, + "num_tokens": 104300851.0, + "step": 4090 + }, + { + "epoch": 0.44926422139248845, + "grad_norm": 2.1407129764556885, + "learning_rate": 1e-06, + "loss": 1.044, + "mean_token_accuracy": 0.6842015981674194, + "num_tokens": 104329462.0, + "step": 4091 + }, + { + "epoch": 0.44937403909510215, + "grad_norm": 1.8169186115264893, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.699271023273468, + "num_tokens": 104366977.0, + "step": 4092 + }, + { + "epoch": 0.4494838567977158, + "grad_norm": 2.1772425174713135, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7168766856193542, + "num_tokens": 104394676.0, + "step": 4093 + }, + { + "epoch": 0.44959367450032944, + "grad_norm": 2.2207188606262207, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7073009014129639, + "num_tokens": 104419854.0, + "step": 4094 + }, + { + "epoch": 0.4497034922029431, + "grad_norm": 2.028289318084717, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7075436115264893, + "num_tokens": 104449544.0, + "step": 4095 + }, + { + "epoch": 0.4498133099055568, + "grad_norm": 2.168022394180298, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7338526248931885, + "num_tokens": 104474036.0, + "step": 4096 + }, + { + "epoch": 0.44992312760817044, + "grad_norm": 2.2777199745178223, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7076489925384521, + "num_tokens": 104500084.0, + "step": 4097 + }, + { + "epoch": 0.4500329453107841, + "grad_norm": 2.170135259628296, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7214965224266052, + "num_tokens": 104523799.0, + "step": 4098 + }, + { + "epoch": 0.4501427630133978, + "grad_norm": 1.9937645196914673, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6934455037117004, + "num_tokens": 104553163.0, + "step": 4099 + }, + { + "epoch": 0.45025258071601143, + "grad_norm": 2.284428119659424, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.7060694694519043, + "num_tokens": 104580021.0, + "step": 4100 + }, + { + "epoch": 0.4503623984186251, + "grad_norm": 2.1423304080963135, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7107573747634888, + "num_tokens": 104608365.0, + "step": 4101 + }, + { + "epoch": 0.4504722161212387, + "grad_norm": 1.9696485996246338, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7241931557655334, + "num_tokens": 104635637.0, + "step": 4102 + }, + { + "epoch": 0.4505820338238524, + "grad_norm": 2.0899271965026855, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6935622692108154, + "num_tokens": 104663952.0, + "step": 4103 + }, + { + "epoch": 0.45069185152646607, + "grad_norm": 2.12048602104187, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6898297667503357, + "num_tokens": 104691540.0, + "step": 4104 + }, + { + "epoch": 0.4508016692290797, + "grad_norm": 2.4440865516662598, + "learning_rate": 1e-06, + "loss": 1.0461, + "mean_token_accuracy": 0.6887988448143005, + "num_tokens": 104714347.0, + "step": 4105 + }, + { + "epoch": 0.45091148693169336, + "grad_norm": 2.269831418991089, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7130550146102905, + "num_tokens": 104736488.0, + "step": 4106 + }, + { + "epoch": 0.45102130463430706, + "grad_norm": 2.0355544090270996, + "learning_rate": 1e-06, + "loss": 1.0473, + "mean_token_accuracy": 0.6980297565460205, + "num_tokens": 104764795.0, + "step": 4107 + }, + { + "epoch": 0.4511311223369207, + "grad_norm": 2.38862943649292, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7215589284896851, + "num_tokens": 104787929.0, + "step": 4108 + }, + { + "epoch": 0.45124094003953436, + "grad_norm": 2.2367618083953857, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7322300672531128, + "num_tokens": 104812571.0, + "step": 4109 + }, + { + "epoch": 0.45135075774214806, + "grad_norm": 2.2939698696136475, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7053402066230774, + "num_tokens": 104835225.0, + "step": 4110 + }, + { + "epoch": 0.4514605754447617, + "grad_norm": 2.25795578956604, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6920777559280396, + "num_tokens": 104861433.0, + "step": 4111 + }, + { + "epoch": 0.45157039314737535, + "grad_norm": 2.0648412704467773, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.7012002468109131, + "num_tokens": 104890159.0, + "step": 4112 + }, + { + "epoch": 0.451680210849989, + "grad_norm": 2.531599521636963, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6963014602661133, + "num_tokens": 104911903.0, + "step": 4113 + }, + { + "epoch": 0.4517900285526027, + "grad_norm": 1.9863746166229248, + "learning_rate": 1e-06, + "loss": 1.0341, + "mean_token_accuracy": 0.6869065761566162, + "num_tokens": 104944771.0, + "step": 4114 + }, + { + "epoch": 0.45189984625521634, + "grad_norm": 1.9363805055618286, + "learning_rate": 1e-06, + "loss": 1.0837, + "mean_token_accuracy": 0.6759925484657288, + "num_tokens": 104980738.0, + "step": 4115 + }, + { + "epoch": 0.45200966395783, + "grad_norm": 2.398268222808838, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6986441016197205, + "num_tokens": 105004007.0, + "step": 4116 + }, + { + "epoch": 0.4521194816604437, + "grad_norm": 2.058016300201416, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.6933630704879761, + "num_tokens": 105033743.0, + "step": 4117 + }, + { + "epoch": 0.45222929936305734, + "grad_norm": 2.2402985095977783, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7071611881256104, + "num_tokens": 105057626.0, + "step": 4118 + }, + { + "epoch": 0.452339117065671, + "grad_norm": 2.0492963790893555, + "learning_rate": 1e-06, + "loss": 1.0511, + "mean_token_accuracy": 0.6803643703460693, + "num_tokens": 105088186.0, + "step": 4119 + }, + { + "epoch": 0.4524489347682846, + "grad_norm": 2.0786378383636475, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6912286281585693, + "num_tokens": 105116703.0, + "step": 4120 + }, + { + "epoch": 0.45255875247089833, + "grad_norm": 2.0510408878326416, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7238487005233765, + "num_tokens": 105142772.0, + "step": 4121 + }, + { + "epoch": 0.452668570173512, + "grad_norm": 2.062819719314575, + "learning_rate": 1e-06, + "loss": 1.1589, + "mean_token_accuracy": 0.6501566767692566, + "num_tokens": 105175833.0, + "step": 4122 + }, + { + "epoch": 0.4527783878761256, + "grad_norm": 2.1115622520446777, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7080004215240479, + "num_tokens": 105203595.0, + "step": 4123 + }, + { + "epoch": 0.45288820557873927, + "grad_norm": 2.474285364151001, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.71635502576828, + "num_tokens": 105223275.0, + "step": 4124 + }, + { + "epoch": 0.45299802328135297, + "grad_norm": 2.0725836753845215, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.7054280042648315, + "num_tokens": 105251197.0, + "step": 4125 + }, + { + "epoch": 0.4531078409839666, + "grad_norm": 2.3504178524017334, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7036571502685547, + "num_tokens": 105275255.0, + "step": 4126 + }, + { + "epoch": 0.45321765868658026, + "grad_norm": 2.4024603366851807, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7064165472984314, + "num_tokens": 105297551.0, + "step": 4127 + }, + { + "epoch": 0.45332747638919396, + "grad_norm": 2.2312681674957275, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.705420732498169, + "num_tokens": 105322412.0, + "step": 4128 + }, + { + "epoch": 0.4534372940918076, + "grad_norm": 2.2205801010131836, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6876217126846313, + "num_tokens": 105347876.0, + "step": 4129 + }, + { + "epoch": 0.45354711179442125, + "grad_norm": 2.380255937576294, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7182438373565674, + "num_tokens": 105370310.0, + "step": 4130 + }, + { + "epoch": 0.4536569294970349, + "grad_norm": 2.2585320472717285, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6948813796043396, + "num_tokens": 105393719.0, + "step": 4131 + }, + { + "epoch": 0.4537667471996486, + "grad_norm": 2.1932873725891113, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7032524347305298, + "num_tokens": 105420184.0, + "step": 4132 + }, + { + "epoch": 0.45387656490226225, + "grad_norm": 2.598119020462036, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7008995413780212, + "num_tokens": 105440106.0, + "step": 4133 + }, + { + "epoch": 0.4539863826048759, + "grad_norm": 2.0858840942382812, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6871815919876099, + "num_tokens": 105468142.0, + "step": 4134 + }, + { + "epoch": 0.4540962003074896, + "grad_norm": 2.2466163635253906, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7042633891105652, + "num_tokens": 105493658.0, + "step": 4135 + }, + { + "epoch": 0.45420601801010324, + "grad_norm": 2.244508743286133, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7046611309051514, + "num_tokens": 105517780.0, + "step": 4136 + }, + { + "epoch": 0.4543158357127169, + "grad_norm": 2.2995340824127197, + "learning_rate": 1e-06, + "loss": 1.0928, + "mean_token_accuracy": 0.6803451776504517, + "num_tokens": 105543085.0, + "step": 4137 + }, + { + "epoch": 0.45442565341533053, + "grad_norm": 2.4278323650360107, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7237152457237244, + "num_tokens": 105562250.0, + "step": 4138 + }, + { + "epoch": 0.45453547111794423, + "grad_norm": 2.494004964828491, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7101204991340637, + "num_tokens": 105583259.0, + "step": 4139 + }, + { + "epoch": 0.4546452888205579, + "grad_norm": 2.228384256362915, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.6919550895690918, + "num_tokens": 105609967.0, + "step": 4140 + }, + { + "epoch": 0.4547551065231715, + "grad_norm": 2.0361292362213135, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.689812183380127, + "num_tokens": 105638835.0, + "step": 4141 + }, + { + "epoch": 0.45486492422578517, + "grad_norm": 2.246882200241089, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6961710453033447, + "num_tokens": 105662262.0, + "step": 4142 + }, + { + "epoch": 0.4549747419283989, + "grad_norm": 2.2869246006011963, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7257683277130127, + "num_tokens": 105685056.0, + "step": 4143 + }, + { + "epoch": 0.4550845596310125, + "grad_norm": 2.4466965198516846, + "learning_rate": 1e-06, + "loss": 1.0443, + "mean_token_accuracy": 0.6888928413391113, + "num_tokens": 105707854.0, + "step": 4144 + }, + { + "epoch": 0.45519437733362617, + "grad_norm": 2.1369271278381348, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7057081460952759, + "num_tokens": 105733355.0, + "step": 4145 + }, + { + "epoch": 0.45530419503623987, + "grad_norm": 2.452873468399048, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7109712958335876, + "num_tokens": 105753756.0, + "step": 4146 + }, + { + "epoch": 0.4554140127388535, + "grad_norm": 2.0432262420654297, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6932504177093506, + "num_tokens": 105784713.0, + "step": 4147 + }, + { + "epoch": 0.45552383044146716, + "grad_norm": 2.0418272018432617, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.7013016939163208, + "num_tokens": 105814050.0, + "step": 4148 + }, + { + "epoch": 0.4556336481440808, + "grad_norm": 2.5325767993927, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7129064798355103, + "num_tokens": 105833192.0, + "step": 4149 + }, + { + "epoch": 0.4557434658466945, + "grad_norm": 2.099512815475464, + "learning_rate": 1e-06, + "loss": 1.0236, + "mean_token_accuracy": 0.7015601396560669, + "num_tokens": 105860879.0, + "step": 4150 + }, + { + "epoch": 0.45585328354930815, + "grad_norm": 2.4223556518554688, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7162973284721375, + "num_tokens": 105882145.0, + "step": 4151 + }, + { + "epoch": 0.4559631012519218, + "grad_norm": 2.2605607509613037, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7025316953659058, + "num_tokens": 105907012.0, + "step": 4152 + }, + { + "epoch": 0.45607291895453544, + "grad_norm": 2.1339218616485596, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.7133776545524597, + "num_tokens": 105934222.0, + "step": 4153 + }, + { + "epoch": 0.45618273665714915, + "grad_norm": 2.2957518100738525, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7077329158782959, + "num_tokens": 105957873.0, + "step": 4154 + }, + { + "epoch": 0.4562925543597628, + "grad_norm": 2.055631399154663, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.712230920791626, + "num_tokens": 105985741.0, + "step": 4155 + }, + { + "epoch": 0.45640237206237644, + "grad_norm": 2.0714662075042725, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.718704104423523, + "num_tokens": 106015244.0, + "step": 4156 + }, + { + "epoch": 0.45651218976499014, + "grad_norm": 2.244150400161743, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7354178428649902, + "num_tokens": 106037984.0, + "step": 4157 + }, + { + "epoch": 0.4566220074676038, + "grad_norm": 2.01789927482605, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.705978512763977, + "num_tokens": 106067042.0, + "step": 4158 + }, + { + "epoch": 0.45673182517021743, + "grad_norm": 2.4479687213897705, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7106539011001587, + "num_tokens": 106086251.0, + "step": 4159 + }, + { + "epoch": 0.4568416428728311, + "grad_norm": 2.157973051071167, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7082424163818359, + "num_tokens": 106112435.0, + "step": 4160 + }, + { + "epoch": 0.4569514605754448, + "grad_norm": 1.7448134422302246, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7206149697303772, + "num_tokens": 106147145.0, + "step": 4161 + }, + { + "epoch": 0.4570612782780584, + "grad_norm": 2.0919911861419678, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.6987442374229431, + "num_tokens": 106173330.0, + "step": 4162 + }, + { + "epoch": 0.45717109598067207, + "grad_norm": 2.042248487472534, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.713456392288208, + "num_tokens": 106200364.0, + "step": 4163 + }, + { + "epoch": 0.4572809136832858, + "grad_norm": 2.3545265197753906, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7267516851425171, + "num_tokens": 106222509.0, + "step": 4164 + }, + { + "epoch": 0.4573907313858994, + "grad_norm": 2.466413974761963, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7065505385398865, + "num_tokens": 106243293.0, + "step": 4165 + }, + { + "epoch": 0.45750054908851306, + "grad_norm": 2.0205655097961426, + "learning_rate": 1e-06, + "loss": 1.0822, + "mean_token_accuracy": 0.6806961297988892, + "num_tokens": 106273818.0, + "step": 4166 + }, + { + "epoch": 0.4576103667911267, + "grad_norm": 2.1488497257232666, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.7063870429992676, + "num_tokens": 106299410.0, + "step": 4167 + }, + { + "epoch": 0.4577201844937404, + "grad_norm": 2.0501859188079834, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6988644003868103, + "num_tokens": 106328501.0, + "step": 4168 + }, + { + "epoch": 0.45783000219635406, + "grad_norm": 2.2263574600219727, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7023991346359253, + "num_tokens": 106353536.0, + "step": 4169 + }, + { + "epoch": 0.4579398198989677, + "grad_norm": 2.4861900806427, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7357909679412842, + "num_tokens": 106373157.0, + "step": 4170 + }, + { + "epoch": 0.45804963760158135, + "grad_norm": 2.384652853012085, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7195876836776733, + "num_tokens": 106393666.0, + "step": 4171 + }, + { + "epoch": 0.45815945530419505, + "grad_norm": 2.362837553024292, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7190766334533691, + "num_tokens": 106414685.0, + "step": 4172 + }, + { + "epoch": 0.4582692730068087, + "grad_norm": 2.1571967601776123, + "learning_rate": 1e-06, + "loss": 1.0327, + "mean_token_accuracy": 0.6895254850387573, + "num_tokens": 106442275.0, + "step": 4173 + }, + { + "epoch": 0.45837909070942234, + "grad_norm": 2.003676176071167, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7045771479606628, + "num_tokens": 106471216.0, + "step": 4174 + }, + { + "epoch": 0.45848890841203604, + "grad_norm": 2.1429226398468018, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.7013953924179077, + "num_tokens": 106498257.0, + "step": 4175 + }, + { + "epoch": 0.4585987261146497, + "grad_norm": 2.252849817276001, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.6983708143234253, + "num_tokens": 106524949.0, + "step": 4176 + }, + { + "epoch": 0.45870854381726334, + "grad_norm": 2.3496110439300537, + "learning_rate": 1e-06, + "loss": 1.0478, + "mean_token_accuracy": 0.6942580938339233, + "num_tokens": 106547658.0, + "step": 4177 + }, + { + "epoch": 0.458818361519877, + "grad_norm": 2.238814115524292, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7100818157196045, + "num_tokens": 106571308.0, + "step": 4178 + }, + { + "epoch": 0.4589281792224907, + "grad_norm": 2.1932992935180664, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6933716535568237, + "num_tokens": 106597461.0, + "step": 4179 + }, + { + "epoch": 0.45903799692510433, + "grad_norm": 2.273008346557617, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.724725604057312, + "num_tokens": 106619572.0, + "step": 4180 + }, + { + "epoch": 0.459147814627718, + "grad_norm": 2.4162344932556152, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7078155875205994, + "num_tokens": 106641855.0, + "step": 4181 + }, + { + "epoch": 0.4592576323303316, + "grad_norm": 2.0724923610687256, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.6937265992164612, + "num_tokens": 106670131.0, + "step": 4182 + }, + { + "epoch": 0.4593674500329453, + "grad_norm": 2.3466391563415527, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6986611485481262, + "num_tokens": 106693795.0, + "step": 4183 + }, + { + "epoch": 0.45947726773555897, + "grad_norm": 2.3193321228027344, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7019423246383667, + "num_tokens": 106718683.0, + "step": 4184 + }, + { + "epoch": 0.4595870854381726, + "grad_norm": 2.1313486099243164, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6985920071601868, + "num_tokens": 106747038.0, + "step": 4185 + }, + { + "epoch": 0.4596969031407863, + "grad_norm": 2.3687236309051514, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.7008030414581299, + "num_tokens": 106770619.0, + "step": 4186 + }, + { + "epoch": 0.45980672084339996, + "grad_norm": 2.1798486709594727, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.6881441473960876, + "num_tokens": 106796623.0, + "step": 4187 + }, + { + "epoch": 0.4599165385460136, + "grad_norm": 2.126676559448242, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7069371938705444, + "num_tokens": 106827125.0, + "step": 4188 + }, + { + "epoch": 0.46002635624862726, + "grad_norm": 2.4365830421447754, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7101086974143982, + "num_tokens": 106849558.0, + "step": 4189 + }, + { + "epoch": 0.46013617395124096, + "grad_norm": 2.332664728164673, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.7024707794189453, + "num_tokens": 106873572.0, + "step": 4190 + }, + { + "epoch": 0.4602459916538546, + "grad_norm": 2.0760953426361084, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6972852349281311, + "num_tokens": 106902043.0, + "step": 4191 + }, + { + "epoch": 0.46035580935646825, + "grad_norm": 2.1965742111206055, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6918530464172363, + "num_tokens": 106929811.0, + "step": 4192 + }, + { + "epoch": 0.46046562705908195, + "grad_norm": 2.3772990703582764, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7112098932266235, + "num_tokens": 106953302.0, + "step": 4193 + }, + { + "epoch": 0.4605754447616956, + "grad_norm": 2.269948959350586, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7342514395713806, + "num_tokens": 106976261.0, + "step": 4194 + }, + { + "epoch": 0.46068526246430924, + "grad_norm": 1.9397555589675903, + "learning_rate": 1e-06, + "loss": 1.0466, + "mean_token_accuracy": 0.678713321685791, + "num_tokens": 107007559.0, + "step": 4195 + }, + { + "epoch": 0.4607950801669229, + "grad_norm": 2.218534469604492, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6939473152160645, + "num_tokens": 107032961.0, + "step": 4196 + }, + { + "epoch": 0.4609048978695366, + "grad_norm": 2.474822759628296, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7112048268318176, + "num_tokens": 107054714.0, + "step": 4197 + }, + { + "epoch": 0.46101471557215024, + "grad_norm": 2.2396061420440674, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7047386169433594, + "num_tokens": 107079603.0, + "step": 4198 + }, + { + "epoch": 0.4611245332747639, + "grad_norm": 2.370217800140381, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.6988480091094971, + "num_tokens": 107102648.0, + "step": 4199 + }, + { + "epoch": 0.4612343509773775, + "grad_norm": 2.289781332015991, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7075483798980713, + "num_tokens": 107126090.0, + "step": 4200 + }, + { + "epoch": 0.46134416867999123, + "grad_norm": 1.9576953649520874, + "learning_rate": 1e-06, + "loss": 1.0408, + "mean_token_accuracy": 0.684205174446106, + "num_tokens": 107159826.0, + "step": 4201 + }, + { + "epoch": 0.4614539863826049, + "grad_norm": 1.8495194911956787, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6876550912857056, + "num_tokens": 107195853.0, + "step": 4202 + }, + { + "epoch": 0.4615638040852185, + "grad_norm": 2.4011778831481934, + "learning_rate": 1e-06, + "loss": 1.1328, + "mean_token_accuracy": 0.6591019034385681, + "num_tokens": 107218595.0, + "step": 4203 + }, + { + "epoch": 0.4616736217878322, + "grad_norm": 1.860905647277832, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7144793272018433, + "num_tokens": 107250421.0, + "step": 4204 + }, + { + "epoch": 0.46178343949044587, + "grad_norm": 2.067326545715332, + "learning_rate": 1e-06, + "loss": 1.0488, + "mean_token_accuracy": 0.684218168258667, + "num_tokens": 107281584.0, + "step": 4205 + }, + { + "epoch": 0.4618932571930595, + "grad_norm": 2.461402416229248, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.699710488319397, + "num_tokens": 107304735.0, + "step": 4206 + }, + { + "epoch": 0.46200307489567316, + "grad_norm": 2.233761787414551, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6942483186721802, + "num_tokens": 107329390.0, + "step": 4207 + }, + { + "epoch": 0.46211289259828686, + "grad_norm": 2.066741704940796, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7100846767425537, + "num_tokens": 107357144.0, + "step": 4208 + }, + { + "epoch": 0.4622227103009005, + "grad_norm": 2.0717756748199463, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.700892448425293, + "num_tokens": 107385810.0, + "step": 4209 + }, + { + "epoch": 0.46233252800351415, + "grad_norm": 2.1247541904449463, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7012361884117126, + "num_tokens": 107410974.0, + "step": 4210 + }, + { + "epoch": 0.46244234570612786, + "grad_norm": 2.3948214054107666, + "learning_rate": 1e-06, + "loss": 1.1175, + "mean_token_accuracy": 0.6717363595962524, + "num_tokens": 107433413.0, + "step": 4211 + }, + { + "epoch": 0.4625521634087415, + "grad_norm": 2.116400718688965, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.6960309743881226, + "num_tokens": 107459772.0, + "step": 4212 + }, + { + "epoch": 0.46266198111135515, + "grad_norm": 2.294187307357788, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7085719108581543, + "num_tokens": 107484240.0, + "step": 4213 + }, + { + "epoch": 0.4627717988139688, + "grad_norm": 2.1628448963165283, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7265627384185791, + "num_tokens": 107508733.0, + "step": 4214 + }, + { + "epoch": 0.4628816165165825, + "grad_norm": 2.3079440593719482, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7169270515441895, + "num_tokens": 107533769.0, + "step": 4215 + }, + { + "epoch": 0.46299143421919614, + "grad_norm": 1.9974521398544312, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7249376773834229, + "num_tokens": 107561815.0, + "step": 4216 + }, + { + "epoch": 0.4631012519218098, + "grad_norm": 2.2051477432250977, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7130535840988159, + "num_tokens": 107586175.0, + "step": 4217 + }, + { + "epoch": 0.46321106962442343, + "grad_norm": 2.435943126678467, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7223819494247437, + "num_tokens": 107609646.0, + "step": 4218 + }, + { + "epoch": 0.46332088732703713, + "grad_norm": 2.065356731414795, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7022216320037842, + "num_tokens": 107639081.0, + "step": 4219 + }, + { + "epoch": 0.4634307050296508, + "grad_norm": 1.9335033893585205, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.702501118183136, + "num_tokens": 107672020.0, + "step": 4220 + }, + { + "epoch": 0.4635405227322644, + "grad_norm": 2.33180832862854, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7058982849121094, + "num_tokens": 107695246.0, + "step": 4221 + }, + { + "epoch": 0.4636503404348781, + "grad_norm": 2.239628791809082, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.71396803855896, + "num_tokens": 107719935.0, + "step": 4222 + }, + { + "epoch": 0.4637601581374918, + "grad_norm": 2.0234386920928955, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6945976614952087, + "num_tokens": 107749830.0, + "step": 4223 + }, + { + "epoch": 0.4638699758401054, + "grad_norm": 2.0664737224578857, + "learning_rate": 1e-06, + "loss": 1.1245, + "mean_token_accuracy": 0.6697559356689453, + "num_tokens": 107779678.0, + "step": 4224 + }, + { + "epoch": 0.46397979354271907, + "grad_norm": 2.231640338897705, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.6997631788253784, + "num_tokens": 107805222.0, + "step": 4225 + }, + { + "epoch": 0.46408961124533277, + "grad_norm": 2.1527230739593506, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6985079050064087, + "num_tokens": 107834370.0, + "step": 4226 + }, + { + "epoch": 0.4641994289479464, + "grad_norm": 2.1497793197631836, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7316627502441406, + "num_tokens": 107859599.0, + "step": 4227 + }, + { + "epoch": 0.46430924665056006, + "grad_norm": 2.401456594467163, + "learning_rate": 1e-06, + "loss": 1.0508, + "mean_token_accuracy": 0.6887497901916504, + "num_tokens": 107881135.0, + "step": 4228 + }, + { + "epoch": 0.4644190643531737, + "grad_norm": 2.241374969482422, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6986536979675293, + "num_tokens": 107905961.0, + "step": 4229 + }, + { + "epoch": 0.4645288820557874, + "grad_norm": 2.0089337825775146, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6977951526641846, + "num_tokens": 107935294.0, + "step": 4230 + }, + { + "epoch": 0.46463869975840105, + "grad_norm": 2.4458258152008057, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7210090160369873, + "num_tokens": 107955473.0, + "step": 4231 + }, + { + "epoch": 0.4647485174610147, + "grad_norm": 2.3671634197235107, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7075809240341187, + "num_tokens": 107979094.0, + "step": 4232 + }, + { + "epoch": 0.4648583351636284, + "grad_norm": 2.083005905151367, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7212111949920654, + "num_tokens": 108006202.0, + "step": 4233 + }, + { + "epoch": 0.46496815286624205, + "grad_norm": 2.0249338150024414, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.7020608186721802, + "num_tokens": 108033865.0, + "step": 4234 + }, + { + "epoch": 0.4650779705688557, + "grad_norm": 2.0696728229522705, + "learning_rate": 1e-06, + "loss": 1.1262, + "mean_token_accuracy": 0.6716614961624146, + "num_tokens": 108064294.0, + "step": 4235 + }, + { + "epoch": 0.46518778827146934, + "grad_norm": 1.981701135635376, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7241359949111938, + "num_tokens": 108092887.0, + "step": 4236 + }, + { + "epoch": 0.46529760597408304, + "grad_norm": 2.352304220199585, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6942687034606934, + "num_tokens": 108116612.0, + "step": 4237 + }, + { + "epoch": 0.4654074236766967, + "grad_norm": 2.1190314292907715, + "learning_rate": 1e-06, + "loss": 1.0684, + "mean_token_accuracy": 0.6773884892463684, + "num_tokens": 108144213.0, + "step": 4238 + }, + { + "epoch": 0.46551724137931033, + "grad_norm": 2.147656202316284, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.7074977159500122, + "num_tokens": 108171891.0, + "step": 4239 + }, + { + "epoch": 0.46562705908192403, + "grad_norm": 2.263979911804199, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.703906774520874, + "num_tokens": 108197146.0, + "step": 4240 + }, + { + "epoch": 0.4657368767845377, + "grad_norm": 2.183718204498291, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.6958838701248169, + "num_tokens": 108221726.0, + "step": 4241 + }, + { + "epoch": 0.4658466944871513, + "grad_norm": 2.306218147277832, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7156420946121216, + "num_tokens": 108242937.0, + "step": 4242 + }, + { + "epoch": 0.46595651218976497, + "grad_norm": 2.3377685546875, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.7002733945846558, + "num_tokens": 108266320.0, + "step": 4243 + }, + { + "epoch": 0.4660663298923787, + "grad_norm": 2.1874217987060547, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.6968902349472046, + "num_tokens": 108292992.0, + "step": 4244 + }, + { + "epoch": 0.4661761475949923, + "grad_norm": 1.9874948263168335, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7085747718811035, + "num_tokens": 108322362.0, + "step": 4245 + }, + { + "epoch": 0.46628596529760596, + "grad_norm": 2.594982624053955, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7063050270080566, + "num_tokens": 108342382.0, + "step": 4246 + }, + { + "epoch": 0.4663957830002196, + "grad_norm": 1.933960199356079, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.702544629573822, + "num_tokens": 108372458.0, + "step": 4247 + }, + { + "epoch": 0.4665056007028333, + "grad_norm": 2.5473594665527344, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.703914999961853, + "num_tokens": 108394192.0, + "step": 4248 + }, + { + "epoch": 0.46661541840544696, + "grad_norm": 2.1638197898864746, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6906439661979675, + "num_tokens": 108422677.0, + "step": 4249 + }, + { + "epoch": 0.4667252361080606, + "grad_norm": 2.3635952472686768, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7027274370193481, + "num_tokens": 108444823.0, + "step": 4250 + }, + { + "epoch": 0.4668350538106743, + "grad_norm": 2.452244520187378, + "learning_rate": 1e-06, + "loss": 1.025, + "mean_token_accuracy": 0.6899847984313965, + "num_tokens": 108467126.0, + "step": 4251 + }, + { + "epoch": 0.46694487151328795, + "grad_norm": 2.419687271118164, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7217759490013123, + "num_tokens": 108486987.0, + "step": 4252 + }, + { + "epoch": 0.4670546892159016, + "grad_norm": 2.091846227645874, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7087829113006592, + "num_tokens": 108513132.0, + "step": 4253 + }, + { + "epoch": 0.46716450691851524, + "grad_norm": 2.2511444091796875, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7138245105743408, + "num_tokens": 108535526.0, + "step": 4254 + }, + { + "epoch": 0.46727432462112894, + "grad_norm": 2.3554112911224365, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7120433449745178, + "num_tokens": 108557103.0, + "step": 4255 + }, + { + "epoch": 0.4673841423237426, + "grad_norm": 2.2718658447265625, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.7192630767822266, + "num_tokens": 108580905.0, + "step": 4256 + }, + { + "epoch": 0.46749396002635624, + "grad_norm": 2.0754895210266113, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6855834722518921, + "num_tokens": 108609434.0, + "step": 4257 + }, + { + "epoch": 0.4676037777289699, + "grad_norm": 2.4336915016174316, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.712250292301178, + "num_tokens": 108631724.0, + "step": 4258 + }, + { + "epoch": 0.4677135954315836, + "grad_norm": 2.2664098739624023, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6985667943954468, + "num_tokens": 108656542.0, + "step": 4259 + }, + { + "epoch": 0.46782341313419723, + "grad_norm": 2.2833969593048096, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7066816091537476, + "num_tokens": 108679622.0, + "step": 4260 + }, + { + "epoch": 0.4679332308368109, + "grad_norm": 2.115760087966919, + "learning_rate": 1e-06, + "loss": 1.0339, + "mean_token_accuracy": 0.6896848678588867, + "num_tokens": 108706515.0, + "step": 4261 + }, + { + "epoch": 0.4680430485394246, + "grad_norm": 2.4287991523742676, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7023538947105408, + "num_tokens": 108727105.0, + "step": 4262 + }, + { + "epoch": 0.4681528662420382, + "grad_norm": 2.5388944149017334, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7039352655410767, + "num_tokens": 108746894.0, + "step": 4263 + }, + { + "epoch": 0.46826268394465187, + "grad_norm": 2.2892041206359863, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7021229267120361, + "num_tokens": 108771054.0, + "step": 4264 + }, + { + "epoch": 0.4683725016472655, + "grad_norm": 2.2535812854766846, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7042809128761292, + "num_tokens": 108796881.0, + "step": 4265 + }, + { + "epoch": 0.4684823193498792, + "grad_norm": 2.1325161457061768, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7162966728210449, + "num_tokens": 108824400.0, + "step": 4266 + }, + { + "epoch": 0.46859213705249286, + "grad_norm": 2.541659355163574, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.727124810218811, + "num_tokens": 108844065.0, + "step": 4267 + }, + { + "epoch": 0.4687019547551065, + "grad_norm": 2.5076050758361816, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7132893204689026, + "num_tokens": 108863667.0, + "step": 4268 + }, + { + "epoch": 0.4688117724577202, + "grad_norm": 2.29868745803833, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7052981853485107, + "num_tokens": 108887514.0, + "step": 4269 + }, + { + "epoch": 0.46892159016033386, + "grad_norm": 2.317901372909546, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7070133686065674, + "num_tokens": 108910393.0, + "step": 4270 + }, + { + "epoch": 0.4690314078629475, + "grad_norm": 2.2647855281829834, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.7035984992980957, + "num_tokens": 108935039.0, + "step": 4271 + }, + { + "epoch": 0.46914122556556115, + "grad_norm": 2.208228588104248, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6889636516571045, + "num_tokens": 108960145.0, + "step": 4272 + }, + { + "epoch": 0.46925104326817485, + "grad_norm": 2.16036319732666, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7141621112823486, + "num_tokens": 108987549.0, + "step": 4273 + }, + { + "epoch": 0.4693608609707885, + "grad_norm": 2.3675448894500732, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7053442597389221, + "num_tokens": 109011290.0, + "step": 4274 + }, + { + "epoch": 0.46947067867340214, + "grad_norm": 2.008384943008423, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.710290789604187, + "num_tokens": 109038122.0, + "step": 4275 + }, + { + "epoch": 0.4695804963760158, + "grad_norm": 2.251173257827759, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6975122094154358, + "num_tokens": 109063079.0, + "step": 4276 + }, + { + "epoch": 0.4696903140786295, + "grad_norm": 2.484919786453247, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.701056718826294, + "num_tokens": 109083815.0, + "step": 4277 + }, + { + "epoch": 0.46980013178124314, + "grad_norm": 2.030869722366333, + "learning_rate": 1e-06, + "loss": 1.0753, + "mean_token_accuracy": 0.6777985095977783, + "num_tokens": 109113269.0, + "step": 4278 + }, + { + "epoch": 0.4699099494838568, + "grad_norm": 2.2668392658233643, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7082416415214539, + "num_tokens": 109137762.0, + "step": 4279 + }, + { + "epoch": 0.4700197671864705, + "grad_norm": 2.4483799934387207, + "learning_rate": 1e-06, + "loss": 1.0533, + "mean_token_accuracy": 0.6829297542572021, + "num_tokens": 109166591.0, + "step": 4280 + }, + { + "epoch": 0.47012958488908413, + "grad_norm": 2.418651580810547, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.6999990344047546, + "num_tokens": 109188243.0, + "step": 4281 + }, + { + "epoch": 0.4702394025916978, + "grad_norm": 2.0149924755096436, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.692889392375946, + "num_tokens": 109217581.0, + "step": 4282 + }, + { + "epoch": 0.4703492202943114, + "grad_norm": 2.4231655597686768, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7240779995918274, + "num_tokens": 109240920.0, + "step": 4283 + }, + { + "epoch": 0.4704590379969251, + "grad_norm": 2.187671661376953, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7086687088012695, + "num_tokens": 109266769.0, + "step": 4284 + }, + { + "epoch": 0.47056885569953877, + "grad_norm": 2.2814481258392334, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6881558299064636, + "num_tokens": 109290539.0, + "step": 4285 + }, + { + "epoch": 0.4706786734021524, + "grad_norm": 2.2528390884399414, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.690559983253479, + "num_tokens": 109317071.0, + "step": 4286 + }, + { + "epoch": 0.4707884911047661, + "grad_norm": 2.330340623855591, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6919901967048645, + "num_tokens": 109342162.0, + "step": 4287 + }, + { + "epoch": 0.47089830880737976, + "grad_norm": 2.161027431488037, + "learning_rate": 1e-06, + "loss": 1.0608, + "mean_token_accuracy": 0.6825532913208008, + "num_tokens": 109370918.0, + "step": 4288 + }, + { + "epoch": 0.4710081265099934, + "grad_norm": 1.8436311483383179, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.6987553834915161, + "num_tokens": 109404554.0, + "step": 4289 + }, + { + "epoch": 0.47111794421260705, + "grad_norm": 2.3144407272338867, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6967049241065979, + "num_tokens": 109430983.0, + "step": 4290 + }, + { + "epoch": 0.47122776191522076, + "grad_norm": 2.398308277130127, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7118828892707825, + "num_tokens": 109451940.0, + "step": 4291 + }, + { + "epoch": 0.4713375796178344, + "grad_norm": 1.9776365756988525, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.6950759887695312, + "num_tokens": 109484137.0, + "step": 4292 + }, + { + "epoch": 0.47144739732044805, + "grad_norm": 2.0154521465301514, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7014232873916626, + "num_tokens": 109516664.0, + "step": 4293 + }, + { + "epoch": 0.4715572150230617, + "grad_norm": 2.153644561767578, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6983722448348999, + "num_tokens": 109542645.0, + "step": 4294 + }, + { + "epoch": 0.4716670327256754, + "grad_norm": 2.140817403793335, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7200630307197571, + "num_tokens": 109566397.0, + "step": 4295 + }, + { + "epoch": 0.47177685042828904, + "grad_norm": 2.112854480743408, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.7000761032104492, + "num_tokens": 109596409.0, + "step": 4296 + }, + { + "epoch": 0.4718866681309027, + "grad_norm": 2.1604256629943848, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7300368547439575, + "num_tokens": 109623209.0, + "step": 4297 + }, + { + "epoch": 0.4719964858335164, + "grad_norm": 2.1964848041534424, + "learning_rate": 1e-06, + "loss": 1.0299, + "mean_token_accuracy": 0.6906725168228149, + "num_tokens": 109648691.0, + "step": 4298 + }, + { + "epoch": 0.47210630353613003, + "grad_norm": 2.277251720428467, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6899751424789429, + "num_tokens": 109675374.0, + "step": 4299 + }, + { + "epoch": 0.4722161212387437, + "grad_norm": 2.0631823539733887, + "learning_rate": 1e-06, + "loss": 1.0439, + "mean_token_accuracy": 0.6876981258392334, + "num_tokens": 109704773.0, + "step": 4300 + }, + { + "epoch": 0.4723259389413573, + "grad_norm": 2.051353931427002, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7100459933280945, + "num_tokens": 109732711.0, + "step": 4301 + }, + { + "epoch": 0.472435756643971, + "grad_norm": 2.367252826690674, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7148718237876892, + "num_tokens": 109755352.0, + "step": 4302 + }, + { + "epoch": 0.4725455743465847, + "grad_norm": 2.275381326675415, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7111210227012634, + "num_tokens": 109779182.0, + "step": 4303 + }, + { + "epoch": 0.4726553920491983, + "grad_norm": 2.0942375659942627, + "learning_rate": 1e-06, + "loss": 1.1096, + "mean_token_accuracy": 0.6725550293922424, + "num_tokens": 109810948.0, + "step": 4304 + }, + { + "epoch": 0.47276520975181197, + "grad_norm": 2.2404282093048096, + "learning_rate": 1e-06, + "loss": 1.0554, + "mean_token_accuracy": 0.6818534135818481, + "num_tokens": 109838769.0, + "step": 4305 + }, + { + "epoch": 0.47287502745442567, + "grad_norm": 2.23211669921875, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7104353904724121, + "num_tokens": 109865342.0, + "step": 4306 + }, + { + "epoch": 0.4729848451570393, + "grad_norm": 2.281761407852173, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6951356530189514, + "num_tokens": 109891739.0, + "step": 4307 + }, + { + "epoch": 0.47309466285965296, + "grad_norm": 2.48336124420166, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7235968112945557, + "num_tokens": 109912793.0, + "step": 4308 + }, + { + "epoch": 0.47320448056226666, + "grad_norm": 2.746370315551758, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.7017987370491028, + "num_tokens": 109931763.0, + "step": 4309 + }, + { + "epoch": 0.4733142982648803, + "grad_norm": 2.020644187927246, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7094498872756958, + "num_tokens": 109960915.0, + "step": 4310 + }, + { + "epoch": 0.47342411596749395, + "grad_norm": 2.122889280319214, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.6977895498275757, + "num_tokens": 109987012.0, + "step": 4311 + }, + { + "epoch": 0.4735339336701076, + "grad_norm": 2.113893747329712, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7075435519218445, + "num_tokens": 110012920.0, + "step": 4312 + }, + { + "epoch": 0.4736437513727213, + "grad_norm": 2.1870322227478027, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7226966619491577, + "num_tokens": 110038676.0, + "step": 4313 + }, + { + "epoch": 0.47375356907533495, + "grad_norm": 1.923624038696289, + "learning_rate": 1e-06, + "loss": 1.084, + "mean_token_accuracy": 0.6828449964523315, + "num_tokens": 110070603.0, + "step": 4314 + }, + { + "epoch": 0.4738633867779486, + "grad_norm": 2.3373751640319824, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.707690954208374, + "num_tokens": 110093749.0, + "step": 4315 + }, + { + "epoch": 0.4739732044805623, + "grad_norm": 2.4342169761657715, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7151954770088196, + "num_tokens": 110115662.0, + "step": 4316 + }, + { + "epoch": 0.47408302218317594, + "grad_norm": 2.0925533771514893, + "learning_rate": 1e-06, + "loss": 1.0524, + "mean_token_accuracy": 0.686596691608429, + "num_tokens": 110144579.0, + "step": 4317 + }, + { + "epoch": 0.4741928398857896, + "grad_norm": 2.050114154815674, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6963838934898376, + "num_tokens": 110171091.0, + "step": 4318 + }, + { + "epoch": 0.47430265758840323, + "grad_norm": 2.0263328552246094, + "learning_rate": 1e-06, + "loss": 1.0605, + "mean_token_accuracy": 0.6826502084732056, + "num_tokens": 110201846.0, + "step": 4319 + }, + { + "epoch": 0.47441247529101693, + "grad_norm": 1.9969491958618164, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7138073444366455, + "num_tokens": 110229538.0, + "step": 4320 + }, + { + "epoch": 0.4745222929936306, + "grad_norm": 2.0556435585021973, + "learning_rate": 1e-06, + "loss": 1.0876, + "mean_token_accuracy": 0.6770588159561157, + "num_tokens": 110260528.0, + "step": 4321 + }, + { + "epoch": 0.4746321106962442, + "grad_norm": 2.1693196296691895, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7189425230026245, + "num_tokens": 110286376.0, + "step": 4322 + }, + { + "epoch": 0.47474192839885787, + "grad_norm": 2.2107925415039062, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.6989772319793701, + "num_tokens": 110311052.0, + "step": 4323 + }, + { + "epoch": 0.4748517461014716, + "grad_norm": 2.2354841232299805, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6923469305038452, + "num_tokens": 110336097.0, + "step": 4324 + }, + { + "epoch": 0.4749615638040852, + "grad_norm": 2.258371114730835, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.7032842636108398, + "num_tokens": 110361333.0, + "step": 4325 + }, + { + "epoch": 0.47507138150669886, + "grad_norm": 2.5661253929138184, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7130123972892761, + "num_tokens": 110380246.0, + "step": 4326 + }, + { + "epoch": 0.47518119920931257, + "grad_norm": 2.0515477657318115, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.68182772397995, + "num_tokens": 110409116.0, + "step": 4327 + }, + { + "epoch": 0.4752910169119262, + "grad_norm": 1.975200891494751, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.6955066323280334, + "num_tokens": 110442490.0, + "step": 4328 + }, + { + "epoch": 0.47540083461453986, + "grad_norm": 2.4208595752716064, + "learning_rate": 1e-06, + "loss": 1.0222, + "mean_token_accuracy": 0.6934217810630798, + "num_tokens": 110465826.0, + "step": 4329 + }, + { + "epoch": 0.4755106523171535, + "grad_norm": 2.2365238666534424, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.6871491074562073, + "num_tokens": 110490770.0, + "step": 4330 + }, + { + "epoch": 0.4756204700197672, + "grad_norm": 2.2047922611236572, + "learning_rate": 1e-06, + "loss": 1.0814, + "mean_token_accuracy": 0.6787360310554504, + "num_tokens": 110516906.0, + "step": 4331 + }, + { + "epoch": 0.47573028772238085, + "grad_norm": 2.0858218669891357, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7151046395301819, + "num_tokens": 110543855.0, + "step": 4332 + }, + { + "epoch": 0.4758401054249945, + "grad_norm": 2.431654691696167, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7044824361801147, + "num_tokens": 110566812.0, + "step": 4333 + }, + { + "epoch": 0.47594992312760814, + "grad_norm": 2.15828537940979, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.718464732170105, + "num_tokens": 110591388.0, + "step": 4334 + }, + { + "epoch": 0.47605974083022184, + "grad_norm": 2.3584377765655518, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7089403867721558, + "num_tokens": 110612936.0, + "step": 4335 + }, + { + "epoch": 0.4761695585328355, + "grad_norm": 2.305129289627075, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6870342493057251, + "num_tokens": 110638402.0, + "step": 4336 + }, + { + "epoch": 0.47627937623544914, + "grad_norm": 2.212446451187134, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7191970944404602, + "num_tokens": 110662115.0, + "step": 4337 + }, + { + "epoch": 0.47638919393806284, + "grad_norm": 2.0757007598876953, + "learning_rate": 1e-06, + "loss": 1.0209, + "mean_token_accuracy": 0.6925093531608582, + "num_tokens": 110689820.0, + "step": 4338 + }, + { + "epoch": 0.4764990116406765, + "grad_norm": 2.6823596954345703, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7165159583091736, + "num_tokens": 110709875.0, + "step": 4339 + }, + { + "epoch": 0.47660882934329013, + "grad_norm": 2.5200958251953125, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.705985426902771, + "num_tokens": 110731438.0, + "step": 4340 + }, + { + "epoch": 0.4767186470459038, + "grad_norm": 2.6947381496429443, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7128984928131104, + "num_tokens": 110749278.0, + "step": 4341 + }, + { + "epoch": 0.4768284647485175, + "grad_norm": 2.018991708755493, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7159494757652283, + "num_tokens": 110777001.0, + "step": 4342 + }, + { + "epoch": 0.4769382824511311, + "grad_norm": 2.4835922718048096, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.724905252456665, + "num_tokens": 110798224.0, + "step": 4343 + }, + { + "epoch": 0.47704810015374477, + "grad_norm": 2.43086314201355, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6894983053207397, + "num_tokens": 110820059.0, + "step": 4344 + }, + { + "epoch": 0.47715791785635847, + "grad_norm": 1.9829925298690796, + "learning_rate": 1e-06, + "loss": 1.0482, + "mean_token_accuracy": 0.6848629117012024, + "num_tokens": 110852162.0, + "step": 4345 + }, + { + "epoch": 0.4772677355589721, + "grad_norm": 2.1288797855377197, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7219407558441162, + "num_tokens": 110877660.0, + "step": 4346 + }, + { + "epoch": 0.47737755326158576, + "grad_norm": 2.2840073108673096, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.7065627574920654, + "num_tokens": 110902430.0, + "step": 4347 + }, + { + "epoch": 0.4774873709641994, + "grad_norm": 2.0818698406219482, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7055950164794922, + "num_tokens": 110929201.0, + "step": 4348 + }, + { + "epoch": 0.4775971886668131, + "grad_norm": 2.4460151195526123, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7141580581665039, + "num_tokens": 110948349.0, + "step": 4349 + }, + { + "epoch": 0.47770700636942676, + "grad_norm": 1.8794506788253784, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6955142617225647, + "num_tokens": 110981287.0, + "step": 4350 + }, + { + "epoch": 0.4778168240720404, + "grad_norm": 2.353410482406616, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7037750482559204, + "num_tokens": 111006167.0, + "step": 4351 + }, + { + "epoch": 0.47792664177465405, + "grad_norm": 2.143519163131714, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6890982389450073, + "num_tokens": 111032136.0, + "step": 4352 + }, + { + "epoch": 0.47803645947726775, + "grad_norm": 1.9968247413635254, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6959143877029419, + "num_tokens": 111065391.0, + "step": 4353 + }, + { + "epoch": 0.4781462771798814, + "grad_norm": 2.46927809715271, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7044857740402222, + "num_tokens": 111086794.0, + "step": 4354 + }, + { + "epoch": 0.47825609488249504, + "grad_norm": 2.453415870666504, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.715616762638092, + "num_tokens": 111106575.0, + "step": 4355 + }, + { + "epoch": 0.47836591258510874, + "grad_norm": 2.2083358764648438, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7227286100387573, + "num_tokens": 111130376.0, + "step": 4356 + }, + { + "epoch": 0.4784757302877224, + "grad_norm": 2.103564739227295, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7011106014251709, + "num_tokens": 111157357.0, + "step": 4357 + }, + { + "epoch": 0.47858554799033604, + "grad_norm": 2.055772066116333, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6975106000900269, + "num_tokens": 111188158.0, + "step": 4358 + }, + { + "epoch": 0.4786953656929497, + "grad_norm": 2.3030853271484375, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.7033083438873291, + "num_tokens": 111212518.0, + "step": 4359 + }, + { + "epoch": 0.4788051833955634, + "grad_norm": 2.156452178955078, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6997330188751221, + "num_tokens": 111240901.0, + "step": 4360 + }, + { + "epoch": 0.47891500109817703, + "grad_norm": 2.602900505065918, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7179691791534424, + "num_tokens": 111259376.0, + "step": 4361 + }, + { + "epoch": 0.4790248188007907, + "grad_norm": 2.4636423587799072, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.711459755897522, + "num_tokens": 111279485.0, + "step": 4362 + }, + { + "epoch": 0.4791346365034044, + "grad_norm": 2.567314624786377, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7009537220001221, + "num_tokens": 111300112.0, + "step": 4363 + }, + { + "epoch": 0.479244454206018, + "grad_norm": 2.300816535949707, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7208971977233887, + "num_tokens": 111323787.0, + "step": 4364 + }, + { + "epoch": 0.47935427190863167, + "grad_norm": 2.1004226207733154, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7068710327148438, + "num_tokens": 111351462.0, + "step": 4365 + }, + { + "epoch": 0.4794640896112453, + "grad_norm": 2.126931667327881, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6894908547401428, + "num_tokens": 111379419.0, + "step": 4366 + }, + { + "epoch": 0.479573907313859, + "grad_norm": 2.045140504837036, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.684228241443634, + "num_tokens": 111408486.0, + "step": 4367 + }, + { + "epoch": 0.47968372501647266, + "grad_norm": 2.2347376346588135, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6830448508262634, + "num_tokens": 111432722.0, + "step": 4368 + }, + { + "epoch": 0.4797935427190863, + "grad_norm": 2.077045202255249, + "learning_rate": 1e-06, + "loss": 1.0479, + "mean_token_accuracy": 0.6833328604698181, + "num_tokens": 111462276.0, + "step": 4369 + }, + { + "epoch": 0.47990336042169995, + "grad_norm": 2.1453897953033447, + "learning_rate": 1e-06, + "loss": 1.0526, + "mean_token_accuracy": 0.6903244853019714, + "num_tokens": 111489809.0, + "step": 4370 + }, + { + "epoch": 0.48001317812431366, + "grad_norm": 2.195859432220459, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7275282144546509, + "num_tokens": 111513713.0, + "step": 4371 + }, + { + "epoch": 0.4801229958269273, + "grad_norm": 2.2177023887634277, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.7028385400772095, + "num_tokens": 111539916.0, + "step": 4372 + }, + { + "epoch": 0.48023281352954095, + "grad_norm": 2.4889872074127197, + "learning_rate": 1e-06, + "loss": 1.0353, + "mean_token_accuracy": 0.6910111308097839, + "num_tokens": 111562229.0, + "step": 4373 + }, + { + "epoch": 0.48034263123215465, + "grad_norm": 2.1312129497528076, + "learning_rate": 1e-06, + "loss": 1.0439, + "mean_token_accuracy": 0.6807054281234741, + "num_tokens": 111591413.0, + "step": 4374 + }, + { + "epoch": 0.4804524489347683, + "grad_norm": 2.1122915744781494, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7059600353240967, + "num_tokens": 111618232.0, + "step": 4375 + }, + { + "epoch": 0.48056226663738194, + "grad_norm": 2.090909004211426, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.6977864503860474, + "num_tokens": 111645708.0, + "step": 4376 + }, + { + "epoch": 0.4806720843399956, + "grad_norm": 2.36511492729187, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7137545943260193, + "num_tokens": 111667397.0, + "step": 4377 + }, + { + "epoch": 0.4807819020426093, + "grad_norm": 2.1291098594665527, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7235939502716064, + "num_tokens": 111692896.0, + "step": 4378 + }, + { + "epoch": 0.48089171974522293, + "grad_norm": 2.423105239868164, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7110780477523804, + "num_tokens": 111715419.0, + "step": 4379 + }, + { + "epoch": 0.4810015374478366, + "grad_norm": 2.4964001178741455, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6911088824272156, + "num_tokens": 111736287.0, + "step": 4380 + }, + { + "epoch": 0.4811113551504502, + "grad_norm": 2.193988561630249, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7063840627670288, + "num_tokens": 111761829.0, + "step": 4381 + }, + { + "epoch": 0.4812211728530639, + "grad_norm": 2.0588436126708984, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7020784616470337, + "num_tokens": 111788296.0, + "step": 4382 + }, + { + "epoch": 0.4813309905556776, + "grad_norm": 2.297271490097046, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.71025550365448, + "num_tokens": 111810589.0, + "step": 4383 + }, + { + "epoch": 0.4814408082582912, + "grad_norm": 2.2743000984191895, + "learning_rate": 1e-06, + "loss": 1.0134, + "mean_token_accuracy": 0.6956234574317932, + "num_tokens": 111836266.0, + "step": 4384 + }, + { + "epoch": 0.4815506259609049, + "grad_norm": 2.3734683990478516, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.6985841989517212, + "num_tokens": 111857930.0, + "step": 4385 + }, + { + "epoch": 0.48166044366351857, + "grad_norm": 2.151519775390625, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7337090373039246, + "num_tokens": 111883139.0, + "step": 4386 + }, + { + "epoch": 0.4817702613661322, + "grad_norm": 2.6320223808288574, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7069604396820068, + "num_tokens": 111902791.0, + "step": 4387 + }, + { + "epoch": 0.48188007906874586, + "grad_norm": 2.2107930183410645, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7163400650024414, + "num_tokens": 111925401.0, + "step": 4388 + }, + { + "epoch": 0.48198989677135956, + "grad_norm": 2.2400553226470947, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7114280462265015, + "num_tokens": 111948884.0, + "step": 4389 + }, + { + "epoch": 0.4820997144739732, + "grad_norm": 2.464995861053467, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.7012428045272827, + "num_tokens": 111971116.0, + "step": 4390 + }, + { + "epoch": 0.48220953217658685, + "grad_norm": 2.1927552223205566, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.7123005986213684, + "num_tokens": 111997259.0, + "step": 4391 + }, + { + "epoch": 0.48231934987920055, + "grad_norm": 2.0714783668518066, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6949408054351807, + "num_tokens": 112025565.0, + "step": 4392 + }, + { + "epoch": 0.4824291675818142, + "grad_norm": 2.4592888355255127, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7118743658065796, + "num_tokens": 112045721.0, + "step": 4393 + }, + { + "epoch": 0.48253898528442785, + "grad_norm": 2.786755084991455, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7046651244163513, + "num_tokens": 112063070.0, + "step": 4394 + }, + { + "epoch": 0.4826488029870415, + "grad_norm": 2.0118231773376465, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7040525078773499, + "num_tokens": 112091681.0, + "step": 4395 + }, + { + "epoch": 0.4827586206896552, + "grad_norm": 2.0565152168273926, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7243555784225464, + "num_tokens": 112116756.0, + "step": 4396 + }, + { + "epoch": 0.48286843839226884, + "grad_norm": 2.5700600147247314, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7088690400123596, + "num_tokens": 112138577.0, + "step": 4397 + }, + { + "epoch": 0.4829782560948825, + "grad_norm": 2.515666961669922, + "learning_rate": 1e-06, + "loss": 0.9999, + "mean_token_accuracy": 0.6961916089057922, + "num_tokens": 112158808.0, + "step": 4398 + }, + { + "epoch": 0.48308807379749613, + "grad_norm": 2.526421546936035, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.7052841186523438, + "num_tokens": 112179485.0, + "step": 4399 + }, + { + "epoch": 0.48319789150010983, + "grad_norm": 2.2745819091796875, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7298367023468018, + "num_tokens": 112201874.0, + "step": 4400 + }, + { + "epoch": 0.4833077092027235, + "grad_norm": 2.1460633277893066, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7086098194122314, + "num_tokens": 112227364.0, + "step": 4401 + }, + { + "epoch": 0.4834175269053371, + "grad_norm": 2.3128654956817627, + "learning_rate": 1e-06, + "loss": 1.0533, + "mean_token_accuracy": 0.6836722493171692, + "num_tokens": 112249823.0, + "step": 4402 + }, + { + "epoch": 0.4835273446079508, + "grad_norm": 2.198575735092163, + "learning_rate": 1e-06, + "loss": 1.0715, + "mean_token_accuracy": 0.677266001701355, + "num_tokens": 112274781.0, + "step": 4403 + }, + { + "epoch": 0.4836371623105645, + "grad_norm": 2.0317652225494385, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7043766379356384, + "num_tokens": 112302514.0, + "step": 4404 + }, + { + "epoch": 0.4837469800131781, + "grad_norm": 2.041696786880493, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7162579298019409, + "num_tokens": 112330585.0, + "step": 4405 + }, + { + "epoch": 0.48385679771579176, + "grad_norm": 2.104022741317749, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.711979866027832, + "num_tokens": 112359683.0, + "step": 4406 + }, + { + "epoch": 0.48396661541840547, + "grad_norm": 2.2320051193237305, + "learning_rate": 1e-06, + "loss": 1.0685, + "mean_token_accuracy": 0.6894564628601074, + "num_tokens": 112387528.0, + "step": 4407 + }, + { + "epoch": 0.4840764331210191, + "grad_norm": 2.219881057739258, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.7003750205039978, + "num_tokens": 112412370.0, + "step": 4408 + }, + { + "epoch": 0.48418625082363276, + "grad_norm": 2.4446308612823486, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7038842439651489, + "num_tokens": 112433234.0, + "step": 4409 + }, + { + "epoch": 0.4842960685262464, + "grad_norm": 2.127147912979126, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.6994791030883789, + "num_tokens": 112462274.0, + "step": 4410 + }, + { + "epoch": 0.4844058862288601, + "grad_norm": 2.585202217102051, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7261176109313965, + "num_tokens": 112482084.0, + "step": 4411 + }, + { + "epoch": 0.48451570393147375, + "grad_norm": 2.307136297225952, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7160030603408813, + "num_tokens": 112505992.0, + "step": 4412 + }, + { + "epoch": 0.4846255216340874, + "grad_norm": 2.361006736755371, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7098859548568726, + "num_tokens": 112527397.0, + "step": 4413 + }, + { + "epoch": 0.4847353393367011, + "grad_norm": 2.3562099933624268, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7109721899032593, + "num_tokens": 112550778.0, + "step": 4414 + }, + { + "epoch": 0.48484515703931474, + "grad_norm": 2.2779476642608643, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6986026763916016, + "num_tokens": 112576576.0, + "step": 4415 + }, + { + "epoch": 0.4849549747419284, + "grad_norm": 2.6841635704040527, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7200403809547424, + "num_tokens": 112594365.0, + "step": 4416 + }, + { + "epoch": 0.48506479244454204, + "grad_norm": 2.3898487091064453, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.710767388343811, + "num_tokens": 112615958.0, + "step": 4417 + }, + { + "epoch": 0.48517461014715574, + "grad_norm": 2.1813459396362305, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6990574598312378, + "num_tokens": 112640980.0, + "step": 4418 + }, + { + "epoch": 0.4852844278497694, + "grad_norm": 2.195129871368408, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7231395244598389, + "num_tokens": 112665248.0, + "step": 4419 + }, + { + "epoch": 0.48539424555238303, + "grad_norm": 2.11044979095459, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6934247612953186, + "num_tokens": 112691482.0, + "step": 4420 + }, + { + "epoch": 0.48550406325499673, + "grad_norm": 2.4703497886657715, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6997954249382019, + "num_tokens": 112712360.0, + "step": 4421 + }, + { + "epoch": 0.4856138809576104, + "grad_norm": 2.5140633583068848, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.6962695121765137, + "num_tokens": 112734658.0, + "step": 4422 + }, + { + "epoch": 0.485723698660224, + "grad_norm": 2.0637598037719727, + "learning_rate": 1e-06, + "loss": 1.0722, + "mean_token_accuracy": 0.6842640042304993, + "num_tokens": 112763133.0, + "step": 4423 + }, + { + "epoch": 0.48583351636283767, + "grad_norm": 2.507812261581421, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7344434857368469, + "num_tokens": 112783794.0, + "step": 4424 + }, + { + "epoch": 0.48594333406545137, + "grad_norm": 2.4980595111846924, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7030422687530518, + "num_tokens": 112802826.0, + "step": 4425 + }, + { + "epoch": 0.486053151768065, + "grad_norm": 2.5723934173583984, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7101240158081055, + "num_tokens": 112822162.0, + "step": 4426 + }, + { + "epoch": 0.48616296947067866, + "grad_norm": 1.8888194561004639, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6858590841293335, + "num_tokens": 112853520.0, + "step": 4427 + }, + { + "epoch": 0.4862727871732923, + "grad_norm": 2.0935583114624023, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.7021385431289673, + "num_tokens": 112882035.0, + "step": 4428 + }, + { + "epoch": 0.486382604875906, + "grad_norm": 2.1594901084899902, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.7020815014839172, + "num_tokens": 112907338.0, + "step": 4429 + }, + { + "epoch": 0.48649242257851966, + "grad_norm": 2.2319982051849365, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.6969256401062012, + "num_tokens": 112933477.0, + "step": 4430 + }, + { + "epoch": 0.4866022402811333, + "grad_norm": 2.0229856967926025, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.6967177391052246, + "num_tokens": 112962154.0, + "step": 4431 + }, + { + "epoch": 0.486712057983747, + "grad_norm": 1.9674725532531738, + "learning_rate": 1e-06, + "loss": 1.108, + "mean_token_accuracy": 0.6742664575576782, + "num_tokens": 112995632.0, + "step": 4432 + }, + { + "epoch": 0.48682187568636065, + "grad_norm": 2.234989643096924, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6810086369514465, + "num_tokens": 113020125.0, + "step": 4433 + }, + { + "epoch": 0.4869316933889743, + "grad_norm": 2.0569941997528076, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7063002586364746, + "num_tokens": 113048443.0, + "step": 4434 + }, + { + "epoch": 0.48704151109158794, + "grad_norm": 2.4393162727355957, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.7005115151405334, + "num_tokens": 113072060.0, + "step": 4435 + }, + { + "epoch": 0.48715132879420164, + "grad_norm": 1.8847843408584595, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.7002053260803223, + "num_tokens": 113104038.0, + "step": 4436 + }, + { + "epoch": 0.4872611464968153, + "grad_norm": 2.322240114212036, + "learning_rate": 1e-06, + "loss": 1.0354, + "mean_token_accuracy": 0.689134955406189, + "num_tokens": 113126778.0, + "step": 4437 + }, + { + "epoch": 0.48737096419942894, + "grad_norm": 1.996114730834961, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7010877728462219, + "num_tokens": 113159064.0, + "step": 4438 + }, + { + "epoch": 0.48748078190204264, + "grad_norm": 2.1893739700317383, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6868486404418945, + "num_tokens": 113185609.0, + "step": 4439 + }, + { + "epoch": 0.4875905996046563, + "grad_norm": 2.231365919113159, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.7001031637191772, + "num_tokens": 113211059.0, + "step": 4440 + }, + { + "epoch": 0.48770041730726993, + "grad_norm": 2.13112735748291, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7328630685806274, + "num_tokens": 113236044.0, + "step": 4441 + }, + { + "epoch": 0.4878102350098836, + "grad_norm": 2.26104736328125, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.714266836643219, + "num_tokens": 113259664.0, + "step": 4442 + }, + { + "epoch": 0.4879200527124973, + "grad_norm": 2.133253574371338, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.698763370513916, + "num_tokens": 113286985.0, + "step": 4443 + }, + { + "epoch": 0.4880298704151109, + "grad_norm": 2.4513378143310547, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7298484444618225, + "num_tokens": 113305677.0, + "step": 4444 + }, + { + "epoch": 0.48813968811772457, + "grad_norm": 2.272123098373413, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.6886545419692993, + "num_tokens": 113329845.0, + "step": 4445 + }, + { + "epoch": 0.4882495058203382, + "grad_norm": 2.339599609375, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7018086910247803, + "num_tokens": 113352127.0, + "step": 4446 + }, + { + "epoch": 0.4883593235229519, + "grad_norm": 2.382392406463623, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.7065315842628479, + "num_tokens": 113375076.0, + "step": 4447 + }, + { + "epoch": 0.48846914122556556, + "grad_norm": 2.442070484161377, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7220410108566284, + "num_tokens": 113394880.0, + "step": 4448 + }, + { + "epoch": 0.4885789589281792, + "grad_norm": 2.271010398864746, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7064496278762817, + "num_tokens": 113418731.0, + "step": 4449 + }, + { + "epoch": 0.4886887766307929, + "grad_norm": 2.361920118331909, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7196031808853149, + "num_tokens": 113440365.0, + "step": 4450 + }, + { + "epoch": 0.48879859433340656, + "grad_norm": 2.113680124282837, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.6957157850265503, + "num_tokens": 113469157.0, + "step": 4451 + }, + { + "epoch": 0.4889084120360202, + "grad_norm": 2.0750555992126465, + "learning_rate": 1e-06, + "loss": 1.036, + "mean_token_accuracy": 0.6808123588562012, + "num_tokens": 113497271.0, + "step": 4452 + }, + { + "epoch": 0.48901822973863385, + "grad_norm": 1.9671225547790527, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7125951051712036, + "num_tokens": 113526870.0, + "step": 4453 + }, + { + "epoch": 0.48912804744124755, + "grad_norm": 2.1735095977783203, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7218838930130005, + "num_tokens": 113550642.0, + "step": 4454 + }, + { + "epoch": 0.4892378651438612, + "grad_norm": 2.202315092086792, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6835695505142212, + "num_tokens": 113578888.0, + "step": 4455 + }, + { + "epoch": 0.48934768284647484, + "grad_norm": 2.425081729888916, + "learning_rate": 1e-06, + "loss": 1.0648, + "mean_token_accuracy": 0.6838072538375854, + "num_tokens": 113602365.0, + "step": 4456 + }, + { + "epoch": 0.4894575005490885, + "grad_norm": 2.1845548152923584, + "learning_rate": 1e-06, + "loss": 1.0493, + "mean_token_accuracy": 0.6829911470413208, + "num_tokens": 113628079.0, + "step": 4457 + }, + { + "epoch": 0.4895673182517022, + "grad_norm": 2.1964287757873535, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7132467031478882, + "num_tokens": 113655998.0, + "step": 4458 + }, + { + "epoch": 0.48967713595431583, + "grad_norm": 2.41605806350708, + "learning_rate": 1e-06, + "loss": 1.0512, + "mean_token_accuracy": 0.6847684979438782, + "num_tokens": 113677089.0, + "step": 4459 + }, + { + "epoch": 0.4897869536569295, + "grad_norm": 2.4790194034576416, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7172290682792664, + "num_tokens": 113697441.0, + "step": 4460 + }, + { + "epoch": 0.4898967713595432, + "grad_norm": 2.2884607315063477, + "learning_rate": 1e-06, + "loss": 1.0591, + "mean_token_accuracy": 0.686306357383728, + "num_tokens": 113722546.0, + "step": 4461 + }, + { + "epoch": 0.4900065890621568, + "grad_norm": 2.0440807342529297, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7001122832298279, + "num_tokens": 113751647.0, + "step": 4462 + }, + { + "epoch": 0.4901164067647705, + "grad_norm": 2.060798168182373, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7087452411651611, + "num_tokens": 113779080.0, + "step": 4463 + }, + { + "epoch": 0.4902262244673841, + "grad_norm": 2.4506325721740723, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7134630680084229, + "num_tokens": 113800604.0, + "step": 4464 + }, + { + "epoch": 0.4903360421699978, + "grad_norm": 2.8913729190826416, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7213786840438843, + "num_tokens": 113816863.0, + "step": 4465 + }, + { + "epoch": 0.49044585987261147, + "grad_norm": 2.123102903366089, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7192637920379639, + "num_tokens": 113841343.0, + "step": 4466 + }, + { + "epoch": 0.4905556775752251, + "grad_norm": 2.1076278686523438, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.7016987204551697, + "num_tokens": 113869427.0, + "step": 4467 + }, + { + "epoch": 0.4906654952778388, + "grad_norm": 1.8674912452697754, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6968557238578796, + "num_tokens": 113902094.0, + "step": 4468 + }, + { + "epoch": 0.49077531298045246, + "grad_norm": 2.130331516265869, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6980266571044922, + "num_tokens": 113929212.0, + "step": 4469 + }, + { + "epoch": 0.4908851306830661, + "grad_norm": 1.7767667770385742, + "learning_rate": 1e-06, + "loss": 1.0713, + "mean_token_accuracy": 0.675952672958374, + "num_tokens": 113969133.0, + "step": 4470 + }, + { + "epoch": 0.49099494838567975, + "grad_norm": 4.528143882751465, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6949443817138672, + "num_tokens": 113992620.0, + "step": 4471 + }, + { + "epoch": 0.49110476608829345, + "grad_norm": 2.034607410430908, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.6878692507743835, + "num_tokens": 114021974.0, + "step": 4472 + }, + { + "epoch": 0.4912145837909071, + "grad_norm": 2.0012636184692383, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.6942939758300781, + "num_tokens": 114050439.0, + "step": 4473 + }, + { + "epoch": 0.49132440149352075, + "grad_norm": 2.7693428993225098, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7197791934013367, + "num_tokens": 114067836.0, + "step": 4474 + }, + { + "epoch": 0.4914342191961344, + "grad_norm": 1.9709033966064453, + "learning_rate": 1e-06, + "loss": 1.0821, + "mean_token_accuracy": 0.6696513891220093, + "num_tokens": 114099138.0, + "step": 4475 + }, + { + "epoch": 0.4915440368987481, + "grad_norm": 2.6023669242858887, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.700253963470459, + "num_tokens": 114117404.0, + "step": 4476 + }, + { + "epoch": 0.49165385460136174, + "grad_norm": 2.1005988121032715, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7113110423088074, + "num_tokens": 114144239.0, + "step": 4477 + }, + { + "epoch": 0.4917636723039754, + "grad_norm": 2.010871648788452, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6909297704696655, + "num_tokens": 114176133.0, + "step": 4478 + }, + { + "epoch": 0.4918734900065891, + "grad_norm": 1.9898676872253418, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6986950039863586, + "num_tokens": 114205844.0, + "step": 4479 + }, + { + "epoch": 0.49198330770920273, + "grad_norm": 2.2995517253875732, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7098366618156433, + "num_tokens": 114230253.0, + "step": 4480 + }, + { + "epoch": 0.4920931254118164, + "grad_norm": 2.118238687515259, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7065514922142029, + "num_tokens": 114256053.0, + "step": 4481 + }, + { + "epoch": 0.49220294311443, + "grad_norm": 2.2047531604766846, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.724197268486023, + "num_tokens": 114282364.0, + "step": 4482 + }, + { + "epoch": 0.4923127608170437, + "grad_norm": 2.020001173019409, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6974986791610718, + "num_tokens": 114312123.0, + "step": 4483 + }, + { + "epoch": 0.4924225785196574, + "grad_norm": 2.218496084213257, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.689975380897522, + "num_tokens": 114336936.0, + "step": 4484 + }, + { + "epoch": 0.492532396222271, + "grad_norm": 2.602642297744751, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7111987471580505, + "num_tokens": 114357001.0, + "step": 4485 + }, + { + "epoch": 0.49264221392488466, + "grad_norm": 2.332707643508911, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6953100562095642, + "num_tokens": 114383837.0, + "step": 4486 + }, + { + "epoch": 0.49275203162749837, + "grad_norm": 2.03755259513855, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.706454336643219, + "num_tokens": 114413283.0, + "step": 4487 + }, + { + "epoch": 0.492861849330112, + "grad_norm": 2.3657608032226562, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.7055878043174744, + "num_tokens": 114437445.0, + "step": 4488 + }, + { + "epoch": 0.49297166703272566, + "grad_norm": 2.2075417041778564, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.70921790599823, + "num_tokens": 114462073.0, + "step": 4489 + }, + { + "epoch": 0.49308148473533936, + "grad_norm": 2.2301828861236572, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7021437883377075, + "num_tokens": 114486566.0, + "step": 4490 + }, + { + "epoch": 0.493191302437953, + "grad_norm": 2.1911516189575195, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7005348801612854, + "num_tokens": 114511988.0, + "step": 4491 + }, + { + "epoch": 0.49330112014056665, + "grad_norm": 2.0680108070373535, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7014439105987549, + "num_tokens": 114539904.0, + "step": 4492 + }, + { + "epoch": 0.4934109378431803, + "grad_norm": 2.1969878673553467, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6974038481712341, + "num_tokens": 114565860.0, + "step": 4493 + }, + { + "epoch": 0.493520755545794, + "grad_norm": 2.3243114948272705, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7100936770439148, + "num_tokens": 114589324.0, + "step": 4494 + }, + { + "epoch": 0.49363057324840764, + "grad_norm": 2.2940049171447754, + "learning_rate": 1e-06, + "loss": 1.08, + "mean_token_accuracy": 0.6770992279052734, + "num_tokens": 114613101.0, + "step": 4495 + }, + { + "epoch": 0.4937403909510213, + "grad_norm": 1.915385127067566, + "learning_rate": 1e-06, + "loss": 1.0592, + "mean_token_accuracy": 0.6811593770980835, + "num_tokens": 114648263.0, + "step": 4496 + }, + { + "epoch": 0.493850208653635, + "grad_norm": 2.4101521968841553, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.7012854814529419, + "num_tokens": 114671403.0, + "step": 4497 + }, + { + "epoch": 0.49396002635624864, + "grad_norm": 2.407362699508667, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7233896255493164, + "num_tokens": 114692865.0, + "step": 4498 + }, + { + "epoch": 0.4940698440588623, + "grad_norm": 2.295358657836914, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7136377096176147, + "num_tokens": 114715824.0, + "step": 4499 + }, + { + "epoch": 0.49417966176147593, + "grad_norm": 2.1982696056365967, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6873499155044556, + "num_tokens": 114742754.0, + "step": 4500 + }, + { + "epoch": 0.49428947946408963, + "grad_norm": 2.2508959770202637, + "learning_rate": 1e-06, + "loss": 1.069, + "mean_token_accuracy": 0.6961073875427246, + "num_tokens": 114766187.0, + "step": 4501 + }, + { + "epoch": 0.4943992971667033, + "grad_norm": 2.0405821800231934, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6994284391403198, + "num_tokens": 114795392.0, + "step": 4502 + }, + { + "epoch": 0.4945091148693169, + "grad_norm": 2.199878215789795, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7087010145187378, + "num_tokens": 114820784.0, + "step": 4503 + }, + { + "epoch": 0.49461893257193057, + "grad_norm": 2.1514742374420166, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7142295241355896, + "num_tokens": 114845733.0, + "step": 4504 + }, + { + "epoch": 0.49472875027454427, + "grad_norm": 2.2842376232147217, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7345670461654663, + "num_tokens": 114868336.0, + "step": 4505 + }, + { + "epoch": 0.4948385679771579, + "grad_norm": 2.228677988052368, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7121051549911499, + "num_tokens": 114890648.0, + "step": 4506 + }, + { + "epoch": 0.49494838567977156, + "grad_norm": 2.2605738639831543, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6880106925964355, + "num_tokens": 114915761.0, + "step": 4507 + }, + { + "epoch": 0.49505820338238526, + "grad_norm": 2.2335734367370605, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7088075876235962, + "num_tokens": 114939245.0, + "step": 4508 + }, + { + "epoch": 0.4951680210849989, + "grad_norm": 2.2318992614746094, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7196868658065796, + "num_tokens": 114967382.0, + "step": 4509 + }, + { + "epoch": 0.49527783878761256, + "grad_norm": 2.3250467777252197, + "learning_rate": 1e-06, + "loss": 1.0929, + "mean_token_accuracy": 0.6802351474761963, + "num_tokens": 114993762.0, + "step": 4510 + }, + { + "epoch": 0.4953876564902262, + "grad_norm": 2.08912992477417, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.7087349891662598, + "num_tokens": 115022447.0, + "step": 4511 + }, + { + "epoch": 0.4954974741928399, + "grad_norm": 2.4429283142089844, + "learning_rate": 1e-06, + "loss": 1.068, + "mean_token_accuracy": 0.6952711343765259, + "num_tokens": 115045994.0, + "step": 4512 + }, + { + "epoch": 0.49560729189545355, + "grad_norm": 2.1912221908569336, + "learning_rate": 1e-06, + "loss": 1.0538, + "mean_token_accuracy": 0.6781548261642456, + "num_tokens": 115071796.0, + "step": 4513 + }, + { + "epoch": 0.4957171095980672, + "grad_norm": 2.2366139888763428, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7461024522781372, + "num_tokens": 115093522.0, + "step": 4514 + }, + { + "epoch": 0.4958269273006809, + "grad_norm": 2.428413152694702, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7138630747795105, + "num_tokens": 115113624.0, + "step": 4515 + }, + { + "epoch": 0.49593674500329454, + "grad_norm": 2.0896029472351074, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.7035413384437561, + "num_tokens": 115142147.0, + "step": 4516 + }, + { + "epoch": 0.4960465627059082, + "grad_norm": 2.3957927227020264, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.701094925403595, + "num_tokens": 115162873.0, + "step": 4517 + }, + { + "epoch": 0.49615638040852184, + "grad_norm": 2.259692907333374, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7042191028594971, + "num_tokens": 115187242.0, + "step": 4518 + }, + { + "epoch": 0.49626619811113554, + "grad_norm": 2.349257230758667, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7024945020675659, + "num_tokens": 115208779.0, + "step": 4519 + }, + { + "epoch": 0.4963760158137492, + "grad_norm": 2.4570207595825195, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7228049635887146, + "num_tokens": 115230516.0, + "step": 4520 + }, + { + "epoch": 0.49648583351636283, + "grad_norm": 2.0230860710144043, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.70574951171875, + "num_tokens": 115259560.0, + "step": 4521 + }, + { + "epoch": 0.4965956512189765, + "grad_norm": 2.130927085876465, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6918302774429321, + "num_tokens": 115289813.0, + "step": 4522 + }, + { + "epoch": 0.4967054689215902, + "grad_norm": 2.316441774368286, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6993153095245361, + "num_tokens": 115313756.0, + "step": 4523 + }, + { + "epoch": 0.4968152866242038, + "grad_norm": 2.0397725105285645, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7175283432006836, + "num_tokens": 115342626.0, + "step": 4524 + }, + { + "epoch": 0.49692510432681747, + "grad_norm": 2.47255277633667, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7164172530174255, + "num_tokens": 115362432.0, + "step": 4525 + }, + { + "epoch": 0.49703492202943117, + "grad_norm": 2.3920397758483887, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.715548038482666, + "num_tokens": 115384425.0, + "step": 4526 + }, + { + "epoch": 0.4971447397320448, + "grad_norm": 2.1888539791107178, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.703021228313446, + "num_tokens": 115409324.0, + "step": 4527 + }, + { + "epoch": 0.49725455743465846, + "grad_norm": 2.4781415462493896, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7022066116333008, + "num_tokens": 115429527.0, + "step": 4528 + }, + { + "epoch": 0.4973643751372721, + "grad_norm": 2.2170662879943848, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.709805965423584, + "num_tokens": 115453128.0, + "step": 4529 + }, + { + "epoch": 0.4974741928398858, + "grad_norm": 2.345019578933716, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7026993036270142, + "num_tokens": 115476677.0, + "step": 4530 + }, + { + "epoch": 0.49758401054249946, + "grad_norm": 2.2226428985595703, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7082817554473877, + "num_tokens": 115500684.0, + "step": 4531 + }, + { + "epoch": 0.4976938282451131, + "grad_norm": 2.206852912902832, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7083272933959961, + "num_tokens": 115526710.0, + "step": 4532 + }, + { + "epoch": 0.49780364594772675, + "grad_norm": 2.2297942638397217, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7006583213806152, + "num_tokens": 115550307.0, + "step": 4533 + }, + { + "epoch": 0.49791346365034045, + "grad_norm": 2.649834632873535, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7417588829994202, + "num_tokens": 115570092.0, + "step": 4534 + }, + { + "epoch": 0.4980232813529541, + "grad_norm": 2.3079488277435303, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6966738700866699, + "num_tokens": 115592167.0, + "step": 4535 + }, + { + "epoch": 0.49813309905556774, + "grad_norm": 2.1191771030426025, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7232579588890076, + "num_tokens": 115619310.0, + "step": 4536 + }, + { + "epoch": 0.49824291675818144, + "grad_norm": 2.3251876831054688, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7081601619720459, + "num_tokens": 115643306.0, + "step": 4537 + }, + { + "epoch": 0.4983527344607951, + "grad_norm": 2.200134515762329, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.7011744379997253, + "num_tokens": 115668353.0, + "step": 4538 + }, + { + "epoch": 0.49846255216340873, + "grad_norm": 2.0970051288604736, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7112618088722229, + "num_tokens": 115695669.0, + "step": 4539 + }, + { + "epoch": 0.4985723698660224, + "grad_norm": 2.2301955223083496, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6878528594970703, + "num_tokens": 115721271.0, + "step": 4540 + }, + { + "epoch": 0.4986821875686361, + "grad_norm": 2.2465641498565674, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7057624459266663, + "num_tokens": 115746085.0, + "step": 4541 + }, + { + "epoch": 0.4987920052712497, + "grad_norm": 2.6832776069641113, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.69460129737854, + "num_tokens": 115764407.0, + "step": 4542 + }, + { + "epoch": 0.4989018229738634, + "grad_norm": 2.3024885654449463, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7064656615257263, + "num_tokens": 115787123.0, + "step": 4543 + }, + { + "epoch": 0.4990116406764771, + "grad_norm": 2.047889471054077, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7131142616271973, + "num_tokens": 115813709.0, + "step": 4544 + }, + { + "epoch": 0.4991214583790907, + "grad_norm": 2.2170987129211426, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6868884563446045, + "num_tokens": 115839817.0, + "step": 4545 + }, + { + "epoch": 0.49923127608170437, + "grad_norm": 2.1324775218963623, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.697830080986023, + "num_tokens": 115867035.0, + "step": 4546 + }, + { + "epoch": 0.499341093784318, + "grad_norm": 2.2561073303222656, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.6876406073570251, + "num_tokens": 115894036.0, + "step": 4547 + }, + { + "epoch": 0.4994509114869317, + "grad_norm": 2.1600754261016846, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7010083198547363, + "num_tokens": 115920748.0, + "step": 4548 + }, + { + "epoch": 0.49956072918954536, + "grad_norm": 2.1879777908325195, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.692082405090332, + "num_tokens": 115944997.0, + "step": 4549 + }, + { + "epoch": 0.499670546892159, + "grad_norm": 1.9823607206344604, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.7052869200706482, + "num_tokens": 115972980.0, + "step": 4550 + }, + { + "epoch": 0.49978036459477265, + "grad_norm": 2.042678117752075, + "learning_rate": 1e-06, + "loss": 1.0588, + "mean_token_accuracy": 0.6797553300857544, + "num_tokens": 116002059.0, + "step": 4551 + }, + { + "epoch": 0.49989018229738635, + "grad_norm": 2.347764492034912, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7063676118850708, + "num_tokens": 116024719.0, + "step": 4552 + }, + { + "epoch": 0.5, + "grad_norm": 2.0623321533203125, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7070022225379944, + "num_tokens": 116051860.0, + "step": 4553 + }, + { + "epoch": 0.5001098177026136, + "grad_norm": 2.167038917541504, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7100043296813965, + "num_tokens": 116076478.0, + "step": 4554 + }, + { + "epoch": 0.5002196354052273, + "grad_norm": 2.144383430480957, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7098073959350586, + "num_tokens": 116101671.0, + "step": 4555 + }, + { + "epoch": 0.5003294531078409, + "grad_norm": 2.165825605392456, + "learning_rate": 1e-06, + "loss": 1.0763, + "mean_token_accuracy": 0.6738517880439758, + "num_tokens": 116128457.0, + "step": 4556 + }, + { + "epoch": 0.5004392708104547, + "grad_norm": 2.3220255374908447, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6896852254867554, + "num_tokens": 116152909.0, + "step": 4557 + }, + { + "epoch": 0.5005490885130683, + "grad_norm": 1.9486576318740845, + "learning_rate": 1e-06, + "loss": 1.1065, + "mean_token_accuracy": 0.6687237620353699, + "num_tokens": 116184145.0, + "step": 4558 + }, + { + "epoch": 0.500658906215682, + "grad_norm": 2.181321620941162, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7033896446228027, + "num_tokens": 116210154.0, + "step": 4559 + }, + { + "epoch": 0.5007687239182956, + "grad_norm": 2.692282199859619, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7317928075790405, + "num_tokens": 116226426.0, + "step": 4560 + }, + { + "epoch": 0.5008785416209093, + "grad_norm": 2.269644021987915, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.6933484673500061, + "num_tokens": 116249622.0, + "step": 4561 + }, + { + "epoch": 0.5009883593235229, + "grad_norm": 2.4833791255950928, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.6967607140541077, + "num_tokens": 116269137.0, + "step": 4562 + }, + { + "epoch": 0.5010981770261366, + "grad_norm": 2.1203503608703613, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.713772177696228, + "num_tokens": 116293971.0, + "step": 4563 + }, + { + "epoch": 0.5012079947287503, + "grad_norm": 2.4748542308807373, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.7096431851387024, + "num_tokens": 116313441.0, + "step": 4564 + }, + { + "epoch": 0.501317812431364, + "grad_norm": 2.194847583770752, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.6971122622489929, + "num_tokens": 116338563.0, + "step": 4565 + }, + { + "epoch": 0.5014276301339776, + "grad_norm": 2.5869486331939697, + "learning_rate": 1e-06, + "loss": 1.0416, + "mean_token_accuracy": 0.687341570854187, + "num_tokens": 116358053.0, + "step": 4566 + }, + { + "epoch": 0.5015374478365913, + "grad_norm": 2.4606823921203613, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6918426752090454, + "num_tokens": 116380984.0, + "step": 4567 + }, + { + "epoch": 0.5016472655392049, + "grad_norm": 2.2869136333465576, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7065647840499878, + "num_tokens": 116405750.0, + "step": 4568 + }, + { + "epoch": 0.5017570832418186, + "grad_norm": 2.291489839553833, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7181473970413208, + "num_tokens": 116429542.0, + "step": 4569 + }, + { + "epoch": 0.5018669009444322, + "grad_norm": 2.378938674926758, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6865718960762024, + "num_tokens": 116452767.0, + "step": 4570 + }, + { + "epoch": 0.5019767186470458, + "grad_norm": 2.253965139389038, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7197239398956299, + "num_tokens": 116477770.0, + "step": 4571 + }, + { + "epoch": 0.5020865363496596, + "grad_norm": 2.2020976543426514, + "learning_rate": 1e-06, + "loss": 1.0443, + "mean_token_accuracy": 0.6855665445327759, + "num_tokens": 116504819.0, + "step": 4572 + }, + { + "epoch": 0.5021963540522733, + "grad_norm": 2.2067291736602783, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.6921154260635376, + "num_tokens": 116530227.0, + "step": 4573 + }, + { + "epoch": 0.5023061717548869, + "grad_norm": 2.3915319442749023, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6920216679573059, + "num_tokens": 116552784.0, + "step": 4574 + }, + { + "epoch": 0.5024159894575005, + "grad_norm": 2.1820597648620605, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6983482837677002, + "num_tokens": 116579969.0, + "step": 4575 + }, + { + "epoch": 0.5025258071601142, + "grad_norm": 2.446296453475952, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7065107822418213, + "num_tokens": 116603025.0, + "step": 4576 + }, + { + "epoch": 0.5026356248627278, + "grad_norm": 2.250333786010742, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7033315300941467, + "num_tokens": 116626285.0, + "step": 4577 + }, + { + "epoch": 0.5027454425653415, + "grad_norm": 2.5323400497436523, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7276511788368225, + "num_tokens": 116646440.0, + "step": 4578 + }, + { + "epoch": 0.5028552602679552, + "grad_norm": 2.509061098098755, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7129729390144348, + "num_tokens": 116668060.0, + "step": 4579 + }, + { + "epoch": 0.5029650779705689, + "grad_norm": 2.2816529273986816, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.6974514722824097, + "num_tokens": 116691149.0, + "step": 4580 + }, + { + "epoch": 0.5030748956731825, + "grad_norm": 2.0212481021881104, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7119846343994141, + "num_tokens": 116717882.0, + "step": 4581 + }, + { + "epoch": 0.5031847133757962, + "grad_norm": 2.32049822807312, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7107466459274292, + "num_tokens": 116741630.0, + "step": 4582 + }, + { + "epoch": 0.5032945310784098, + "grad_norm": 2.430476188659668, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.7011375427246094, + "num_tokens": 116764931.0, + "step": 4583 + }, + { + "epoch": 0.5034043487810235, + "grad_norm": 2.3713507652282715, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7089200019836426, + "num_tokens": 116787398.0, + "step": 4584 + }, + { + "epoch": 0.5035141664836371, + "grad_norm": 1.8806861639022827, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7134907245635986, + "num_tokens": 116819092.0, + "step": 4585 + }, + { + "epoch": 0.5036239841862509, + "grad_norm": 1.9843974113464355, + "learning_rate": 1e-06, + "loss": 1.0801, + "mean_token_accuracy": 0.6794205904006958, + "num_tokens": 116851055.0, + "step": 4586 + }, + { + "epoch": 0.5037338018888645, + "grad_norm": 1.8818144798278809, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.6955275535583496, + "num_tokens": 116882253.0, + "step": 4587 + }, + { + "epoch": 0.5038436195914782, + "grad_norm": 2.22914457321167, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.705003559589386, + "num_tokens": 116907302.0, + "step": 4588 + }, + { + "epoch": 0.5039534372940918, + "grad_norm": 2.1906821727752686, + "learning_rate": 1e-06, + "loss": 1.0452, + "mean_token_accuracy": 0.6948920488357544, + "num_tokens": 116933679.0, + "step": 4589 + }, + { + "epoch": 0.5040632549967055, + "grad_norm": 1.8406846523284912, + "learning_rate": 1e-06, + "loss": 1.1405, + "mean_token_accuracy": 0.6728166341781616, + "num_tokens": 116971498.0, + "step": 4590 + }, + { + "epoch": 0.5041730726993191, + "grad_norm": 2.0252084732055664, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.702238917350769, + "num_tokens": 117002038.0, + "step": 4591 + }, + { + "epoch": 0.5042828904019327, + "grad_norm": 2.1975626945495605, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7152794003486633, + "num_tokens": 117027436.0, + "step": 4592 + }, + { + "epoch": 0.5043927081045465, + "grad_norm": 2.1343953609466553, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.702170729637146, + "num_tokens": 117055338.0, + "step": 4593 + }, + { + "epoch": 0.5045025258071602, + "grad_norm": 2.4709882736206055, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7121544480323792, + "num_tokens": 117078043.0, + "step": 4594 + }, + { + "epoch": 0.5046123435097738, + "grad_norm": 2.2196807861328125, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.693699300289154, + "num_tokens": 117105042.0, + "step": 4595 + }, + { + "epoch": 0.5047221612123874, + "grad_norm": 2.1429948806762695, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7285685539245605, + "num_tokens": 117131368.0, + "step": 4596 + }, + { + "epoch": 0.5048319789150011, + "grad_norm": 2.2113025188446045, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.7040940523147583, + "num_tokens": 117156967.0, + "step": 4597 + }, + { + "epoch": 0.5049417966176147, + "grad_norm": 2.2654001712799072, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7026723027229309, + "num_tokens": 117179817.0, + "step": 4598 + }, + { + "epoch": 0.5050516143202284, + "grad_norm": 2.1857588291168213, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.6978307962417603, + "num_tokens": 117205782.0, + "step": 4599 + }, + { + "epoch": 0.505161432022842, + "grad_norm": 2.3037853240966797, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6895781755447388, + "num_tokens": 117230627.0, + "step": 4600 + }, + { + "epoch": 0.5052712497254558, + "grad_norm": 2.4445013999938965, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.6931823492050171, + "num_tokens": 117252591.0, + "step": 4601 + }, + { + "epoch": 0.5053810674280694, + "grad_norm": 2.104004144668579, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6913268566131592, + "num_tokens": 117279238.0, + "step": 4602 + }, + { + "epoch": 0.5054908851306831, + "grad_norm": 2.1610865592956543, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6908218860626221, + "num_tokens": 117305644.0, + "step": 4603 + }, + { + "epoch": 0.5056007028332967, + "grad_norm": 2.286452054977417, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7157329320907593, + "num_tokens": 117328814.0, + "step": 4604 + }, + { + "epoch": 0.5057105205359104, + "grad_norm": 2.2634971141815186, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7186634540557861, + "num_tokens": 117354080.0, + "step": 4605 + }, + { + "epoch": 0.505820338238524, + "grad_norm": 2.1910979747772217, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.6893458366394043, + "num_tokens": 117381381.0, + "step": 4606 + }, + { + "epoch": 0.5059301559411377, + "grad_norm": 2.0981051921844482, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7084826231002808, + "num_tokens": 117408294.0, + "step": 4607 + }, + { + "epoch": 0.5060399736437514, + "grad_norm": 2.1010499000549316, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6947576999664307, + "num_tokens": 117434769.0, + "step": 4608 + }, + { + "epoch": 0.5061497913463651, + "grad_norm": 2.2633216381073, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.6988828182220459, + "num_tokens": 117460165.0, + "step": 4609 + }, + { + "epoch": 0.5062596090489787, + "grad_norm": 2.0039255619049072, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6984536647796631, + "num_tokens": 117491725.0, + "step": 4610 + }, + { + "epoch": 0.5063694267515924, + "grad_norm": 2.202346086502075, + "learning_rate": 1e-06, + "loss": 1.1119, + "mean_token_accuracy": 0.6683831214904785, + "num_tokens": 117520226.0, + "step": 4611 + }, + { + "epoch": 0.506479244454206, + "grad_norm": 1.9508593082427979, + "learning_rate": 1e-06, + "loss": 1.0362, + "mean_token_accuracy": 0.6927490234375, + "num_tokens": 117552738.0, + "step": 4612 + }, + { + "epoch": 0.5065890621568196, + "grad_norm": 2.040940284729004, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.724094033241272, + "num_tokens": 117580304.0, + "step": 4613 + }, + { + "epoch": 0.5066988798594333, + "grad_norm": 2.218503952026367, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7042279839515686, + "num_tokens": 117605776.0, + "step": 4614 + }, + { + "epoch": 0.506808697562047, + "grad_norm": 2.1878809928894043, + "learning_rate": 1e-06, + "loss": 1.0509, + "mean_token_accuracy": 0.6797601580619812, + "num_tokens": 117632888.0, + "step": 4615 + }, + { + "epoch": 0.5069185152646607, + "grad_norm": 2.2421836853027344, + "learning_rate": 1e-06, + "loss": 1.0641, + "mean_token_accuracy": 0.6843932271003723, + "num_tokens": 117659102.0, + "step": 4616 + }, + { + "epoch": 0.5070283329672743, + "grad_norm": 2.0285751819610596, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.6923166513442993, + "num_tokens": 117687725.0, + "step": 4617 + }, + { + "epoch": 0.507138150669888, + "grad_norm": 2.1111693382263184, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6878426671028137, + "num_tokens": 117715227.0, + "step": 4618 + }, + { + "epoch": 0.5072479683725016, + "grad_norm": 2.2321648597717285, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7136317491531372, + "num_tokens": 117740894.0, + "step": 4619 + }, + { + "epoch": 0.5073577860751153, + "grad_norm": 2.3193273544311523, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7260428667068481, + "num_tokens": 117761840.0, + "step": 4620 + }, + { + "epoch": 0.5074676037777289, + "grad_norm": 2.313234567642212, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7023500204086304, + "num_tokens": 117785559.0, + "step": 4621 + }, + { + "epoch": 0.5075774214803427, + "grad_norm": 2.4006290435791016, + "learning_rate": 1e-06, + "loss": 1.0279, + "mean_token_accuracy": 0.6912297010421753, + "num_tokens": 117809085.0, + "step": 4622 + }, + { + "epoch": 0.5076872391829563, + "grad_norm": 2.086287498474121, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.710397481918335, + "num_tokens": 117835824.0, + "step": 4623 + }, + { + "epoch": 0.50779705688557, + "grad_norm": 2.172233819961548, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7069758176803589, + "num_tokens": 117860721.0, + "step": 4624 + }, + { + "epoch": 0.5079068745881836, + "grad_norm": 2.0513484477996826, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.703819751739502, + "num_tokens": 117889856.0, + "step": 4625 + }, + { + "epoch": 0.5080166922907973, + "grad_norm": 2.4153428077697754, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6939414143562317, + "num_tokens": 117912476.0, + "step": 4626 + }, + { + "epoch": 0.5081265099934109, + "grad_norm": 2.147826910018921, + "learning_rate": 1e-06, + "loss": 1.0488, + "mean_token_accuracy": 0.686058759689331, + "num_tokens": 117939851.0, + "step": 4627 + }, + { + "epoch": 0.5082363276960246, + "grad_norm": 2.0149412155151367, + "learning_rate": 1e-06, + "loss": 1.0695, + "mean_token_accuracy": 0.6784789562225342, + "num_tokens": 117969171.0, + "step": 4628 + }, + { + "epoch": 0.5083461453986382, + "grad_norm": 1.954176425933838, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7081047296524048, + "num_tokens": 117999473.0, + "step": 4629 + }, + { + "epoch": 0.508455963101252, + "grad_norm": 2.171656608581543, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7055543661117554, + "num_tokens": 118026151.0, + "step": 4630 + }, + { + "epoch": 0.5085657808038656, + "grad_norm": 2.1660854816436768, + "learning_rate": 1e-06, + "loss": 1.0477, + "mean_token_accuracy": 0.6868722438812256, + "num_tokens": 118054388.0, + "step": 4631 + }, + { + "epoch": 0.5086755985064793, + "grad_norm": 2.1519064903259277, + "learning_rate": 1e-06, + "loss": 1.0159, + "mean_token_accuracy": 0.6950365900993347, + "num_tokens": 118082081.0, + "step": 4632 + }, + { + "epoch": 0.5087854162090929, + "grad_norm": 2.266299247741699, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6984066963195801, + "num_tokens": 118106795.0, + "step": 4633 + }, + { + "epoch": 0.5088952339117065, + "grad_norm": 2.3727333545684814, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.7031282186508179, + "num_tokens": 118129970.0, + "step": 4634 + }, + { + "epoch": 0.5090050516143202, + "grad_norm": 2.232581615447998, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7304456233978271, + "num_tokens": 118152877.0, + "step": 4635 + }, + { + "epoch": 0.5091148693169338, + "grad_norm": 2.310838460922241, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7061507701873779, + "num_tokens": 118176609.0, + "step": 4636 + }, + { + "epoch": 0.5092246870195476, + "grad_norm": 2.2190611362457275, + "learning_rate": 1e-06, + "loss": 1.0419, + "mean_token_accuracy": 0.6868364214897156, + "num_tokens": 118202736.0, + "step": 4637 + }, + { + "epoch": 0.5093345047221612, + "grad_norm": 2.0787243843078613, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7147732973098755, + "num_tokens": 118228449.0, + "step": 4638 + }, + { + "epoch": 0.5094443224247749, + "grad_norm": 1.9596086740493774, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6967388391494751, + "num_tokens": 118261439.0, + "step": 4639 + }, + { + "epoch": 0.5095541401273885, + "grad_norm": 2.2896456718444824, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6899672150611877, + "num_tokens": 118286537.0, + "step": 4640 + }, + { + "epoch": 0.5096639578300022, + "grad_norm": 2.278761386871338, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7036232948303223, + "num_tokens": 118311223.0, + "step": 4641 + }, + { + "epoch": 0.5097737755326158, + "grad_norm": 2.1418020725250244, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.7013943195343018, + "num_tokens": 118337968.0, + "step": 4642 + }, + { + "epoch": 0.5098835932352295, + "grad_norm": 1.9182889461517334, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7161091566085815, + "num_tokens": 118369077.0, + "step": 4643 + }, + { + "epoch": 0.5099934109378432, + "grad_norm": 2.4562594890594482, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7126160860061646, + "num_tokens": 118390173.0, + "step": 4644 + }, + { + "epoch": 0.5101032286404569, + "grad_norm": 1.9531188011169434, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7040430903434753, + "num_tokens": 118421604.0, + "step": 4645 + }, + { + "epoch": 0.5102130463430705, + "grad_norm": 2.5449650287628174, + "learning_rate": 1e-06, + "loss": 1.0551, + "mean_token_accuracy": 0.6833284497261047, + "num_tokens": 118443257.0, + "step": 4646 + }, + { + "epoch": 0.5103228640456842, + "grad_norm": 2.607633590698242, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6975107192993164, + "num_tokens": 118463869.0, + "step": 4647 + }, + { + "epoch": 0.5104326817482978, + "grad_norm": 1.9966590404510498, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6999070048332214, + "num_tokens": 118492377.0, + "step": 4648 + }, + { + "epoch": 0.5105424994509115, + "grad_norm": 2.2166202068328857, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6898723840713501, + "num_tokens": 118517923.0, + "step": 4649 + }, + { + "epoch": 0.5106523171535251, + "grad_norm": 2.072775363922119, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7040764093399048, + "num_tokens": 118544870.0, + "step": 4650 + }, + { + "epoch": 0.5107621348561389, + "grad_norm": 2.3097593784332275, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7304080128669739, + "num_tokens": 118566692.0, + "step": 4651 + }, + { + "epoch": 0.5108719525587525, + "grad_norm": 2.4523556232452393, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6923738718032837, + "num_tokens": 118587786.0, + "step": 4652 + }, + { + "epoch": 0.5109817702613662, + "grad_norm": 2.2233328819274902, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.7032862305641174, + "num_tokens": 118612432.0, + "step": 4653 + }, + { + "epoch": 0.5110915879639798, + "grad_norm": 2.108112096786499, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7152749300003052, + "num_tokens": 118638283.0, + "step": 4654 + }, + { + "epoch": 0.5112014056665934, + "grad_norm": 2.1119422912597656, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7223408818244934, + "num_tokens": 118663764.0, + "step": 4655 + }, + { + "epoch": 0.5113112233692071, + "grad_norm": 1.8654215335845947, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7282682657241821, + "num_tokens": 118693425.0, + "step": 4656 + }, + { + "epoch": 0.5114210410718207, + "grad_norm": 1.90770423412323, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6988502144813538, + "num_tokens": 118726027.0, + "step": 4657 + }, + { + "epoch": 0.5115308587744345, + "grad_norm": 2.1731934547424316, + "learning_rate": 1e-06, + "loss": 1.0323, + "mean_token_accuracy": 0.683962881565094, + "num_tokens": 118753379.0, + "step": 4658 + }, + { + "epoch": 0.5116406764770481, + "grad_norm": 2.6080710887908936, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7039193511009216, + "num_tokens": 118772643.0, + "step": 4659 + }, + { + "epoch": 0.5117504941796618, + "grad_norm": 2.0682876110076904, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6922503709793091, + "num_tokens": 118803082.0, + "step": 4660 + }, + { + "epoch": 0.5118603118822754, + "grad_norm": 1.969226360321045, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7098672389984131, + "num_tokens": 118832221.0, + "step": 4661 + }, + { + "epoch": 0.5119701295848891, + "grad_norm": 2.1142303943634033, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6939278244972229, + "num_tokens": 118860429.0, + "step": 4662 + }, + { + "epoch": 0.5120799472875027, + "grad_norm": 2.0034244060516357, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6998443007469177, + "num_tokens": 118891335.0, + "step": 4663 + }, + { + "epoch": 0.5121897649901164, + "grad_norm": 2.300483465194702, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7166321277618408, + "num_tokens": 118913968.0, + "step": 4664 + }, + { + "epoch": 0.51229958269273, + "grad_norm": 2.4306557178497314, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7077735066413879, + "num_tokens": 118936096.0, + "step": 4665 + }, + { + "epoch": 0.5124094003953438, + "grad_norm": 1.9833459854125977, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7017249464988708, + "num_tokens": 118967133.0, + "step": 4666 + }, + { + "epoch": 0.5125192180979574, + "grad_norm": 2.373988151550293, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.7084046602249146, + "num_tokens": 118991141.0, + "step": 4667 + }, + { + "epoch": 0.5126290358005711, + "grad_norm": 2.1063520908355713, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7162487506866455, + "num_tokens": 119017841.0, + "step": 4668 + }, + { + "epoch": 0.5127388535031847, + "grad_norm": 2.5598671436309814, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7352362871170044, + "num_tokens": 119035686.0, + "step": 4669 + }, + { + "epoch": 0.5128486712057984, + "grad_norm": 2.1602418422698975, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.709840714931488, + "num_tokens": 119062138.0, + "step": 4670 + }, + { + "epoch": 0.512958488908412, + "grad_norm": 2.2954647541046143, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7023722529411316, + "num_tokens": 119085160.0, + "step": 4671 + }, + { + "epoch": 0.5130683066110256, + "grad_norm": 2.5929009914398193, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.69354248046875, + "num_tokens": 119106137.0, + "step": 4672 + }, + { + "epoch": 0.5131781243136394, + "grad_norm": 2.38948130607605, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7073955535888672, + "num_tokens": 119129369.0, + "step": 4673 + }, + { + "epoch": 0.513287942016253, + "grad_norm": 2.1362602710723877, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6952906847000122, + "num_tokens": 119157034.0, + "step": 4674 + }, + { + "epoch": 0.5133977597188667, + "grad_norm": 2.4579479694366455, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7158544659614563, + "num_tokens": 119177726.0, + "step": 4675 + }, + { + "epoch": 0.5135075774214803, + "grad_norm": 2.431621789932251, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7089710235595703, + "num_tokens": 119198934.0, + "step": 4676 + }, + { + "epoch": 0.513617395124094, + "grad_norm": 1.9660353660583496, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7123665809631348, + "num_tokens": 119227992.0, + "step": 4677 + }, + { + "epoch": 0.5137272128267076, + "grad_norm": 2.298363208770752, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7303909659385681, + "num_tokens": 119251424.0, + "step": 4678 + }, + { + "epoch": 0.5138370305293213, + "grad_norm": 2.86274790763855, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7138506174087524, + "num_tokens": 119267534.0, + "step": 4679 + }, + { + "epoch": 0.513946848231935, + "grad_norm": 2.6631813049316406, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7128681540489197, + "num_tokens": 119286857.0, + "step": 4680 + }, + { + "epoch": 0.5140566659345487, + "grad_norm": 2.492001533508301, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7213019132614136, + "num_tokens": 119306020.0, + "step": 4681 + }, + { + "epoch": 0.5141664836371623, + "grad_norm": 2.420440196990967, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7109720706939697, + "num_tokens": 119326985.0, + "step": 4682 + }, + { + "epoch": 0.514276301339776, + "grad_norm": 2.0849051475524902, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.7069330215454102, + "num_tokens": 119354299.0, + "step": 4683 + }, + { + "epoch": 0.5143861190423896, + "grad_norm": 2.1584157943725586, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7163586616516113, + "num_tokens": 119380468.0, + "step": 4684 + }, + { + "epoch": 0.5144959367450033, + "grad_norm": 2.1563751697540283, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.728203296661377, + "num_tokens": 119406577.0, + "step": 4685 + }, + { + "epoch": 0.5146057544476169, + "grad_norm": 2.0665040016174316, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.7078772783279419, + "num_tokens": 119435201.0, + "step": 4686 + }, + { + "epoch": 0.5147155721502307, + "grad_norm": 2.3703324794769287, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.6922783255577087, + "num_tokens": 119457502.0, + "step": 4687 + }, + { + "epoch": 0.5148253898528443, + "grad_norm": 2.6589529514312744, + "learning_rate": 1e-06, + "loss": 1.0264, + "mean_token_accuracy": 0.6923333406448364, + "num_tokens": 119477519.0, + "step": 4688 + }, + { + "epoch": 0.514935207555458, + "grad_norm": 2.1850202083587646, + "learning_rate": 1e-06, + "loss": 1.0249, + "mean_token_accuracy": 0.6930384039878845, + "num_tokens": 119502802.0, + "step": 4689 + }, + { + "epoch": 0.5150450252580716, + "grad_norm": 2.1540908813476562, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7132636308670044, + "num_tokens": 119527961.0, + "step": 4690 + }, + { + "epoch": 0.5151548429606853, + "grad_norm": 2.068033456802368, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7116701602935791, + "num_tokens": 119555341.0, + "step": 4691 + }, + { + "epoch": 0.5152646606632989, + "grad_norm": 1.9881995916366577, + "learning_rate": 1e-06, + "loss": 1.062, + "mean_token_accuracy": 0.6745823621749878, + "num_tokens": 119585433.0, + "step": 4692 + }, + { + "epoch": 0.5153744783659125, + "grad_norm": 2.3792643547058105, + "learning_rate": 1e-06, + "loss": 1.0786, + "mean_token_accuracy": 0.6782779693603516, + "num_tokens": 119609632.0, + "step": 4693 + }, + { + "epoch": 0.5154842960685262, + "grad_norm": 2.527503728866577, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.72220778465271, + "num_tokens": 119628986.0, + "step": 4694 + }, + { + "epoch": 0.51559411377114, + "grad_norm": 2.3829848766326904, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.6942704319953918, + "num_tokens": 119652699.0, + "step": 4695 + }, + { + "epoch": 0.5157039314737536, + "grad_norm": 2.297757625579834, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7270911931991577, + "num_tokens": 119674773.0, + "step": 4696 + }, + { + "epoch": 0.5158137491763672, + "grad_norm": 2.204301595687866, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7211208343505859, + "num_tokens": 119699503.0, + "step": 4697 + }, + { + "epoch": 0.5159235668789809, + "grad_norm": 2.241769552230835, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7353514432907104, + "num_tokens": 119723495.0, + "step": 4698 + }, + { + "epoch": 0.5160333845815945, + "grad_norm": 2.0014772415161133, + "learning_rate": 1e-06, + "loss": 1.0177, + "mean_token_accuracy": 0.6897373199462891, + "num_tokens": 119756236.0, + "step": 4699 + }, + { + "epoch": 0.5161432022842082, + "grad_norm": 2.121649742126465, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7209939956665039, + "num_tokens": 119782548.0, + "step": 4700 + }, + { + "epoch": 0.5162530199868218, + "grad_norm": 2.1895205974578857, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7020401358604431, + "num_tokens": 119808279.0, + "step": 4701 + }, + { + "epoch": 0.5163628376894356, + "grad_norm": 2.1190223693847656, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7041743993759155, + "num_tokens": 119835656.0, + "step": 4702 + }, + { + "epoch": 0.5164726553920492, + "grad_norm": 2.165982723236084, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7180283665657043, + "num_tokens": 119861946.0, + "step": 4703 + }, + { + "epoch": 0.5165824730946629, + "grad_norm": 1.8402994871139526, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6974477767944336, + "num_tokens": 119895611.0, + "step": 4704 + }, + { + "epoch": 0.5166922907972765, + "grad_norm": 2.1920647621154785, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7176356315612793, + "num_tokens": 119918694.0, + "step": 4705 + }, + { + "epoch": 0.5168021084998902, + "grad_norm": 2.1232776641845703, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.710261344909668, + "num_tokens": 119944867.0, + "step": 4706 + }, + { + "epoch": 0.5169119262025038, + "grad_norm": 2.337246894836426, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7099357843399048, + "num_tokens": 119966757.0, + "step": 4707 + }, + { + "epoch": 0.5170217439051175, + "grad_norm": 2.636044502258301, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7154992818832397, + "num_tokens": 119988314.0, + "step": 4708 + }, + { + "epoch": 0.5171315616077312, + "grad_norm": 2.325866937637329, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7201343774795532, + "num_tokens": 120010005.0, + "step": 4709 + }, + { + "epoch": 0.5172413793103449, + "grad_norm": 2.447906494140625, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7262972593307495, + "num_tokens": 120029486.0, + "step": 4710 + }, + { + "epoch": 0.5173511970129585, + "grad_norm": 2.0878682136535645, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7230722904205322, + "num_tokens": 120056519.0, + "step": 4711 + }, + { + "epoch": 0.5174610147155722, + "grad_norm": 2.4239351749420166, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.7018924951553345, + "num_tokens": 120078324.0, + "step": 4712 + }, + { + "epoch": 0.5175708324181858, + "grad_norm": 2.407104730606079, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7172142267227173, + "num_tokens": 120098835.0, + "step": 4713 + }, + { + "epoch": 0.5176806501207994, + "grad_norm": 2.159186601638794, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.6931225061416626, + "num_tokens": 120123943.0, + "step": 4714 + }, + { + "epoch": 0.5177904678234131, + "grad_norm": 2.2600033283233643, + "learning_rate": 1e-06, + "loss": 1.006, + "mean_token_accuracy": 0.7007055282592773, + "num_tokens": 120147951.0, + "step": 4715 + }, + { + "epoch": 0.5179002855260268, + "grad_norm": 2.207048177719116, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.6986856460571289, + "num_tokens": 120173115.0, + "step": 4716 + }, + { + "epoch": 0.5180101032286405, + "grad_norm": 2.093801975250244, + "learning_rate": 1e-06, + "loss": 1.0351, + "mean_token_accuracy": 0.6908937692642212, + "num_tokens": 120201732.0, + "step": 4717 + }, + { + "epoch": 0.5181199209312541, + "grad_norm": 2.526719093322754, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7090517282485962, + "num_tokens": 120223037.0, + "step": 4718 + }, + { + "epoch": 0.5182297386338678, + "grad_norm": 2.0280776023864746, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.7022351026535034, + "num_tokens": 120253697.0, + "step": 4719 + }, + { + "epoch": 0.5183395563364814, + "grad_norm": 2.081122636795044, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.7045568823814392, + "num_tokens": 120280438.0, + "step": 4720 + }, + { + "epoch": 0.5184493740390951, + "grad_norm": 1.836210012435913, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.7040141224861145, + "num_tokens": 120311675.0, + "step": 4721 + }, + { + "epoch": 0.5185591917417087, + "grad_norm": 2.0886030197143555, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7224456071853638, + "num_tokens": 120340129.0, + "step": 4722 + }, + { + "epoch": 0.5186690094443224, + "grad_norm": 2.1332449913024902, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.7014533281326294, + "num_tokens": 120366868.0, + "step": 4723 + }, + { + "epoch": 0.5187788271469361, + "grad_norm": 2.1389663219451904, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.6954555511474609, + "num_tokens": 120392923.0, + "step": 4724 + }, + { + "epoch": 0.5188886448495498, + "grad_norm": 2.050917863845825, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6907758116722107, + "num_tokens": 120422573.0, + "step": 4725 + }, + { + "epoch": 0.5189984625521634, + "grad_norm": 2.3620786666870117, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.702439546585083, + "num_tokens": 120445175.0, + "step": 4726 + }, + { + "epoch": 0.5191082802547771, + "grad_norm": 2.447969913482666, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7195005416870117, + "num_tokens": 120465770.0, + "step": 4727 + }, + { + "epoch": 0.5192180979573907, + "grad_norm": 2.1213085651397705, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6968222260475159, + "num_tokens": 120494655.0, + "step": 4728 + }, + { + "epoch": 0.5193279156600044, + "grad_norm": 2.2546279430389404, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6948932409286499, + "num_tokens": 120519926.0, + "step": 4729 + }, + { + "epoch": 0.519437733362618, + "grad_norm": 2.3899126052856445, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7174327373504639, + "num_tokens": 120540992.0, + "step": 4730 + }, + { + "epoch": 0.5195475510652318, + "grad_norm": 2.1997315883636475, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7301488518714905, + "num_tokens": 120565395.0, + "step": 4731 + }, + { + "epoch": 0.5196573687678454, + "grad_norm": 2.1669695377349854, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7140824794769287, + "num_tokens": 120592274.0, + "step": 4732 + }, + { + "epoch": 0.519767186470459, + "grad_norm": 2.0674643516540527, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6945043206214905, + "num_tokens": 120618509.0, + "step": 4733 + }, + { + "epoch": 0.5198770041730727, + "grad_norm": 2.1831071376800537, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7271232008934021, + "num_tokens": 120644951.0, + "step": 4734 + }, + { + "epoch": 0.5199868218756863, + "grad_norm": 2.313279628753662, + "learning_rate": 1e-06, + "loss": 1.1048, + "mean_token_accuracy": 0.670077383518219, + "num_tokens": 120667940.0, + "step": 4735 + }, + { + "epoch": 0.5200966395783, + "grad_norm": 2.1104373931884766, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.688718318939209, + "num_tokens": 120696680.0, + "step": 4736 + }, + { + "epoch": 0.5202064572809136, + "grad_norm": 2.2292895317077637, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7175118923187256, + "num_tokens": 120720342.0, + "step": 4737 + }, + { + "epoch": 0.5203162749835274, + "grad_norm": 2.125347375869751, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7155573964118958, + "num_tokens": 120746411.0, + "step": 4738 + }, + { + "epoch": 0.520426092686141, + "grad_norm": 2.1550633907318115, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7001461386680603, + "num_tokens": 120773197.0, + "step": 4739 + }, + { + "epoch": 0.5205359103887547, + "grad_norm": 2.3106744289398193, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7028714418411255, + "num_tokens": 120797844.0, + "step": 4740 + }, + { + "epoch": 0.5206457280913683, + "grad_norm": 2.3119544982910156, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7189760804176331, + "num_tokens": 120823282.0, + "step": 4741 + }, + { + "epoch": 0.520755545793982, + "grad_norm": 2.2709922790527344, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7125377655029297, + "num_tokens": 120845872.0, + "step": 4742 + }, + { + "epoch": 0.5208653634965956, + "grad_norm": 2.023078203201294, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7174602746963501, + "num_tokens": 120876426.0, + "step": 4743 + }, + { + "epoch": 0.5209751811992093, + "grad_norm": 2.32686710357666, + "learning_rate": 1e-06, + "loss": 1.0892, + "mean_token_accuracy": 0.675873875617981, + "num_tokens": 120901072.0, + "step": 4744 + }, + { + "epoch": 0.521084998901823, + "grad_norm": 2.155869722366333, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7124155759811401, + "num_tokens": 120926506.0, + "step": 4745 + }, + { + "epoch": 0.5211948166044367, + "grad_norm": 2.265249013900757, + "learning_rate": 1e-06, + "loss": 1.0687, + "mean_token_accuracy": 0.6871939301490784, + "num_tokens": 120951017.0, + "step": 4746 + }, + { + "epoch": 0.5213046343070503, + "grad_norm": 2.2750253677368164, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7173727750778198, + "num_tokens": 120974117.0, + "step": 4747 + }, + { + "epoch": 0.521414452009664, + "grad_norm": 2.350687026977539, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7002809643745422, + "num_tokens": 120996805.0, + "step": 4748 + }, + { + "epoch": 0.5215242697122776, + "grad_norm": 1.8718994855880737, + "learning_rate": 1e-06, + "loss": 1.0303, + "mean_token_accuracy": 0.6910308599472046, + "num_tokens": 121031498.0, + "step": 4749 + }, + { + "epoch": 0.5216340874148913, + "grad_norm": 2.4378669261932373, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7065224647521973, + "num_tokens": 121051435.0, + "step": 4750 + }, + { + "epoch": 0.5217439051175049, + "grad_norm": 2.292771577835083, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.710923433303833, + "num_tokens": 121074366.0, + "step": 4751 + }, + { + "epoch": 0.5218537228201185, + "grad_norm": 2.2976176738739014, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7193260192871094, + "num_tokens": 121098341.0, + "step": 4752 + }, + { + "epoch": 0.5219635405227323, + "grad_norm": 2.145777940750122, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6895856857299805, + "num_tokens": 121126561.0, + "step": 4753 + }, + { + "epoch": 0.522073358225346, + "grad_norm": 2.027438163757324, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7044007778167725, + "num_tokens": 121156779.0, + "step": 4754 + }, + { + "epoch": 0.5221831759279596, + "grad_norm": 1.9891088008880615, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6892051696777344, + "num_tokens": 121189393.0, + "step": 4755 + }, + { + "epoch": 0.5222929936305732, + "grad_norm": 2.234368085861206, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7013421058654785, + "num_tokens": 121214184.0, + "step": 4756 + }, + { + "epoch": 0.5224028113331869, + "grad_norm": 2.088864803314209, + "learning_rate": 1e-06, + "loss": 1.0835, + "mean_token_accuracy": 0.6735910177230835, + "num_tokens": 121244155.0, + "step": 4757 + }, + { + "epoch": 0.5225126290358005, + "grad_norm": 2.077972173690796, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.6906759142875671, + "num_tokens": 121271084.0, + "step": 4758 + }, + { + "epoch": 0.5226224467384142, + "grad_norm": 1.9478992223739624, + "learning_rate": 1e-06, + "loss": 1.0725, + "mean_token_accuracy": 0.6808108687400818, + "num_tokens": 121305519.0, + "step": 4759 + }, + { + "epoch": 0.5227322644410279, + "grad_norm": 2.092777967453003, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7279422283172607, + "num_tokens": 121331322.0, + "step": 4760 + }, + { + "epoch": 0.5228420821436416, + "grad_norm": 2.0409340858459473, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7112853527069092, + "num_tokens": 121359803.0, + "step": 4761 + }, + { + "epoch": 0.5229518998462552, + "grad_norm": 2.2318849563598633, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7192428112030029, + "num_tokens": 121384127.0, + "step": 4762 + }, + { + "epoch": 0.5230617175488689, + "grad_norm": 2.1895902156829834, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.731816291809082, + "num_tokens": 121407223.0, + "step": 4763 + }, + { + "epoch": 0.5231715352514825, + "grad_norm": 2.3871207237243652, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7149806022644043, + "num_tokens": 121429172.0, + "step": 4764 + }, + { + "epoch": 0.5232813529540962, + "grad_norm": 2.3660552501678467, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7105344533920288, + "num_tokens": 121449759.0, + "step": 4765 + }, + { + "epoch": 0.5233911706567098, + "grad_norm": 2.216367244720459, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6991084218025208, + "num_tokens": 121475941.0, + "step": 4766 + }, + { + "epoch": 0.5235009883593236, + "grad_norm": 2.0984251499176025, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.6998696327209473, + "num_tokens": 121504021.0, + "step": 4767 + }, + { + "epoch": 0.5236108060619372, + "grad_norm": 2.528589963912964, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7019796967506409, + "num_tokens": 121524045.0, + "step": 4768 + }, + { + "epoch": 0.5237206237645509, + "grad_norm": 2.2167139053344727, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7057431936264038, + "num_tokens": 121550958.0, + "step": 4769 + }, + { + "epoch": 0.5238304414671645, + "grad_norm": 2.007664680480957, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.710976779460907, + "num_tokens": 121580633.0, + "step": 4770 + }, + { + "epoch": 0.5239402591697782, + "grad_norm": 2.237966775894165, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7349188327789307, + "num_tokens": 121603817.0, + "step": 4771 + }, + { + "epoch": 0.5240500768723918, + "grad_norm": 1.9052635431289673, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7266033887863159, + "num_tokens": 121634699.0, + "step": 4772 + }, + { + "epoch": 0.5241598945750054, + "grad_norm": 2.2076661586761475, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7213093042373657, + "num_tokens": 121656693.0, + "step": 4773 + }, + { + "epoch": 0.5242697122776192, + "grad_norm": 2.148542642593384, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7081013321876526, + "num_tokens": 121681836.0, + "step": 4774 + }, + { + "epoch": 0.5243795299802329, + "grad_norm": 2.1670992374420166, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.6965532302856445, + "num_tokens": 121708808.0, + "step": 4775 + }, + { + "epoch": 0.5244893476828465, + "grad_norm": 2.0333213806152344, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.6935602426528931, + "num_tokens": 121735991.0, + "step": 4776 + }, + { + "epoch": 0.5245991653854601, + "grad_norm": 2.3292529582977295, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7342357039451599, + "num_tokens": 121757516.0, + "step": 4777 + }, + { + "epoch": 0.5247089830880738, + "grad_norm": 2.188948631286621, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7399516701698303, + "num_tokens": 121781559.0, + "step": 4778 + }, + { + "epoch": 0.5248188007906874, + "grad_norm": 2.7553517818450928, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6930269598960876, + "num_tokens": 121800746.0, + "step": 4779 + }, + { + "epoch": 0.5249286184933011, + "grad_norm": 2.1132194995880127, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.7068026065826416, + "num_tokens": 121828327.0, + "step": 4780 + }, + { + "epoch": 0.5250384361959147, + "grad_norm": 2.1576874256134033, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6939431428909302, + "num_tokens": 121856140.0, + "step": 4781 + }, + { + "epoch": 0.5251482538985285, + "grad_norm": 2.2954161167144775, + "learning_rate": 1e-06, + "loss": 1.0345, + "mean_token_accuracy": 0.7001583576202393, + "num_tokens": 121880163.0, + "step": 4782 + }, + { + "epoch": 0.5252580716011421, + "grad_norm": 2.2103192806243896, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6957765817642212, + "num_tokens": 121907886.0, + "step": 4783 + }, + { + "epoch": 0.5253678893037558, + "grad_norm": 2.0991899967193604, + "learning_rate": 1e-06, + "loss": 1.0725, + "mean_token_accuracy": 0.6803179979324341, + "num_tokens": 121939824.0, + "step": 4784 + }, + { + "epoch": 0.5254777070063694, + "grad_norm": 2.3929693698883057, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7211973667144775, + "num_tokens": 121959551.0, + "step": 4785 + }, + { + "epoch": 0.5255875247089831, + "grad_norm": 2.317838430404663, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6951054334640503, + "num_tokens": 121983217.0, + "step": 4786 + }, + { + "epoch": 0.5256973424115967, + "grad_norm": 2.2662835121154785, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.6949833035469055, + "num_tokens": 122005712.0, + "step": 4787 + }, + { + "epoch": 0.5258071601142104, + "grad_norm": 2.378622531890869, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7088965177536011, + "num_tokens": 122027311.0, + "step": 4788 + }, + { + "epoch": 0.5259169778168241, + "grad_norm": 2.223829507827759, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7247472405433655, + "num_tokens": 122050021.0, + "step": 4789 + }, + { + "epoch": 0.5260267955194378, + "grad_norm": 2.3731343746185303, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.703175961971283, + "num_tokens": 122071780.0, + "step": 4790 + }, + { + "epoch": 0.5261366132220514, + "grad_norm": 2.326929807662964, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7043936848640442, + "num_tokens": 122096737.0, + "step": 4791 + }, + { + "epoch": 0.526246430924665, + "grad_norm": 2.3488218784332275, + "learning_rate": 1e-06, + "loss": 1.0196, + "mean_token_accuracy": 0.6962509751319885, + "num_tokens": 122121438.0, + "step": 4792 + }, + { + "epoch": 0.5263562486272787, + "grad_norm": 2.3452250957489014, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7080366611480713, + "num_tokens": 122143831.0, + "step": 4793 + }, + { + "epoch": 0.5264660663298923, + "grad_norm": 2.1835973262786865, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6983712911605835, + "num_tokens": 122169295.0, + "step": 4794 + }, + { + "epoch": 0.526575884032506, + "grad_norm": 2.0615885257720947, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7000964879989624, + "num_tokens": 122196960.0, + "step": 4795 + }, + { + "epoch": 0.5266857017351197, + "grad_norm": 2.176663875579834, + "learning_rate": 1e-06, + "loss": 1.0561, + "mean_token_accuracy": 0.6775420904159546, + "num_tokens": 122222445.0, + "step": 4796 + }, + { + "epoch": 0.5267955194377334, + "grad_norm": 2.178410530090332, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7180649042129517, + "num_tokens": 122246055.0, + "step": 4797 + }, + { + "epoch": 0.526905337140347, + "grad_norm": 2.2797341346740723, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7173473238945007, + "num_tokens": 122268403.0, + "step": 4798 + }, + { + "epoch": 0.5270151548429607, + "grad_norm": 2.2261319160461426, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6999633312225342, + "num_tokens": 122295408.0, + "step": 4799 + }, + { + "epoch": 0.5271249725455743, + "grad_norm": 2.4063103199005127, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.6996520757675171, + "num_tokens": 122317858.0, + "step": 4800 + }, + { + "epoch": 0.527234790248188, + "grad_norm": 2.094658613204956, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6938191652297974, + "num_tokens": 122342971.0, + "step": 4801 + }, + { + "epoch": 0.5273446079508016, + "grad_norm": 2.418470621109009, + "learning_rate": 1e-06, + "loss": 1.0189, + "mean_token_accuracy": 0.7007572054862976, + "num_tokens": 122366167.0, + "step": 4802 + }, + { + "epoch": 0.5274544256534154, + "grad_norm": 2.3353166580200195, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6959990859031677, + "num_tokens": 122389809.0, + "step": 4803 + }, + { + "epoch": 0.527564243356029, + "grad_norm": 2.1294989585876465, + "learning_rate": 1e-06, + "loss": 1.0552, + "mean_token_accuracy": 0.677675724029541, + "num_tokens": 122416786.0, + "step": 4804 + }, + { + "epoch": 0.5276740610586427, + "grad_norm": 2.145293712615967, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7165063619613647, + "num_tokens": 122442593.0, + "step": 4805 + }, + { + "epoch": 0.5277838787612563, + "grad_norm": 2.290956974029541, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7181800603866577, + "num_tokens": 122466669.0, + "step": 4806 + }, + { + "epoch": 0.52789369646387, + "grad_norm": 2.1801917552948, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7098134160041809, + "num_tokens": 122496554.0, + "step": 4807 + }, + { + "epoch": 0.5280035141664836, + "grad_norm": 2.37516713142395, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7005250453948975, + "num_tokens": 122518522.0, + "step": 4808 + }, + { + "epoch": 0.5281133318690973, + "grad_norm": 1.978040337562561, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6979520320892334, + "num_tokens": 122547696.0, + "step": 4809 + }, + { + "epoch": 0.528223149571711, + "grad_norm": 2.104513645172119, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7072571516036987, + "num_tokens": 122576057.0, + "step": 4810 + }, + { + "epoch": 0.5283329672743247, + "grad_norm": 2.175158977508545, + "learning_rate": 1e-06, + "loss": 1.0576, + "mean_token_accuracy": 0.6902841329574585, + "num_tokens": 122602608.0, + "step": 4811 + }, + { + "epoch": 0.5284427849769383, + "grad_norm": 2.2848281860351562, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7130946516990662, + "num_tokens": 122627641.0, + "step": 4812 + }, + { + "epoch": 0.528552602679552, + "grad_norm": 2.177640676498413, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7023622989654541, + "num_tokens": 122654835.0, + "step": 4813 + }, + { + "epoch": 0.5286624203821656, + "grad_norm": 2.4445106983184814, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7088170051574707, + "num_tokens": 122675944.0, + "step": 4814 + }, + { + "epoch": 0.5287722380847792, + "grad_norm": 1.9993313550949097, + "learning_rate": 1e-06, + "loss": 1.0832, + "mean_token_accuracy": 0.6821466088294983, + "num_tokens": 122705056.0, + "step": 4815 + }, + { + "epoch": 0.5288820557873929, + "grad_norm": 2.1919689178466797, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6937472224235535, + "num_tokens": 122731516.0, + "step": 4816 + }, + { + "epoch": 0.5289918734900065, + "grad_norm": 2.2862513065338135, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.705929696559906, + "num_tokens": 122753731.0, + "step": 4817 + }, + { + "epoch": 0.5291016911926203, + "grad_norm": 1.893104910850525, + "learning_rate": 1e-06, + "loss": 1.0714, + "mean_token_accuracy": 0.6771621108055115, + "num_tokens": 122786224.0, + "step": 4818 + }, + { + "epoch": 0.5292115088952339, + "grad_norm": 2.225517749786377, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.696209192276001, + "num_tokens": 122813778.0, + "step": 4819 + }, + { + "epoch": 0.5293213265978476, + "grad_norm": 2.2810473442077637, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7030702829360962, + "num_tokens": 122839470.0, + "step": 4820 + }, + { + "epoch": 0.5294311443004612, + "grad_norm": 2.1318557262420654, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6932333707809448, + "num_tokens": 122867059.0, + "step": 4821 + }, + { + "epoch": 0.5295409620030749, + "grad_norm": 1.9688204526901245, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6975034475326538, + "num_tokens": 122897177.0, + "step": 4822 + }, + { + "epoch": 0.5296507797056885, + "grad_norm": 2.1062655448913574, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.7018927335739136, + "num_tokens": 122925631.0, + "step": 4823 + }, + { + "epoch": 0.5297605974083022, + "grad_norm": 2.14416766166687, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7145500779151917, + "num_tokens": 122952119.0, + "step": 4824 + }, + { + "epoch": 0.5298704151109159, + "grad_norm": 2.1988229751586914, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7050731182098389, + "num_tokens": 122978510.0, + "step": 4825 + }, + { + "epoch": 0.5299802328135296, + "grad_norm": 2.2382123470306396, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7133249640464783, + "num_tokens": 123002481.0, + "step": 4826 + }, + { + "epoch": 0.5300900505161432, + "grad_norm": 2.3913803100585938, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7167438864707947, + "num_tokens": 123025956.0, + "step": 4827 + }, + { + "epoch": 0.5301998682187569, + "grad_norm": 2.1519722938537598, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.7010815143585205, + "num_tokens": 123050993.0, + "step": 4828 + }, + { + "epoch": 0.5303096859213705, + "grad_norm": 2.292402982711792, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7163407206535339, + "num_tokens": 123072644.0, + "step": 4829 + }, + { + "epoch": 0.5304195036239842, + "grad_norm": 2.075568675994873, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.700308084487915, + "num_tokens": 123103264.0, + "step": 4830 + }, + { + "epoch": 0.5305293213265978, + "grad_norm": 2.0808868408203125, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7251404523849487, + "num_tokens": 123130215.0, + "step": 4831 + }, + { + "epoch": 0.5306391390292116, + "grad_norm": 2.184863805770874, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6969066858291626, + "num_tokens": 123155098.0, + "step": 4832 + }, + { + "epoch": 0.5307489567318252, + "grad_norm": 2.583077907562256, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.6982821226119995, + "num_tokens": 123175722.0, + "step": 4833 + }, + { + "epoch": 0.5308587744344389, + "grad_norm": 2.3245153427124023, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7378958463668823, + "num_tokens": 123198443.0, + "step": 4834 + }, + { + "epoch": 0.5309685921370525, + "grad_norm": 2.3271048069000244, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.698013186454773, + "num_tokens": 123224206.0, + "step": 4835 + }, + { + "epoch": 0.5310784098396661, + "grad_norm": 2.3563411235809326, + "learning_rate": 1e-06, + "loss": 1.0569, + "mean_token_accuracy": 0.694357693195343, + "num_tokens": 123247759.0, + "step": 4836 + }, + { + "epoch": 0.5311882275422798, + "grad_norm": 2.2129998207092285, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7199727892875671, + "num_tokens": 123272620.0, + "step": 4837 + }, + { + "epoch": 0.5312980452448934, + "grad_norm": 2.1633591651916504, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7131489515304565, + "num_tokens": 123298246.0, + "step": 4838 + }, + { + "epoch": 0.5314078629475072, + "grad_norm": 2.190430164337158, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6922999620437622, + "num_tokens": 123324393.0, + "step": 4839 + }, + { + "epoch": 0.5315176806501208, + "grad_norm": 2.3857457637786865, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7163646221160889, + "num_tokens": 123344370.0, + "step": 4840 + }, + { + "epoch": 0.5316274983527345, + "grad_norm": 1.9819732904434204, + "learning_rate": 1e-06, + "loss": 1.0184, + "mean_token_accuracy": 0.6818690896034241, + "num_tokens": 123374660.0, + "step": 4841 + }, + { + "epoch": 0.5317373160553481, + "grad_norm": 2.301522970199585, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.702224612236023, + "num_tokens": 123397668.0, + "step": 4842 + }, + { + "epoch": 0.5318471337579618, + "grad_norm": 2.1152474880218506, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7212107181549072, + "num_tokens": 123421593.0, + "step": 4843 + }, + { + "epoch": 0.5319569514605754, + "grad_norm": 2.4655988216400146, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.7083615064620972, + "num_tokens": 123441591.0, + "step": 4844 + }, + { + "epoch": 0.5320667691631891, + "grad_norm": 2.246528148651123, + "learning_rate": 1e-06, + "loss": 1.0519, + "mean_token_accuracy": 0.6792389750480652, + "num_tokens": 123467230.0, + "step": 4845 + }, + { + "epoch": 0.5321765868658027, + "grad_norm": 2.2979795932769775, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.6938250064849854, + "num_tokens": 123490751.0, + "step": 4846 + }, + { + "epoch": 0.5322864045684165, + "grad_norm": 2.1908504962921143, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.7076128721237183, + "num_tokens": 123513889.0, + "step": 4847 + }, + { + "epoch": 0.5323962222710301, + "grad_norm": 1.9192605018615723, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.6996933817863464, + "num_tokens": 123543920.0, + "step": 4848 + }, + { + "epoch": 0.5325060399736438, + "grad_norm": 2.2694709300994873, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.725990355014801, + "num_tokens": 123567226.0, + "step": 4849 + }, + { + "epoch": 0.5326158576762574, + "grad_norm": 2.2295031547546387, + "learning_rate": 1e-06, + "loss": 1.0552, + "mean_token_accuracy": 0.6821856498718262, + "num_tokens": 123592948.0, + "step": 4850 + }, + { + "epoch": 0.532725675378871, + "grad_norm": 2.2283384799957275, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.7043830752372742, + "num_tokens": 123619265.0, + "step": 4851 + }, + { + "epoch": 0.5328354930814847, + "grad_norm": 2.300978183746338, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.7017660140991211, + "num_tokens": 123641106.0, + "step": 4852 + }, + { + "epoch": 0.5329453107840983, + "grad_norm": 2.138544797897339, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6822177767753601, + "num_tokens": 123670084.0, + "step": 4853 + }, + { + "epoch": 0.5330551284867121, + "grad_norm": 2.679965019226074, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7024624943733215, + "num_tokens": 123689392.0, + "step": 4854 + }, + { + "epoch": 0.5331649461893258, + "grad_norm": 2.117504358291626, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.7000952959060669, + "num_tokens": 123718047.0, + "step": 4855 + }, + { + "epoch": 0.5332747638919394, + "grad_norm": 2.0528321266174316, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7052451968193054, + "num_tokens": 123745018.0, + "step": 4856 + }, + { + "epoch": 0.533384581594553, + "grad_norm": 2.109995126724243, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7069042325019836, + "num_tokens": 123771264.0, + "step": 4857 + }, + { + "epoch": 0.5334943992971667, + "grad_norm": 2.2542927265167236, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7202234864234924, + "num_tokens": 123796082.0, + "step": 4858 + }, + { + "epoch": 0.5336042169997803, + "grad_norm": 1.9962819814682007, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7021551132202148, + "num_tokens": 123823835.0, + "step": 4859 + }, + { + "epoch": 0.533714034702394, + "grad_norm": 2.015770673751831, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6992647647857666, + "num_tokens": 123854564.0, + "step": 4860 + }, + { + "epoch": 0.5338238524050077, + "grad_norm": 2.058441162109375, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6961377859115601, + "num_tokens": 123883576.0, + "step": 4861 + }, + { + "epoch": 0.5339336701076214, + "grad_norm": 2.211998224258423, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7055535912513733, + "num_tokens": 123909155.0, + "step": 4862 + }, + { + "epoch": 0.534043487810235, + "grad_norm": 1.8691892623901367, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6956338882446289, + "num_tokens": 123942725.0, + "step": 4863 + }, + { + "epoch": 0.5341533055128487, + "grad_norm": 2.028428554534912, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6999814510345459, + "num_tokens": 123969572.0, + "step": 4864 + }, + { + "epoch": 0.5342631232154623, + "grad_norm": 2.115710973739624, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6981335282325745, + "num_tokens": 123997035.0, + "step": 4865 + }, + { + "epoch": 0.534372940918076, + "grad_norm": 2.549241542816162, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7078959941864014, + "num_tokens": 124017928.0, + "step": 4866 + }, + { + "epoch": 0.5344827586206896, + "grad_norm": 2.4921517372131348, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7060941457748413, + "num_tokens": 124041519.0, + "step": 4867 + }, + { + "epoch": 0.5345925763233034, + "grad_norm": 2.4857017993927, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7087060213088989, + "num_tokens": 124062882.0, + "step": 4868 + }, + { + "epoch": 0.534702394025917, + "grad_norm": 2.0930967330932617, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7214410305023193, + "num_tokens": 124088181.0, + "step": 4869 + }, + { + "epoch": 0.5348122117285307, + "grad_norm": 2.359910249710083, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.6977263689041138, + "num_tokens": 124111580.0, + "step": 4870 + }, + { + "epoch": 0.5349220294311443, + "grad_norm": 2.2712836265563965, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6937793493270874, + "num_tokens": 124138322.0, + "step": 4871 + }, + { + "epoch": 0.535031847133758, + "grad_norm": 2.192193031311035, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7221524715423584, + "num_tokens": 124160555.0, + "step": 4872 + }, + { + "epoch": 0.5351416648363716, + "grad_norm": 2.1303281784057617, + "learning_rate": 1e-06, + "loss": 1.1041, + "mean_token_accuracy": 0.6762128472328186, + "num_tokens": 124191329.0, + "step": 4873 + }, + { + "epoch": 0.5352514825389852, + "grad_norm": 2.149766445159912, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7071877717971802, + "num_tokens": 124216589.0, + "step": 4874 + }, + { + "epoch": 0.5353613002415989, + "grad_norm": 2.7037501335144043, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7039520144462585, + "num_tokens": 124236032.0, + "step": 4875 + }, + { + "epoch": 0.5354711179442126, + "grad_norm": 2.731093645095825, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7146904468536377, + "num_tokens": 124254813.0, + "step": 4876 + }, + { + "epoch": 0.5355809356468263, + "grad_norm": 2.336561918258667, + "learning_rate": 1e-06, + "loss": 1.048, + "mean_token_accuracy": 0.6838122606277466, + "num_tokens": 124278390.0, + "step": 4877 + }, + { + "epoch": 0.5356907533494399, + "grad_norm": 2.100640058517456, + "learning_rate": 1e-06, + "loss": 1.0899, + "mean_token_accuracy": 0.6740058660507202, + "num_tokens": 124310563.0, + "step": 4878 + }, + { + "epoch": 0.5358005710520536, + "grad_norm": 2.053307294845581, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6952041387557983, + "num_tokens": 124338411.0, + "step": 4879 + }, + { + "epoch": 0.5359103887546672, + "grad_norm": 2.215829372406006, + "learning_rate": 1e-06, + "loss": 1.0887, + "mean_token_accuracy": 0.6794157028198242, + "num_tokens": 124363436.0, + "step": 4880 + }, + { + "epoch": 0.5360202064572809, + "grad_norm": 2.066236734390259, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7093509435653687, + "num_tokens": 124389619.0, + "step": 4881 + }, + { + "epoch": 0.5361300241598945, + "grad_norm": 2.384859800338745, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7039639949798584, + "num_tokens": 124410704.0, + "step": 4882 + }, + { + "epoch": 0.5362398418625083, + "grad_norm": 2.332087755203247, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7251459360122681, + "num_tokens": 124433386.0, + "step": 4883 + }, + { + "epoch": 0.5363496595651219, + "grad_norm": 2.4163036346435547, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7217142581939697, + "num_tokens": 124453785.0, + "step": 4884 + }, + { + "epoch": 0.5364594772677356, + "grad_norm": 2.387122869491577, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7060884833335876, + "num_tokens": 124476703.0, + "step": 4885 + }, + { + "epoch": 0.5365692949703492, + "grad_norm": 2.150226354598999, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.703697681427002, + "num_tokens": 124501131.0, + "step": 4886 + }, + { + "epoch": 0.5366791126729629, + "grad_norm": 2.3205082416534424, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.7060494422912598, + "num_tokens": 124525002.0, + "step": 4887 + }, + { + "epoch": 0.5367889303755765, + "grad_norm": 2.0417795181274414, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.7115451097488403, + "num_tokens": 124556001.0, + "step": 4888 + }, + { + "epoch": 0.5368987480781902, + "grad_norm": 2.356259346008301, + "learning_rate": 1e-06, + "loss": 1.0582, + "mean_token_accuracy": 0.6829901933670044, + "num_tokens": 124581476.0, + "step": 4889 + }, + { + "epoch": 0.5370085657808039, + "grad_norm": 1.9267568588256836, + "learning_rate": 1e-06, + "loss": 1.0239, + "mean_token_accuracy": 0.6887027025222778, + "num_tokens": 124613923.0, + "step": 4890 + }, + { + "epoch": 0.5371183834834176, + "grad_norm": 2.42020845413208, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.703250527381897, + "num_tokens": 124637172.0, + "step": 4891 + }, + { + "epoch": 0.5372282011860312, + "grad_norm": 2.393782377243042, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7153230905532837, + "num_tokens": 124660478.0, + "step": 4892 + }, + { + "epoch": 0.5373380188886449, + "grad_norm": 1.9092130661010742, + "learning_rate": 1e-06, + "loss": 1.0443, + "mean_token_accuracy": 0.6845622062683105, + "num_tokens": 124693341.0, + "step": 4893 + }, + { + "epoch": 0.5374478365912585, + "grad_norm": 2.2522993087768555, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7129228115081787, + "num_tokens": 124716407.0, + "step": 4894 + }, + { + "epoch": 0.5375576542938721, + "grad_norm": 2.125953197479248, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6902252435684204, + "num_tokens": 124743410.0, + "step": 4895 + }, + { + "epoch": 0.5376674719964858, + "grad_norm": 2.3398749828338623, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7133573293685913, + "num_tokens": 124764841.0, + "step": 4896 + }, + { + "epoch": 0.5377772896990995, + "grad_norm": 2.2092740535736084, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.698577344417572, + "num_tokens": 124793997.0, + "step": 4897 + }, + { + "epoch": 0.5378871074017132, + "grad_norm": 2.3427507877349854, + "learning_rate": 1e-06, + "loss": 1.0425, + "mean_token_accuracy": 0.6860019564628601, + "num_tokens": 124818420.0, + "step": 4898 + }, + { + "epoch": 0.5379969251043268, + "grad_norm": 2.5069217681884766, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7019644975662231, + "num_tokens": 124840043.0, + "step": 4899 + }, + { + "epoch": 0.5381067428069405, + "grad_norm": 2.251725435256958, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7082117199897766, + "num_tokens": 124864865.0, + "step": 4900 + }, + { + "epoch": 0.5382165605095541, + "grad_norm": 2.2835450172424316, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.693055272102356, + "num_tokens": 124891419.0, + "step": 4901 + }, + { + "epoch": 0.5383263782121678, + "grad_norm": 2.1707751750946045, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.6916261315345764, + "num_tokens": 124916697.0, + "step": 4902 + }, + { + "epoch": 0.5384361959147814, + "grad_norm": 2.0110514163970947, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6930548548698425, + "num_tokens": 124947514.0, + "step": 4903 + }, + { + "epoch": 0.5385460136173951, + "grad_norm": 2.1420340538024902, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7031623125076294, + "num_tokens": 124976619.0, + "step": 4904 + }, + { + "epoch": 0.5386558313200088, + "grad_norm": 2.315788507461548, + "learning_rate": 1e-06, + "loss": 1.0887, + "mean_token_accuracy": 0.6771318912506104, + "num_tokens": 125001258.0, + "step": 4905 + }, + { + "epoch": 0.5387656490226225, + "grad_norm": 2.4357151985168457, + "learning_rate": 1e-06, + "loss": 1.0912, + "mean_token_accuracy": 0.6774531006813049, + "num_tokens": 125024017.0, + "step": 4906 + }, + { + "epoch": 0.5388754667252361, + "grad_norm": 2.3990025520324707, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7232621908187866, + "num_tokens": 125045473.0, + "step": 4907 + }, + { + "epoch": 0.5389852844278498, + "grad_norm": 2.1890430450439453, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7268513441085815, + "num_tokens": 125069771.0, + "step": 4908 + }, + { + "epoch": 0.5390951021304634, + "grad_norm": 1.9616892337799072, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7119478583335876, + "num_tokens": 125099530.0, + "step": 4909 + }, + { + "epoch": 0.5392049198330771, + "grad_norm": 2.2003564834594727, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7197633981704712, + "num_tokens": 125126512.0, + "step": 4910 + }, + { + "epoch": 0.5393147375356907, + "grad_norm": 2.260125160217285, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.7055961489677429, + "num_tokens": 125149625.0, + "step": 4911 + }, + { + "epoch": 0.5394245552383045, + "grad_norm": 2.468374490737915, + "learning_rate": 1e-06, + "loss": 1.0108, + "mean_token_accuracy": 0.6931841373443604, + "num_tokens": 125171188.0, + "step": 4912 + }, + { + "epoch": 0.5395343729409181, + "grad_norm": 2.2123520374298096, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.697679340839386, + "num_tokens": 125196893.0, + "step": 4913 + }, + { + "epoch": 0.5396441906435318, + "grad_norm": 2.1654040813446045, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.7001407146453857, + "num_tokens": 125223440.0, + "step": 4914 + }, + { + "epoch": 0.5397540083461454, + "grad_norm": 2.0825088024139404, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.702523410320282, + "num_tokens": 125249325.0, + "step": 4915 + }, + { + "epoch": 0.539863826048759, + "grad_norm": 2.0848634243011475, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7139734029769897, + "num_tokens": 125276931.0, + "step": 4916 + }, + { + "epoch": 0.5399736437513727, + "grad_norm": 2.149256706237793, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.6913616061210632, + "num_tokens": 125303063.0, + "step": 4917 + }, + { + "epoch": 0.5400834614539863, + "grad_norm": 2.278691053390503, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7116255760192871, + "num_tokens": 125325614.0, + "step": 4918 + }, + { + "epoch": 0.5401932791566001, + "grad_norm": 2.438776731491089, + "learning_rate": 1e-06, + "loss": 1.0509, + "mean_token_accuracy": 0.672531247138977, + "num_tokens": 125347968.0, + "step": 4919 + }, + { + "epoch": 0.5403030968592137, + "grad_norm": 2.692686080932617, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.6951209306716919, + "num_tokens": 125366552.0, + "step": 4920 + }, + { + "epoch": 0.5404129145618274, + "grad_norm": 1.9980313777923584, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7034104466438293, + "num_tokens": 125396885.0, + "step": 4921 + }, + { + "epoch": 0.540522732264441, + "grad_norm": 1.991631031036377, + "learning_rate": 1e-06, + "loss": 1.0304, + "mean_token_accuracy": 0.6873557567596436, + "num_tokens": 125426685.0, + "step": 4922 + }, + { + "epoch": 0.5406325499670547, + "grad_norm": 2.3037188053131104, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.6956766843795776, + "num_tokens": 125451700.0, + "step": 4923 + }, + { + "epoch": 0.5407423676696683, + "grad_norm": 2.5884318351745605, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.697107195854187, + "num_tokens": 125471601.0, + "step": 4924 + }, + { + "epoch": 0.540852185372282, + "grad_norm": 2.075831651687622, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.6849184036254883, + "num_tokens": 125501801.0, + "step": 4925 + }, + { + "epoch": 0.5409620030748957, + "grad_norm": 2.308637857437134, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.722980797290802, + "num_tokens": 125524703.0, + "step": 4926 + }, + { + "epoch": 0.5410718207775094, + "grad_norm": 2.135133981704712, + "learning_rate": 1e-06, + "loss": 1.0764, + "mean_token_accuracy": 0.6815334558486938, + "num_tokens": 125550846.0, + "step": 4927 + }, + { + "epoch": 0.541181638480123, + "grad_norm": 2.2290782928466797, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.723818838596344, + "num_tokens": 125576556.0, + "step": 4928 + }, + { + "epoch": 0.5412914561827367, + "grad_norm": 2.2553317546844482, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6947272419929504, + "num_tokens": 125600875.0, + "step": 4929 + }, + { + "epoch": 0.5414012738853503, + "grad_norm": 2.5906946659088135, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7104165554046631, + "num_tokens": 125620104.0, + "step": 4930 + }, + { + "epoch": 0.541511091587964, + "grad_norm": 2.3345508575439453, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7140926122665405, + "num_tokens": 125641388.0, + "step": 4931 + }, + { + "epoch": 0.5416209092905776, + "grad_norm": 2.256073236465454, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6990504264831543, + "num_tokens": 125665313.0, + "step": 4932 + }, + { + "epoch": 0.5417307269931912, + "grad_norm": 2.0619757175445557, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7060010433197021, + "num_tokens": 125692118.0, + "step": 4933 + }, + { + "epoch": 0.541840544695805, + "grad_norm": 2.1334476470947266, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.6893552541732788, + "num_tokens": 125720294.0, + "step": 4934 + }, + { + "epoch": 0.5419503623984187, + "grad_norm": 2.503154754638672, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7130612730979919, + "num_tokens": 125740120.0, + "step": 4935 + }, + { + "epoch": 0.5420601801010323, + "grad_norm": 2.0597941875457764, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7272089719772339, + "num_tokens": 125767254.0, + "step": 4936 + }, + { + "epoch": 0.5421699978036459, + "grad_norm": 2.280315637588501, + "learning_rate": 1e-06, + "loss": 1.0947, + "mean_token_accuracy": 0.6752040386199951, + "num_tokens": 125791370.0, + "step": 4937 + }, + { + "epoch": 0.5422798155062596, + "grad_norm": 2.108424425125122, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.6975666284561157, + "num_tokens": 125820490.0, + "step": 4938 + }, + { + "epoch": 0.5423896332088732, + "grad_norm": 2.1851253509521484, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.7007856965065002, + "num_tokens": 125847243.0, + "step": 4939 + }, + { + "epoch": 0.5424994509114869, + "grad_norm": 2.2090609073638916, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7028582096099854, + "num_tokens": 125872593.0, + "step": 4940 + }, + { + "epoch": 0.5426092686141006, + "grad_norm": 2.398700714111328, + "learning_rate": 1e-06, + "loss": 1.0542, + "mean_token_accuracy": 0.6832764148712158, + "num_tokens": 125897071.0, + "step": 4941 + }, + { + "epoch": 0.5427190863167143, + "grad_norm": 2.2271728515625, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7269864678382874, + "num_tokens": 125920558.0, + "step": 4942 + }, + { + "epoch": 0.5428289040193279, + "grad_norm": 2.057295560836792, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7093803882598877, + "num_tokens": 125947503.0, + "step": 4943 + }, + { + "epoch": 0.5429387217219416, + "grad_norm": 2.281156063079834, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7153136730194092, + "num_tokens": 125970680.0, + "step": 4944 + }, + { + "epoch": 0.5430485394245552, + "grad_norm": 2.212939500808716, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.717511773109436, + "num_tokens": 125995157.0, + "step": 4945 + }, + { + "epoch": 0.5431583571271689, + "grad_norm": 1.9689728021621704, + "learning_rate": 1e-06, + "loss": 1.0464, + "mean_token_accuracy": 0.6923748254776001, + "num_tokens": 126026407.0, + "step": 4946 + }, + { + "epoch": 0.5432681748297825, + "grad_norm": 2.359985113143921, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7028030157089233, + "num_tokens": 126048769.0, + "step": 4947 + }, + { + "epoch": 0.5433779925323963, + "grad_norm": 2.3021538257598877, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7072519063949585, + "num_tokens": 126071757.0, + "step": 4948 + }, + { + "epoch": 0.5434878102350099, + "grad_norm": 2.2170615196228027, + "learning_rate": 1e-06, + "loss": 1.0358, + "mean_token_accuracy": 0.6906962394714355, + "num_tokens": 126097078.0, + "step": 4949 + }, + { + "epoch": 0.5435976279376236, + "grad_norm": 2.050971508026123, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.7018314599990845, + "num_tokens": 126125492.0, + "step": 4950 + }, + { + "epoch": 0.5437074456402372, + "grad_norm": 2.2677366733551025, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.7001936435699463, + "num_tokens": 126149359.0, + "step": 4951 + }, + { + "epoch": 0.5438172633428509, + "grad_norm": 2.5337438583374023, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7029958963394165, + "num_tokens": 126169708.0, + "step": 4952 + }, + { + "epoch": 0.5439270810454645, + "grad_norm": 2.1080431938171387, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7092703580856323, + "num_tokens": 126196996.0, + "step": 4953 + }, + { + "epoch": 0.5440368987480781, + "grad_norm": 2.007756471633911, + "learning_rate": 1e-06, + "loss": 1.0673, + "mean_token_accuracy": 0.6761147975921631, + "num_tokens": 126228521.0, + "step": 4954 + }, + { + "epoch": 0.5441467164506919, + "grad_norm": 1.939350962638855, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.7017713189125061, + "num_tokens": 126258661.0, + "step": 4955 + }, + { + "epoch": 0.5442565341533055, + "grad_norm": 2.2169365882873535, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.705814003944397, + "num_tokens": 126284595.0, + "step": 4956 + }, + { + "epoch": 0.5443663518559192, + "grad_norm": 2.3730719089508057, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7164033055305481, + "num_tokens": 126306423.0, + "step": 4957 + }, + { + "epoch": 0.5444761695585328, + "grad_norm": 2.331807851791382, + "learning_rate": 1e-06, + "loss": 1.0439, + "mean_token_accuracy": 0.6932221055030823, + "num_tokens": 126329845.0, + "step": 4958 + }, + { + "epoch": 0.5445859872611465, + "grad_norm": 2.5136213302612305, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7205246686935425, + "num_tokens": 126349534.0, + "step": 4959 + }, + { + "epoch": 0.5446958049637601, + "grad_norm": 2.218930959701538, + "learning_rate": 1e-06, + "loss": 1.0952, + "mean_token_accuracy": 0.6724966764450073, + "num_tokens": 126376283.0, + "step": 4960 + }, + { + "epoch": 0.5448056226663738, + "grad_norm": 2.194249391555786, + "learning_rate": 1e-06, + "loss": 1.0598, + "mean_token_accuracy": 0.6811538338661194, + "num_tokens": 126401467.0, + "step": 4961 + }, + { + "epoch": 0.5449154403689875, + "grad_norm": 2.5426065921783447, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7162209749221802, + "num_tokens": 126421944.0, + "step": 4962 + }, + { + "epoch": 0.5450252580716012, + "grad_norm": 2.3361992835998535, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7135351896286011, + "num_tokens": 126444909.0, + "step": 4963 + }, + { + "epoch": 0.5451350757742148, + "grad_norm": 2.2841134071350098, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7003515362739563, + "num_tokens": 126469291.0, + "step": 4964 + }, + { + "epoch": 0.5452448934768285, + "grad_norm": 2.5907342433929443, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.718522310256958, + "num_tokens": 126489611.0, + "step": 4965 + }, + { + "epoch": 0.5453547111794421, + "grad_norm": 2.223341226577759, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7034667730331421, + "num_tokens": 126514335.0, + "step": 4966 + }, + { + "epoch": 0.5454645288820558, + "grad_norm": 2.0320346355438232, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.7055627703666687, + "num_tokens": 126545435.0, + "step": 4967 + }, + { + "epoch": 0.5455743465846694, + "grad_norm": 2.445010185241699, + "learning_rate": 1e-06, + "loss": 1.0677, + "mean_token_accuracy": 0.6783764362335205, + "num_tokens": 126567318.0, + "step": 4968 + }, + { + "epoch": 0.5456841642872831, + "grad_norm": 2.090283155441284, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7098695039749146, + "num_tokens": 126596212.0, + "step": 4969 + }, + { + "epoch": 0.5457939819898968, + "grad_norm": 2.1871705055236816, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7135652303695679, + "num_tokens": 126621707.0, + "step": 4970 + }, + { + "epoch": 0.5459037996925105, + "grad_norm": 2.6212031841278076, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.722070574760437, + "num_tokens": 126641458.0, + "step": 4971 + }, + { + "epoch": 0.5460136173951241, + "grad_norm": 2.0402488708496094, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7044433951377869, + "num_tokens": 126672605.0, + "step": 4972 + }, + { + "epoch": 0.5461234350977378, + "grad_norm": 2.4289095401763916, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.7009665966033936, + "num_tokens": 126695086.0, + "step": 4973 + }, + { + "epoch": 0.5462332528003514, + "grad_norm": 2.1432945728302, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7013334631919861, + "num_tokens": 126721750.0, + "step": 4974 + }, + { + "epoch": 0.546343070502965, + "grad_norm": 2.1126813888549805, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7071365714073181, + "num_tokens": 126747199.0, + "step": 4975 + }, + { + "epoch": 0.5464528882055787, + "grad_norm": 2.1040310859680176, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.7002713680267334, + "num_tokens": 126774857.0, + "step": 4976 + }, + { + "epoch": 0.5465627059081924, + "grad_norm": 2.223076105117798, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6900978088378906, + "num_tokens": 126800837.0, + "step": 4977 + }, + { + "epoch": 0.5466725236108061, + "grad_norm": 2.0827691555023193, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6910749673843384, + "num_tokens": 126829905.0, + "step": 4978 + }, + { + "epoch": 0.5467823413134197, + "grad_norm": 2.1071364879608154, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6864217519760132, + "num_tokens": 126857004.0, + "step": 4979 + }, + { + "epoch": 0.5468921590160334, + "grad_norm": 2.060218334197998, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.716166615486145, + "num_tokens": 126881429.0, + "step": 4980 + }, + { + "epoch": 0.547001976718647, + "grad_norm": 2.4726359844207764, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.6944931149482727, + "num_tokens": 126902726.0, + "step": 4981 + }, + { + "epoch": 0.5471117944212607, + "grad_norm": 2.3510262966156006, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7175713777542114, + "num_tokens": 126924128.0, + "step": 4982 + }, + { + "epoch": 0.5472216121238743, + "grad_norm": 2.5930395126342773, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.695641279220581, + "num_tokens": 126942034.0, + "step": 4983 + }, + { + "epoch": 0.5473314298264881, + "grad_norm": 2.0793299674987793, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7189282178878784, + "num_tokens": 126969822.0, + "step": 4984 + }, + { + "epoch": 0.5474412475291017, + "grad_norm": 2.575610876083374, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7108810544013977, + "num_tokens": 126990373.0, + "step": 4985 + }, + { + "epoch": 0.5475510652317154, + "grad_norm": 2.1185476779937744, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7253572940826416, + "num_tokens": 127016621.0, + "step": 4986 + }, + { + "epoch": 0.547660882934329, + "grad_norm": 2.5025105476379395, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7104605436325073, + "num_tokens": 127039650.0, + "step": 4987 + }, + { + "epoch": 0.5477707006369427, + "grad_norm": 2.1298909187316895, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6937378644943237, + "num_tokens": 127067057.0, + "step": 4988 + }, + { + "epoch": 0.5478805183395563, + "grad_norm": 2.117025852203369, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7130218148231506, + "num_tokens": 127094003.0, + "step": 4989 + }, + { + "epoch": 0.54799033604217, + "grad_norm": 2.237527847290039, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7182203531265259, + "num_tokens": 127119843.0, + "step": 4990 + }, + { + "epoch": 0.5481001537447837, + "grad_norm": 2.547495126724243, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6992959976196289, + "num_tokens": 127140190.0, + "step": 4991 + }, + { + "epoch": 0.5482099714473974, + "grad_norm": 2.3397600650787354, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7108222246170044, + "num_tokens": 127163613.0, + "step": 4992 + }, + { + "epoch": 0.548319789150011, + "grad_norm": 2.085538148880005, + "learning_rate": 1e-06, + "loss": 1.1003, + "mean_token_accuracy": 0.6721348762512207, + "num_tokens": 127192647.0, + "step": 4993 + }, + { + "epoch": 0.5484296068526247, + "grad_norm": 2.2060153484344482, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7240198254585266, + "num_tokens": 127216404.0, + "step": 4994 + }, + { + "epoch": 0.5485394245552383, + "grad_norm": 1.883094072341919, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.706531822681427, + "num_tokens": 127248112.0, + "step": 4995 + }, + { + "epoch": 0.5486492422578519, + "grad_norm": 2.122847318649292, + "learning_rate": 1e-06, + "loss": 1.1006, + "mean_token_accuracy": 0.667656421661377, + "num_tokens": 127277769.0, + "step": 4996 + }, + { + "epoch": 0.5487590599604656, + "grad_norm": 2.2259418964385986, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.710271954536438, + "num_tokens": 127302336.0, + "step": 4997 + }, + { + "epoch": 0.5488688776630792, + "grad_norm": 2.3393712043762207, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7135292887687683, + "num_tokens": 127326469.0, + "step": 4998 + }, + { + "epoch": 0.548978695365693, + "grad_norm": 2.059910774230957, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6933327913284302, + "num_tokens": 127353721.0, + "step": 4999 + }, + { + "epoch": 0.5490885130683066, + "grad_norm": 2.6318819522857666, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6944096088409424, + "num_tokens": 127372678.0, + "step": 5000 + }, + { + "epoch": 0.5491983307709203, + "grad_norm": 2.0816733837127686, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7139386534690857, + "num_tokens": 127400848.0, + "step": 5001 + }, + { + "epoch": 0.5493081484735339, + "grad_norm": 2.1313087940216064, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.709822416305542, + "num_tokens": 127427583.0, + "step": 5002 + }, + { + "epoch": 0.5494179661761476, + "grad_norm": 2.301957368850708, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.722525954246521, + "num_tokens": 127448898.0, + "step": 5003 + }, + { + "epoch": 0.5495277838787612, + "grad_norm": 2.044830322265625, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7235116362571716, + "num_tokens": 127477875.0, + "step": 5004 + }, + { + "epoch": 0.5496376015813749, + "grad_norm": 2.177672863006592, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7288037538528442, + "num_tokens": 127502213.0, + "step": 5005 + }, + { + "epoch": 0.5497474192839886, + "grad_norm": 2.121635913848877, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7193319797515869, + "num_tokens": 127526652.0, + "step": 5006 + }, + { + "epoch": 0.5498572369866023, + "grad_norm": 2.22249174118042, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.714052677154541, + "num_tokens": 127551211.0, + "step": 5007 + }, + { + "epoch": 0.5499670546892159, + "grad_norm": 2.391458749771118, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6954677104949951, + "num_tokens": 127574175.0, + "step": 5008 + }, + { + "epoch": 0.5500768723918296, + "grad_norm": 2.2799994945526123, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7081663012504578, + "num_tokens": 127596991.0, + "step": 5009 + }, + { + "epoch": 0.5501866900944432, + "grad_norm": 2.4677577018737793, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7078588008880615, + "num_tokens": 127617091.0, + "step": 5010 + }, + { + "epoch": 0.5502965077970569, + "grad_norm": 2.3216521739959717, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7229526042938232, + "num_tokens": 127637645.0, + "step": 5011 + }, + { + "epoch": 0.5504063254996705, + "grad_norm": 2.1297380924224854, + "learning_rate": 1e-06, + "loss": 1.0628, + "mean_token_accuracy": 0.6905398964881897, + "num_tokens": 127666396.0, + "step": 5012 + }, + { + "epoch": 0.5505161432022843, + "grad_norm": 2.176677942276001, + "learning_rate": 1e-06, + "loss": 1.012, + "mean_token_accuracy": 0.6893043518066406, + "num_tokens": 127691036.0, + "step": 5013 + }, + { + "epoch": 0.5506259609048979, + "grad_norm": 2.136507034301758, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7336513996124268, + "num_tokens": 127715857.0, + "step": 5014 + }, + { + "epoch": 0.5507357786075116, + "grad_norm": 2.1870925426483154, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7200514674186707, + "num_tokens": 127741048.0, + "step": 5015 + }, + { + "epoch": 0.5508455963101252, + "grad_norm": 2.1256325244903564, + "learning_rate": 1e-06, + "loss": 1.0567, + "mean_token_accuracy": 0.6844537854194641, + "num_tokens": 127767530.0, + "step": 5016 + }, + { + "epoch": 0.5509554140127388, + "grad_norm": 2.425626277923584, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7147083878517151, + "num_tokens": 127789306.0, + "step": 5017 + }, + { + "epoch": 0.5510652317153525, + "grad_norm": 2.374091863632202, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.712077260017395, + "num_tokens": 127812313.0, + "step": 5018 + }, + { + "epoch": 0.5511750494179661, + "grad_norm": 2.076535224914551, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7023954391479492, + "num_tokens": 127839317.0, + "step": 5019 + }, + { + "epoch": 0.5512848671205799, + "grad_norm": 2.2940399646759033, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.718045711517334, + "num_tokens": 127864384.0, + "step": 5020 + }, + { + "epoch": 0.5513946848231935, + "grad_norm": 2.161126136779785, + "learning_rate": 1e-06, + "loss": 1.1033, + "mean_token_accuracy": 0.6779552698135376, + "num_tokens": 127893749.0, + "step": 5021 + }, + { + "epoch": 0.5515045025258072, + "grad_norm": 2.3791637420654297, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.711971640586853, + "num_tokens": 127914328.0, + "step": 5022 + }, + { + "epoch": 0.5516143202284208, + "grad_norm": 2.526937246322632, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7443701028823853, + "num_tokens": 127933416.0, + "step": 5023 + }, + { + "epoch": 0.5517241379310345, + "grad_norm": 2.0115342140197754, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6926649212837219, + "num_tokens": 127963282.0, + "step": 5024 + }, + { + "epoch": 0.5518339556336481, + "grad_norm": 2.1264190673828125, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.6968791484832764, + "num_tokens": 127990806.0, + "step": 5025 + }, + { + "epoch": 0.5519437733362618, + "grad_norm": 2.382436513900757, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.713279128074646, + "num_tokens": 128013139.0, + "step": 5026 + }, + { + "epoch": 0.5520535910388754, + "grad_norm": 2.1288278102874756, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7135026454925537, + "num_tokens": 128040248.0, + "step": 5027 + }, + { + "epoch": 0.5521634087414892, + "grad_norm": 2.347489595413208, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7118347883224487, + "num_tokens": 128064959.0, + "step": 5028 + }, + { + "epoch": 0.5522732264441028, + "grad_norm": 2.603747844696045, + "learning_rate": 1e-06, + "loss": 1.0522, + "mean_token_accuracy": 0.6779855489730835, + "num_tokens": 128085141.0, + "step": 5029 + }, + { + "epoch": 0.5523830441467165, + "grad_norm": 2.3823325634002686, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.6998518705368042, + "num_tokens": 128106806.0, + "step": 5030 + }, + { + "epoch": 0.5524928618493301, + "grad_norm": 2.4194984436035156, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6956319808959961, + "num_tokens": 128130139.0, + "step": 5031 + }, + { + "epoch": 0.5526026795519438, + "grad_norm": 1.9863674640655518, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7105000019073486, + "num_tokens": 128159355.0, + "step": 5032 + }, + { + "epoch": 0.5527124972545574, + "grad_norm": 2.136326551437378, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6964657306671143, + "num_tokens": 128187483.0, + "step": 5033 + }, + { + "epoch": 0.552822314957171, + "grad_norm": 2.0286705493927, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7052655220031738, + "num_tokens": 128214268.0, + "step": 5034 + }, + { + "epoch": 0.5529321326597848, + "grad_norm": 2.4270927906036377, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.712257981300354, + "num_tokens": 128234085.0, + "step": 5035 + }, + { + "epoch": 0.5530419503623984, + "grad_norm": 2.0942223072052, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7091366052627563, + "num_tokens": 128262704.0, + "step": 5036 + }, + { + "epoch": 0.5531517680650121, + "grad_norm": 2.5500290393829346, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7283625602722168, + "num_tokens": 128280751.0, + "step": 5037 + }, + { + "epoch": 0.5532615857676257, + "grad_norm": 2.0927255153656006, + "learning_rate": 1e-06, + "loss": 1.0552, + "mean_token_accuracy": 0.6859005689620972, + "num_tokens": 128309562.0, + "step": 5038 + }, + { + "epoch": 0.5533714034702394, + "grad_norm": 2.0634171962738037, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.6988519430160522, + "num_tokens": 128339041.0, + "step": 5039 + }, + { + "epoch": 0.553481221172853, + "grad_norm": 2.147549867630005, + "learning_rate": 1e-06, + "loss": 1.1354, + "mean_token_accuracy": 0.6620981693267822, + "num_tokens": 128367793.0, + "step": 5040 + }, + { + "epoch": 0.5535910388754667, + "grad_norm": 2.062565565109253, + "learning_rate": 1e-06, + "loss": 1.0345, + "mean_token_accuracy": 0.6914701461791992, + "num_tokens": 128398113.0, + "step": 5041 + }, + { + "epoch": 0.5537008565780804, + "grad_norm": 2.427353858947754, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7168681621551514, + "num_tokens": 128419894.0, + "step": 5042 + }, + { + "epoch": 0.5538106742806941, + "grad_norm": 2.190748929977417, + "learning_rate": 1e-06, + "loss": 1.0556, + "mean_token_accuracy": 0.6777773499488831, + "num_tokens": 128445399.0, + "step": 5043 + }, + { + "epoch": 0.5539204919833077, + "grad_norm": 2.170480728149414, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7044821977615356, + "num_tokens": 128471214.0, + "step": 5044 + }, + { + "epoch": 0.5540303096859214, + "grad_norm": 1.980412483215332, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7009479403495789, + "num_tokens": 128500939.0, + "step": 5045 + }, + { + "epoch": 0.554140127388535, + "grad_norm": 2.336470127105713, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7093791961669922, + "num_tokens": 128524476.0, + "step": 5046 + }, + { + "epoch": 0.5542499450911487, + "grad_norm": 2.2295217514038086, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7092753648757935, + "num_tokens": 128552235.0, + "step": 5047 + }, + { + "epoch": 0.5543597627937623, + "grad_norm": 2.14736270904541, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7164281606674194, + "num_tokens": 128578564.0, + "step": 5048 + }, + { + "epoch": 0.5544695804963761, + "grad_norm": 2.2217442989349365, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7149077653884888, + "num_tokens": 128601866.0, + "step": 5049 + }, + { + "epoch": 0.5545793981989897, + "grad_norm": 2.434511423110962, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6988529562950134, + "num_tokens": 128624218.0, + "step": 5050 + }, + { + "epoch": 0.5546892159016034, + "grad_norm": 2.663844347000122, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.706513524055481, + "num_tokens": 128641647.0, + "step": 5051 + }, + { + "epoch": 0.554799033604217, + "grad_norm": 1.9506900310516357, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7195076942443848, + "num_tokens": 128671465.0, + "step": 5052 + }, + { + "epoch": 0.5549088513068307, + "grad_norm": 2.1653475761413574, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6991294026374817, + "num_tokens": 128698154.0, + "step": 5053 + }, + { + "epoch": 0.5550186690094443, + "grad_norm": 2.154264450073242, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7097935080528259, + "num_tokens": 128724161.0, + "step": 5054 + }, + { + "epoch": 0.555128486712058, + "grad_norm": 2.3378853797912598, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6950951814651489, + "num_tokens": 128749271.0, + "step": 5055 + }, + { + "epoch": 0.5552383044146716, + "grad_norm": 2.1432321071624756, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.697609007358551, + "num_tokens": 128775790.0, + "step": 5056 + }, + { + "epoch": 0.5553481221172853, + "grad_norm": 1.989410161972046, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.6965999603271484, + "num_tokens": 128807518.0, + "step": 5057 + }, + { + "epoch": 0.555457939819899, + "grad_norm": 2.268108606338501, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6972607374191284, + "num_tokens": 128831125.0, + "step": 5058 + }, + { + "epoch": 0.5555677575225126, + "grad_norm": 2.3981406688690186, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.6903600692749023, + "num_tokens": 128853691.0, + "step": 5059 + }, + { + "epoch": 0.5556775752251263, + "grad_norm": 2.0875298976898193, + "learning_rate": 1e-06, + "loss": 1.0448, + "mean_token_accuracy": 0.6875119209289551, + "num_tokens": 128881624.0, + "step": 5060 + }, + { + "epoch": 0.5557873929277399, + "grad_norm": 2.3912601470947266, + "learning_rate": 1e-06, + "loss": 1.058, + "mean_token_accuracy": 0.6924154758453369, + "num_tokens": 128904002.0, + "step": 5061 + }, + { + "epoch": 0.5558972106303536, + "grad_norm": 2.2229528427124023, + "learning_rate": 1e-06, + "loss": 1.046, + "mean_token_accuracy": 0.6897525787353516, + "num_tokens": 128931125.0, + "step": 5062 + }, + { + "epoch": 0.5560070283329672, + "grad_norm": 2.272456169128418, + "learning_rate": 1e-06, + "loss": 1.0541, + "mean_token_accuracy": 0.6785474419593811, + "num_tokens": 128957663.0, + "step": 5063 + }, + { + "epoch": 0.556116846035581, + "grad_norm": 2.314006805419922, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7043375968933105, + "num_tokens": 128982104.0, + "step": 5064 + }, + { + "epoch": 0.5562266637381946, + "grad_norm": 1.971584439277649, + "learning_rate": 1e-06, + "loss": 1.0556, + "mean_token_accuracy": 0.6879315376281738, + "num_tokens": 129014498.0, + "step": 5065 + }, + { + "epoch": 0.5563364814408083, + "grad_norm": 2.3855297565460205, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6964676976203918, + "num_tokens": 129037851.0, + "step": 5066 + }, + { + "epoch": 0.5564462991434219, + "grad_norm": 3.025446891784668, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7180790901184082, + "num_tokens": 129062160.0, + "step": 5067 + }, + { + "epoch": 0.5565561168460356, + "grad_norm": 2.125776529312134, + "learning_rate": 1e-06, + "loss": 1.0874, + "mean_token_accuracy": 0.6806830167770386, + "num_tokens": 129093422.0, + "step": 5068 + }, + { + "epoch": 0.5566659345486492, + "grad_norm": 2.2429721355438232, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6931629180908203, + "num_tokens": 129116546.0, + "step": 5069 + }, + { + "epoch": 0.5567757522512629, + "grad_norm": 2.169922351837158, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.6982944011688232, + "num_tokens": 129142617.0, + "step": 5070 + }, + { + "epoch": 0.5568855699538766, + "grad_norm": 2.5226433277130127, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7362291216850281, + "num_tokens": 129161155.0, + "step": 5071 + }, + { + "epoch": 0.5569953876564903, + "grad_norm": 2.1784517765045166, + "learning_rate": 1e-06, + "loss": 1.0461, + "mean_token_accuracy": 0.6836312413215637, + "num_tokens": 129185678.0, + "step": 5072 + }, + { + "epoch": 0.5571052053591039, + "grad_norm": 2.106693744659424, + "learning_rate": 1e-06, + "loss": 1.0851, + "mean_token_accuracy": 0.6757122278213501, + "num_tokens": 129212457.0, + "step": 5073 + }, + { + "epoch": 0.5572150230617176, + "grad_norm": 2.056474447250366, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.694674551486969, + "num_tokens": 129240720.0, + "step": 5074 + }, + { + "epoch": 0.5573248407643312, + "grad_norm": 2.2802343368530273, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.695022463798523, + "num_tokens": 129264509.0, + "step": 5075 + }, + { + "epoch": 0.5574346584669448, + "grad_norm": 2.438887357711792, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7132323980331421, + "num_tokens": 129286275.0, + "step": 5076 + }, + { + "epoch": 0.5575444761695585, + "grad_norm": 2.211780071258545, + "learning_rate": 1e-06, + "loss": 1.0296, + "mean_token_accuracy": 0.683510422706604, + "num_tokens": 129311097.0, + "step": 5077 + }, + { + "epoch": 0.5576542938721722, + "grad_norm": 2.1521615982055664, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7170761227607727, + "num_tokens": 129337533.0, + "step": 5078 + }, + { + "epoch": 0.5577641115747859, + "grad_norm": 2.229062795639038, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.7002377510070801, + "num_tokens": 129362478.0, + "step": 5079 + }, + { + "epoch": 0.5578739292773995, + "grad_norm": 2.235866069793701, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.7032368183135986, + "num_tokens": 129387249.0, + "step": 5080 + }, + { + "epoch": 0.5579837469800132, + "grad_norm": 2.211988925933838, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7132937908172607, + "num_tokens": 129412362.0, + "step": 5081 + }, + { + "epoch": 0.5580935646826268, + "grad_norm": 2.1794509887695312, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7194147109985352, + "num_tokens": 129436011.0, + "step": 5082 + }, + { + "epoch": 0.5582033823852405, + "grad_norm": 2.0984256267547607, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7012225389480591, + "num_tokens": 129461917.0, + "step": 5083 + }, + { + "epoch": 0.5583132000878541, + "grad_norm": 2.1685585975646973, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6997519731521606, + "num_tokens": 129487344.0, + "step": 5084 + }, + { + "epoch": 0.5584230177904678, + "grad_norm": 2.127697229385376, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7047222852706909, + "num_tokens": 129513855.0, + "step": 5085 + }, + { + "epoch": 0.5585328354930815, + "grad_norm": 2.4011924266815186, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6880044341087341, + "num_tokens": 129535741.0, + "step": 5086 + }, + { + "epoch": 0.5586426531956952, + "grad_norm": 2.0347931385040283, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7062370777130127, + "num_tokens": 129564526.0, + "step": 5087 + }, + { + "epoch": 0.5587524708983088, + "grad_norm": 2.3984830379486084, + "learning_rate": 1e-06, + "loss": 1.0427, + "mean_token_accuracy": 0.6893433928489685, + "num_tokens": 129587298.0, + "step": 5088 + }, + { + "epoch": 0.5588622886009225, + "grad_norm": 2.4921631813049316, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6933639049530029, + "num_tokens": 129607915.0, + "step": 5089 + }, + { + "epoch": 0.5589721063035361, + "grad_norm": 2.167771816253662, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6893824934959412, + "num_tokens": 129633298.0, + "step": 5090 + }, + { + "epoch": 0.5590819240061498, + "grad_norm": 2.190497636795044, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7117547988891602, + "num_tokens": 129656918.0, + "step": 5091 + }, + { + "epoch": 0.5591917417087634, + "grad_norm": 2.3059213161468506, + "learning_rate": 1e-06, + "loss": 1.0682, + "mean_token_accuracy": 0.6856346130371094, + "num_tokens": 129684497.0, + "step": 5092 + }, + { + "epoch": 0.5593015594113772, + "grad_norm": 1.9866304397583008, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.710084080696106, + "num_tokens": 129712199.0, + "step": 5093 + }, + { + "epoch": 0.5594113771139908, + "grad_norm": 2.098632335662842, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.6934145092964172, + "num_tokens": 129741927.0, + "step": 5094 + }, + { + "epoch": 0.5595211948166045, + "grad_norm": 2.0916061401367188, + "learning_rate": 1e-06, + "loss": 1.0359, + "mean_token_accuracy": 0.6916701197624207, + "num_tokens": 129770360.0, + "step": 5095 + }, + { + "epoch": 0.5596310125192181, + "grad_norm": 2.0447444915771484, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.7012022137641907, + "num_tokens": 129800124.0, + "step": 5096 + }, + { + "epoch": 0.5597408302218317, + "grad_norm": 1.9931024312973022, + "learning_rate": 1e-06, + "loss": 1.0519, + "mean_token_accuracy": 0.6792375445365906, + "num_tokens": 129830287.0, + "step": 5097 + }, + { + "epoch": 0.5598506479244454, + "grad_norm": 2.2870728969573975, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.691139817237854, + "num_tokens": 129854999.0, + "step": 5098 + }, + { + "epoch": 0.559960465627059, + "grad_norm": 2.1509158611297607, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7046322822570801, + "num_tokens": 129885050.0, + "step": 5099 + }, + { + "epoch": 0.5600702833296728, + "grad_norm": 2.1903443336486816, + "learning_rate": 1e-06, + "loss": 1.0408, + "mean_token_accuracy": 0.6984431743621826, + "num_tokens": 129912546.0, + "step": 5100 + }, + { + "epoch": 0.5601801010322864, + "grad_norm": 2.407245635986328, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.7037336826324463, + "num_tokens": 129934268.0, + "step": 5101 + }, + { + "epoch": 0.5602899187349001, + "grad_norm": 2.272707939147949, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6932873725891113, + "num_tokens": 129958017.0, + "step": 5102 + }, + { + "epoch": 0.5603997364375137, + "grad_norm": 2.1712117195129395, + "learning_rate": 1e-06, + "loss": 1.0247, + "mean_token_accuracy": 0.6965575814247131, + "num_tokens": 129985251.0, + "step": 5103 + }, + { + "epoch": 0.5605095541401274, + "grad_norm": 2.319458484649658, + "learning_rate": 1e-06, + "loss": 1.0228, + "mean_token_accuracy": 0.6910478472709656, + "num_tokens": 130007997.0, + "step": 5104 + }, + { + "epoch": 0.560619371842741, + "grad_norm": 2.0437397956848145, + "learning_rate": 1e-06, + "loss": 1.0545, + "mean_token_accuracy": 0.6867531538009644, + "num_tokens": 130038993.0, + "step": 5105 + }, + { + "epoch": 0.5607291895453547, + "grad_norm": 2.175976514816284, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6956979632377625, + "num_tokens": 130064125.0, + "step": 5106 + }, + { + "epoch": 0.5608390072479684, + "grad_norm": 2.227463722229004, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.733022153377533, + "num_tokens": 130091010.0, + "step": 5107 + }, + { + "epoch": 0.5609488249505821, + "grad_norm": 2.408325672149658, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7041369676589966, + "num_tokens": 130113028.0, + "step": 5108 + }, + { + "epoch": 0.5610586426531957, + "grad_norm": 2.2541022300720215, + "learning_rate": 1e-06, + "loss": 1.0601, + "mean_token_accuracy": 0.6749649047851562, + "num_tokens": 130138482.0, + "step": 5109 + }, + { + "epoch": 0.5611684603558094, + "grad_norm": 2.485623836517334, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7051980495452881, + "num_tokens": 130159075.0, + "step": 5110 + }, + { + "epoch": 0.561278278058423, + "grad_norm": 1.9775601625442505, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6941401958465576, + "num_tokens": 130187641.0, + "step": 5111 + }, + { + "epoch": 0.5613880957610367, + "grad_norm": 2.147578001022339, + "learning_rate": 1e-06, + "loss": 1.0727, + "mean_token_accuracy": 0.6946074962615967, + "num_tokens": 130213595.0, + "step": 5112 + }, + { + "epoch": 0.5614979134636503, + "grad_norm": 2.3946220874786377, + "learning_rate": 1e-06, + "loss": 1.0904, + "mean_token_accuracy": 0.677237868309021, + "num_tokens": 130237042.0, + "step": 5113 + }, + { + "epoch": 0.5616077311662641, + "grad_norm": 2.3441078662872314, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7057303786277771, + "num_tokens": 130258640.0, + "step": 5114 + }, + { + "epoch": 0.5617175488688777, + "grad_norm": 2.1240017414093018, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.6960259079933167, + "num_tokens": 130285735.0, + "step": 5115 + }, + { + "epoch": 0.5618273665714913, + "grad_norm": 2.083014965057373, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7298115491867065, + "num_tokens": 130312368.0, + "step": 5116 + }, + { + "epoch": 0.561937184274105, + "grad_norm": 2.2287657260894775, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7271766662597656, + "num_tokens": 130336293.0, + "step": 5117 + }, + { + "epoch": 0.5620470019767186, + "grad_norm": 1.9251376390457153, + "learning_rate": 1e-06, + "loss": 1.0887, + "mean_token_accuracy": 0.674142599105835, + "num_tokens": 130370573.0, + "step": 5118 + }, + { + "epoch": 0.5621568196793323, + "grad_norm": 2.5341596603393555, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7118710279464722, + "num_tokens": 130391569.0, + "step": 5119 + }, + { + "epoch": 0.5622666373819459, + "grad_norm": 2.1129586696624756, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.7029224038124084, + "num_tokens": 130420578.0, + "step": 5120 + }, + { + "epoch": 0.5623764550845596, + "grad_norm": 2.0008630752563477, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6929091215133667, + "num_tokens": 130450747.0, + "step": 5121 + }, + { + "epoch": 0.5624862727871733, + "grad_norm": 2.1510415077209473, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7205432057380676, + "num_tokens": 130475454.0, + "step": 5122 + }, + { + "epoch": 0.562596090489787, + "grad_norm": 1.9808787107467651, + "learning_rate": 1e-06, + "loss": 1.0858, + "mean_token_accuracy": 0.6737369298934937, + "num_tokens": 130508968.0, + "step": 5123 + }, + { + "epoch": 0.5627059081924006, + "grad_norm": 2.2089922428131104, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7196816205978394, + "num_tokens": 130534062.0, + "step": 5124 + }, + { + "epoch": 0.5628157258950143, + "grad_norm": 2.3856327533721924, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.6930265426635742, + "num_tokens": 130558747.0, + "step": 5125 + }, + { + "epoch": 0.5629255435976279, + "grad_norm": 2.10599946975708, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7012003660202026, + "num_tokens": 130586525.0, + "step": 5126 + }, + { + "epoch": 0.5630353613002416, + "grad_norm": 2.327697515487671, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7106432914733887, + "num_tokens": 130609028.0, + "step": 5127 + }, + { + "epoch": 0.5631451790028552, + "grad_norm": 2.5155768394470215, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7176546454429626, + "num_tokens": 130629316.0, + "step": 5128 + }, + { + "epoch": 0.563254996705469, + "grad_norm": 2.083313465118408, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7098019123077393, + "num_tokens": 130658106.0, + "step": 5129 + }, + { + "epoch": 0.5633648144080826, + "grad_norm": 2.0753695964813232, + "learning_rate": 1e-06, + "loss": 1.111, + "mean_token_accuracy": 0.6742972135543823, + "num_tokens": 130689352.0, + "step": 5130 + }, + { + "epoch": 0.5634746321106963, + "grad_norm": 1.857393741607666, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7221176028251648, + "num_tokens": 130722656.0, + "step": 5131 + }, + { + "epoch": 0.5635844498133099, + "grad_norm": 2.218791961669922, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6916085481643677, + "num_tokens": 130749781.0, + "step": 5132 + }, + { + "epoch": 0.5636942675159236, + "grad_norm": 2.3122916221618652, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7062805891036987, + "num_tokens": 130772356.0, + "step": 5133 + }, + { + "epoch": 0.5638040852185372, + "grad_norm": 2.151000738143921, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7308138012886047, + "num_tokens": 130799048.0, + "step": 5134 + }, + { + "epoch": 0.5639139029211508, + "grad_norm": 2.077349901199341, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7048230171203613, + "num_tokens": 130827231.0, + "step": 5135 + }, + { + "epoch": 0.5640237206237646, + "grad_norm": 2.2582905292510986, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7012019157409668, + "num_tokens": 130850939.0, + "step": 5136 + }, + { + "epoch": 0.5641335383263782, + "grad_norm": 2.070481777191162, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.706659734249115, + "num_tokens": 130878630.0, + "step": 5137 + }, + { + "epoch": 0.5642433560289919, + "grad_norm": 2.4166131019592285, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.7041993141174316, + "num_tokens": 130902492.0, + "step": 5138 + }, + { + "epoch": 0.5643531737316055, + "grad_norm": 2.081484317779541, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7155393362045288, + "num_tokens": 130928790.0, + "step": 5139 + }, + { + "epoch": 0.5644629914342192, + "grad_norm": 2.175475835800171, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6900338530540466, + "num_tokens": 130956888.0, + "step": 5140 + }, + { + "epoch": 0.5645728091368328, + "grad_norm": 2.204275608062744, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7269702553749084, + "num_tokens": 130983461.0, + "step": 5141 + }, + { + "epoch": 0.5646826268394465, + "grad_norm": 2.2771034240722656, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.6846311092376709, + "num_tokens": 131008757.0, + "step": 5142 + }, + { + "epoch": 0.5647924445420602, + "grad_norm": 2.0806994438171387, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.7061693668365479, + "num_tokens": 131038642.0, + "step": 5143 + }, + { + "epoch": 0.5649022622446739, + "grad_norm": 2.3515284061431885, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7204309701919556, + "num_tokens": 131059473.0, + "step": 5144 + }, + { + "epoch": 0.5650120799472875, + "grad_norm": 2.2628347873687744, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.7041193246841431, + "num_tokens": 131082761.0, + "step": 5145 + }, + { + "epoch": 0.5651218976499012, + "grad_norm": 2.144944190979004, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7029841542243958, + "num_tokens": 131110528.0, + "step": 5146 + }, + { + "epoch": 0.5652317153525148, + "grad_norm": 2.234933853149414, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6829797029495239, + "num_tokens": 131135374.0, + "step": 5147 + }, + { + "epoch": 0.5653415330551285, + "grad_norm": 2.1245670318603516, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7063417434692383, + "num_tokens": 131161723.0, + "step": 5148 + }, + { + "epoch": 0.5654513507577421, + "grad_norm": 2.0710208415985107, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7188420295715332, + "num_tokens": 131189657.0, + "step": 5149 + }, + { + "epoch": 0.5655611684603558, + "grad_norm": 2.2682430744171143, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7073262929916382, + "num_tokens": 131212924.0, + "step": 5150 + }, + { + "epoch": 0.5656709861629695, + "grad_norm": 2.3353233337402344, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7079726457595825, + "num_tokens": 131236772.0, + "step": 5151 + }, + { + "epoch": 0.5657808038655832, + "grad_norm": 2.185546875, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7135343551635742, + "num_tokens": 131261071.0, + "step": 5152 + }, + { + "epoch": 0.5658906215681968, + "grad_norm": 2.1731626987457275, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6985375881195068, + "num_tokens": 131288202.0, + "step": 5153 + }, + { + "epoch": 0.5660004392708105, + "grad_norm": 2.00893235206604, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7130793333053589, + "num_tokens": 131317208.0, + "step": 5154 + }, + { + "epoch": 0.5661102569734241, + "grad_norm": 2.144012689590454, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7095615863800049, + "num_tokens": 131345365.0, + "step": 5155 + }, + { + "epoch": 0.5662200746760377, + "grad_norm": 2.2327182292938232, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6864994764328003, + "num_tokens": 131372245.0, + "step": 5156 + }, + { + "epoch": 0.5663298923786514, + "grad_norm": 2.3037850856781006, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7146398425102234, + "num_tokens": 131397731.0, + "step": 5157 + }, + { + "epoch": 0.5664397100812651, + "grad_norm": 2.150073528289795, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7092946767807007, + "num_tokens": 131424581.0, + "step": 5158 + }, + { + "epoch": 0.5665495277838788, + "grad_norm": 2.675292491912842, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7001781463623047, + "num_tokens": 131442574.0, + "step": 5159 + }, + { + "epoch": 0.5666593454864924, + "grad_norm": 2.3304085731506348, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7286425232887268, + "num_tokens": 131466855.0, + "step": 5160 + }, + { + "epoch": 0.5667691631891061, + "grad_norm": 2.7528884410858154, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7340997457504272, + "num_tokens": 131482550.0, + "step": 5161 + }, + { + "epoch": 0.5668789808917197, + "grad_norm": 2.266721248626709, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7189358472824097, + "num_tokens": 131505190.0, + "step": 5162 + }, + { + "epoch": 0.5669887985943334, + "grad_norm": 1.9506279230117798, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.6995540261268616, + "num_tokens": 131536274.0, + "step": 5163 + }, + { + "epoch": 0.567098616296947, + "grad_norm": 1.9883736371994019, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7149649858474731, + "num_tokens": 131566618.0, + "step": 5164 + }, + { + "epoch": 0.5672084339995608, + "grad_norm": 2.365579128265381, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.7006419897079468, + "num_tokens": 131588505.0, + "step": 5165 + }, + { + "epoch": 0.5673182517021744, + "grad_norm": 2.0704259872436523, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.7000544667243958, + "num_tokens": 131619574.0, + "step": 5166 + }, + { + "epoch": 0.5674280694047881, + "grad_norm": 2.1021273136138916, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.697611927986145, + "num_tokens": 131648526.0, + "step": 5167 + }, + { + "epoch": 0.5675378871074017, + "grad_norm": 2.560479164123535, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7254937887191772, + "num_tokens": 131667215.0, + "step": 5168 + }, + { + "epoch": 0.5676477048100154, + "grad_norm": 1.9811352491378784, + "learning_rate": 1e-06, + "loss": 1.0244, + "mean_token_accuracy": 0.6847327351570129, + "num_tokens": 131697128.0, + "step": 5169 + }, + { + "epoch": 0.567757522512629, + "grad_norm": 2.8404886722564697, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7092763185501099, + "num_tokens": 131717477.0, + "step": 5170 + }, + { + "epoch": 0.5678673402152427, + "grad_norm": 2.174111843109131, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7148408889770508, + "num_tokens": 131743414.0, + "step": 5171 + }, + { + "epoch": 0.5679771579178564, + "grad_norm": 2.4669744968414307, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.6972398161888123, + "num_tokens": 131765215.0, + "step": 5172 + }, + { + "epoch": 0.5680869756204701, + "grad_norm": 2.0787625312805176, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7036269903182983, + "num_tokens": 131792984.0, + "step": 5173 + }, + { + "epoch": 0.5681967933230837, + "grad_norm": 2.082676410675049, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7005239725112915, + "num_tokens": 131822354.0, + "step": 5174 + }, + { + "epoch": 0.5683066110256974, + "grad_norm": 2.436507225036621, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7228002548217773, + "num_tokens": 131843357.0, + "step": 5175 + }, + { + "epoch": 0.568416428728311, + "grad_norm": 2.12221360206604, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6952749490737915, + "num_tokens": 131871987.0, + "step": 5176 + }, + { + "epoch": 0.5685262464309246, + "grad_norm": 2.6247036457061768, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7072811126708984, + "num_tokens": 131893402.0, + "step": 5177 + }, + { + "epoch": 0.5686360641335383, + "grad_norm": 1.8647741079330444, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6870533227920532, + "num_tokens": 131928130.0, + "step": 5178 + }, + { + "epoch": 0.5687458818361519, + "grad_norm": 2.156479597091675, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.7015513181686401, + "num_tokens": 131955007.0, + "step": 5179 + }, + { + "epoch": 0.5688556995387657, + "grad_norm": 2.0177133083343506, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6945134997367859, + "num_tokens": 131984111.0, + "step": 5180 + }, + { + "epoch": 0.5689655172413793, + "grad_norm": 2.11801815032959, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6917584538459778, + "num_tokens": 132011654.0, + "step": 5181 + }, + { + "epoch": 0.569075334943993, + "grad_norm": 2.1374335289001465, + "learning_rate": 1e-06, + "loss": 1.0514, + "mean_token_accuracy": 0.6863277554512024, + "num_tokens": 132040115.0, + "step": 5182 + }, + { + "epoch": 0.5691851526466066, + "grad_norm": 1.8837885856628418, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7037312984466553, + "num_tokens": 132072733.0, + "step": 5183 + }, + { + "epoch": 0.5692949703492203, + "grad_norm": 2.3646414279937744, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6998450756072998, + "num_tokens": 132097305.0, + "step": 5184 + }, + { + "epoch": 0.5694047880518339, + "grad_norm": 2.0167453289031982, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7010546922683716, + "num_tokens": 132127404.0, + "step": 5185 + }, + { + "epoch": 0.5695146057544476, + "grad_norm": 2.2090811729431152, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7298576831817627, + "num_tokens": 132150474.0, + "step": 5186 + }, + { + "epoch": 0.5696244234570613, + "grad_norm": 2.3210058212280273, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6943273544311523, + "num_tokens": 132174884.0, + "step": 5187 + }, + { + "epoch": 0.569734241159675, + "grad_norm": 2.157076835632324, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7289679646492004, + "num_tokens": 132201025.0, + "step": 5188 + }, + { + "epoch": 0.5698440588622886, + "grad_norm": 2.5068397521972656, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7217481136322021, + "num_tokens": 132221265.0, + "step": 5189 + }, + { + "epoch": 0.5699538765649023, + "grad_norm": 2.2753281593322754, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7067268490791321, + "num_tokens": 132244012.0, + "step": 5190 + }, + { + "epoch": 0.5700636942675159, + "grad_norm": 2.0602965354919434, + "learning_rate": 1e-06, + "loss": 1.0286, + "mean_token_accuracy": 0.6897619962692261, + "num_tokens": 132272399.0, + "step": 5191 + }, + { + "epoch": 0.5701735119701296, + "grad_norm": 2.006600856781006, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6964991092681885, + "num_tokens": 132300366.0, + "step": 5192 + }, + { + "epoch": 0.5702833296727432, + "grad_norm": 2.318286657333374, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.6989606022834778, + "num_tokens": 132323868.0, + "step": 5193 + }, + { + "epoch": 0.570393147375357, + "grad_norm": 2.448211908340454, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7276678085327148, + "num_tokens": 132343285.0, + "step": 5194 + }, + { + "epoch": 0.5705029650779706, + "grad_norm": 1.8558766841888428, + "learning_rate": 1e-06, + "loss": 1.0434, + "mean_token_accuracy": 0.6819092035293579, + "num_tokens": 132379270.0, + "step": 5195 + }, + { + "epoch": 0.5706127827805842, + "grad_norm": 2.0721817016601562, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7077013254165649, + "num_tokens": 132405945.0, + "step": 5196 + }, + { + "epoch": 0.5707226004831979, + "grad_norm": 2.157562732696533, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.6959898471832275, + "num_tokens": 132431397.0, + "step": 5197 + }, + { + "epoch": 0.5708324181858115, + "grad_norm": 2.3184618949890137, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.7010560631752014, + "num_tokens": 132453139.0, + "step": 5198 + }, + { + "epoch": 0.5709422358884252, + "grad_norm": 2.2444467544555664, + "learning_rate": 1e-06, + "loss": 0.9913, + "mean_token_accuracy": 0.699222207069397, + "num_tokens": 132478157.0, + "step": 5199 + }, + { + "epoch": 0.5710520535910388, + "grad_norm": 2.214481830596924, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7177157998085022, + "num_tokens": 132502594.0, + "step": 5200 + }, + { + "epoch": 0.5711618712936526, + "grad_norm": 2.1878063678741455, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6918822526931763, + "num_tokens": 132530685.0, + "step": 5201 + }, + { + "epoch": 0.5712716889962662, + "grad_norm": 1.9778870344161987, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.6990845799446106, + "num_tokens": 132563044.0, + "step": 5202 + }, + { + "epoch": 0.5713815066988799, + "grad_norm": 2.113999366760254, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7099546790122986, + "num_tokens": 132588094.0, + "step": 5203 + }, + { + "epoch": 0.5714913244014935, + "grad_norm": 2.132652521133423, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7023540735244751, + "num_tokens": 132614622.0, + "step": 5204 + }, + { + "epoch": 0.5716011421041072, + "grad_norm": 2.6273069381713867, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7124475240707397, + "num_tokens": 132634156.0, + "step": 5205 + }, + { + "epoch": 0.5717109598067208, + "grad_norm": 2.209235668182373, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.6910459399223328, + "num_tokens": 132660684.0, + "step": 5206 + }, + { + "epoch": 0.5718207775093345, + "grad_norm": 2.068369150161743, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7312484979629517, + "num_tokens": 132687420.0, + "step": 5207 + }, + { + "epoch": 0.5719305952119481, + "grad_norm": 2.348304510116577, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.6997594833374023, + "num_tokens": 132710633.0, + "step": 5208 + }, + { + "epoch": 0.5720404129145619, + "grad_norm": 2.205044984817505, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6849792003631592, + "num_tokens": 132736848.0, + "step": 5209 + }, + { + "epoch": 0.5721502306171755, + "grad_norm": 2.156435251235962, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.6959943175315857, + "num_tokens": 132763119.0, + "step": 5210 + }, + { + "epoch": 0.5722600483197892, + "grad_norm": 2.4996817111968994, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6974755525588989, + "num_tokens": 132784519.0, + "step": 5211 + }, + { + "epoch": 0.5723698660224028, + "grad_norm": 2.384580612182617, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7111233472824097, + "num_tokens": 132805818.0, + "step": 5212 + }, + { + "epoch": 0.5724796837250165, + "grad_norm": 2.1993401050567627, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6985836625099182, + "num_tokens": 132834131.0, + "step": 5213 + }, + { + "epoch": 0.5725895014276301, + "grad_norm": 2.0776240825653076, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7146229147911072, + "num_tokens": 132860747.0, + "step": 5214 + }, + { + "epoch": 0.5726993191302437, + "grad_norm": 2.299874782562256, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.6930004358291626, + "num_tokens": 132885354.0, + "step": 5215 + }, + { + "epoch": 0.5728091368328575, + "grad_norm": 2.0961923599243164, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6971170902252197, + "num_tokens": 132912650.0, + "step": 5216 + }, + { + "epoch": 0.5729189545354711, + "grad_norm": 2.24491810798645, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.686447024345398, + "num_tokens": 132939260.0, + "step": 5217 + }, + { + "epoch": 0.5730287722380848, + "grad_norm": 2.093095541000366, + "learning_rate": 1e-06, + "loss": 1.0506, + "mean_token_accuracy": 0.6903430223464966, + "num_tokens": 132969739.0, + "step": 5218 + }, + { + "epoch": 0.5731385899406984, + "grad_norm": 2.3291056156158447, + "learning_rate": 1e-06, + "loss": 1.0171, + "mean_token_accuracy": 0.7029440402984619, + "num_tokens": 132993336.0, + "step": 5219 + }, + { + "epoch": 0.5732484076433121, + "grad_norm": 2.4441981315612793, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7200412750244141, + "num_tokens": 133013805.0, + "step": 5220 + }, + { + "epoch": 0.5733582253459257, + "grad_norm": 2.214062452316284, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.6996303796768188, + "num_tokens": 133039485.0, + "step": 5221 + }, + { + "epoch": 0.5734680430485394, + "grad_norm": 2.122161626815796, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6970516443252563, + "num_tokens": 133065392.0, + "step": 5222 + }, + { + "epoch": 0.5735778607511531, + "grad_norm": 2.691786766052246, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7031771540641785, + "num_tokens": 133083977.0, + "step": 5223 + }, + { + "epoch": 0.5736876784537668, + "grad_norm": 2.231409788131714, + "learning_rate": 1e-06, + "loss": 1.0482, + "mean_token_accuracy": 0.68961101770401, + "num_tokens": 133111193.0, + "step": 5224 + }, + { + "epoch": 0.5737974961563804, + "grad_norm": 2.00504732131958, + "learning_rate": 1e-06, + "loss": 1.079, + "mean_token_accuracy": 0.6767928004264832, + "num_tokens": 133141828.0, + "step": 5225 + }, + { + "epoch": 0.5739073138589941, + "grad_norm": 2.157954216003418, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7045691013336182, + "num_tokens": 133169795.0, + "step": 5226 + }, + { + "epoch": 0.5740171315616077, + "grad_norm": 2.382530689239502, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7158598899841309, + "num_tokens": 133191814.0, + "step": 5227 + }, + { + "epoch": 0.5741269492642214, + "grad_norm": 2.112531900405884, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.691520094871521, + "num_tokens": 133220683.0, + "step": 5228 + }, + { + "epoch": 0.574236766966835, + "grad_norm": 2.4871654510498047, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7048372626304626, + "num_tokens": 133241340.0, + "step": 5229 + }, + { + "epoch": 0.5743465846694488, + "grad_norm": 2.0231270790100098, + "learning_rate": 1e-06, + "loss": 1.0502, + "mean_token_accuracy": 0.686834990978241, + "num_tokens": 133273876.0, + "step": 5230 + }, + { + "epoch": 0.5744564023720624, + "grad_norm": 2.260611057281494, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.7073425650596619, + "num_tokens": 133299212.0, + "step": 5231 + }, + { + "epoch": 0.5745662200746761, + "grad_norm": 2.4087307453155518, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.714553713798523, + "num_tokens": 133319912.0, + "step": 5232 + }, + { + "epoch": 0.5746760377772897, + "grad_norm": 2.302811622619629, + "learning_rate": 1e-06, + "loss": 1.0583, + "mean_token_accuracy": 0.6823376417160034, + "num_tokens": 133343927.0, + "step": 5233 + }, + { + "epoch": 0.5747858554799034, + "grad_norm": 2.0216870307922363, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7055160403251648, + "num_tokens": 133374579.0, + "step": 5234 + }, + { + "epoch": 0.574895673182517, + "grad_norm": 2.2777035236358643, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7117000222206116, + "num_tokens": 133398544.0, + "step": 5235 + }, + { + "epoch": 0.5750054908851306, + "grad_norm": 2.0404820442199707, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.714467465877533, + "num_tokens": 133426486.0, + "step": 5236 + }, + { + "epoch": 0.5751153085877443, + "grad_norm": 2.5021660327911377, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7136412858963013, + "num_tokens": 133445920.0, + "step": 5237 + }, + { + "epoch": 0.575225126290358, + "grad_norm": 1.9881162643432617, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7074400186538696, + "num_tokens": 133475761.0, + "step": 5238 + }, + { + "epoch": 0.5753349439929717, + "grad_norm": 2.281769275665283, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.6977335214614868, + "num_tokens": 133500120.0, + "step": 5239 + }, + { + "epoch": 0.5754447616955853, + "grad_norm": 1.9989583492279053, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.6981135606765747, + "num_tokens": 133531446.0, + "step": 5240 + }, + { + "epoch": 0.575554579398199, + "grad_norm": 2.4114112854003906, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7076876163482666, + "num_tokens": 133552867.0, + "step": 5241 + }, + { + "epoch": 0.5756643971008126, + "grad_norm": 1.8798319101333618, + "learning_rate": 1e-06, + "loss": 1.044, + "mean_token_accuracy": 0.6868451833724976, + "num_tokens": 133587045.0, + "step": 5242 + }, + { + "epoch": 0.5757742148034263, + "grad_norm": 2.3865320682525635, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.6999738216400146, + "num_tokens": 133611736.0, + "step": 5243 + }, + { + "epoch": 0.5758840325060399, + "grad_norm": 2.4895739555358887, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.722665548324585, + "num_tokens": 133631325.0, + "step": 5244 + }, + { + "epoch": 0.5759938502086537, + "grad_norm": 2.369495153427124, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.71089768409729, + "num_tokens": 133653672.0, + "step": 5245 + }, + { + "epoch": 0.5761036679112673, + "grad_norm": 2.236130952835083, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7253720760345459, + "num_tokens": 133677116.0, + "step": 5246 + }, + { + "epoch": 0.576213485613881, + "grad_norm": 2.271392583847046, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.705071747303009, + "num_tokens": 133701318.0, + "step": 5247 + }, + { + "epoch": 0.5763233033164946, + "grad_norm": 1.9884945154190063, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.7003072500228882, + "num_tokens": 133730692.0, + "step": 5248 + }, + { + "epoch": 0.5764331210191083, + "grad_norm": 1.9997483491897583, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6846693158149719, + "num_tokens": 133764442.0, + "step": 5249 + }, + { + "epoch": 0.5765429387217219, + "grad_norm": 2.3998241424560547, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.704757034778595, + "num_tokens": 133788314.0, + "step": 5250 + }, + { + "epoch": 0.5766527564243356, + "grad_norm": 2.267282247543335, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7032233476638794, + "num_tokens": 133811493.0, + "step": 5251 + }, + { + "epoch": 0.5767625741269493, + "grad_norm": 2.226483106613159, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.6993905305862427, + "num_tokens": 133835057.0, + "step": 5252 + }, + { + "epoch": 0.576872391829563, + "grad_norm": 2.1151821613311768, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.6955996751785278, + "num_tokens": 133862905.0, + "step": 5253 + }, + { + "epoch": 0.5769822095321766, + "grad_norm": 2.3996834754943848, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.7041679620742798, + "num_tokens": 133887257.0, + "step": 5254 + }, + { + "epoch": 0.5770920272347903, + "grad_norm": 2.3346304893493652, + "learning_rate": 1e-06, + "loss": 1.0201, + "mean_token_accuracy": 0.6882215738296509, + "num_tokens": 133910673.0, + "step": 5255 + }, + { + "epoch": 0.5772018449374039, + "grad_norm": 2.2204575538635254, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7012366056442261, + "num_tokens": 133935291.0, + "step": 5256 + }, + { + "epoch": 0.5773116626400175, + "grad_norm": 2.5520882606506348, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7191218137741089, + "num_tokens": 133956723.0, + "step": 5257 + }, + { + "epoch": 0.5774214803426312, + "grad_norm": 2.3136112689971924, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6946799755096436, + "num_tokens": 133982255.0, + "step": 5258 + }, + { + "epoch": 0.577531298045245, + "grad_norm": 2.154599905014038, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7212687134742737, + "num_tokens": 134008331.0, + "step": 5259 + }, + { + "epoch": 0.5776411157478586, + "grad_norm": 1.9724736213684082, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7172806859016418, + "num_tokens": 134038300.0, + "step": 5260 + }, + { + "epoch": 0.5777509334504722, + "grad_norm": 2.032412528991699, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6949436664581299, + "num_tokens": 134067680.0, + "step": 5261 + }, + { + "epoch": 0.5778607511530859, + "grad_norm": 2.5138871669769287, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7161370515823364, + "num_tokens": 134086633.0, + "step": 5262 + }, + { + "epoch": 0.5779705688556995, + "grad_norm": 2.1814699172973633, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7172390222549438, + "num_tokens": 134112143.0, + "step": 5263 + }, + { + "epoch": 0.5780803865583132, + "grad_norm": 2.355119466781616, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7042107582092285, + "num_tokens": 134134722.0, + "step": 5264 + }, + { + "epoch": 0.5781902042609268, + "grad_norm": 2.640594482421875, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.704139232635498, + "num_tokens": 134153920.0, + "step": 5265 + }, + { + "epoch": 0.5783000219635406, + "grad_norm": 2.1633007526397705, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7038464546203613, + "num_tokens": 134182619.0, + "step": 5266 + }, + { + "epoch": 0.5784098396661542, + "grad_norm": 2.379894495010376, + "learning_rate": 1e-06, + "loss": 1.1208, + "mean_token_accuracy": 0.6739621162414551, + "num_tokens": 134208412.0, + "step": 5267 + }, + { + "epoch": 0.5785196573687679, + "grad_norm": 1.9424331188201904, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7038541436195374, + "num_tokens": 134240488.0, + "step": 5268 + }, + { + "epoch": 0.5786294750713815, + "grad_norm": 2.6059513092041016, + "learning_rate": 1e-06, + "loss": 1.0224, + "mean_token_accuracy": 0.6861161589622498, + "num_tokens": 134261526.0, + "step": 5269 + }, + { + "epoch": 0.5787392927739952, + "grad_norm": 2.3145227432250977, + "learning_rate": 1e-06, + "loss": 1.0697, + "mean_token_accuracy": 0.6843755841255188, + "num_tokens": 134285228.0, + "step": 5270 + }, + { + "epoch": 0.5788491104766088, + "grad_norm": 2.1506383419036865, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7004333138465881, + "num_tokens": 134312143.0, + "step": 5271 + }, + { + "epoch": 0.5789589281792225, + "grad_norm": 2.2280375957489014, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7091678380966187, + "num_tokens": 134337134.0, + "step": 5272 + }, + { + "epoch": 0.5790687458818361, + "grad_norm": 2.079396963119507, + "learning_rate": 1e-06, + "loss": 1.062, + "mean_token_accuracy": 0.6832972764968872, + "num_tokens": 134367167.0, + "step": 5273 + }, + { + "epoch": 0.5791785635844499, + "grad_norm": 2.365216016769409, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7101856470108032, + "num_tokens": 134389557.0, + "step": 5274 + }, + { + "epoch": 0.5792883812870635, + "grad_norm": 2.3682029247283936, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7313294410705566, + "num_tokens": 134410859.0, + "step": 5275 + }, + { + "epoch": 0.5793981989896771, + "grad_norm": 2.385601043701172, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7168387174606323, + "num_tokens": 134431600.0, + "step": 5276 + }, + { + "epoch": 0.5795080166922908, + "grad_norm": 2.32867169380188, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6956132054328918, + "num_tokens": 134454831.0, + "step": 5277 + }, + { + "epoch": 0.5796178343949044, + "grad_norm": 2.3517696857452393, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7174496054649353, + "num_tokens": 134476773.0, + "step": 5278 + }, + { + "epoch": 0.5797276520975181, + "grad_norm": 2.0662243366241455, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6928979158401489, + "num_tokens": 134505059.0, + "step": 5279 + }, + { + "epoch": 0.5798374698001317, + "grad_norm": 2.4148972034454346, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.6974491477012634, + "num_tokens": 134527436.0, + "step": 5280 + }, + { + "epoch": 0.5799472875027455, + "grad_norm": 2.1412036418914795, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7087935209274292, + "num_tokens": 134553503.0, + "step": 5281 + }, + { + "epoch": 0.5800571052053591, + "grad_norm": 2.274996042251587, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7281283140182495, + "num_tokens": 134575833.0, + "step": 5282 + }, + { + "epoch": 0.5801669229079728, + "grad_norm": 2.543484926223755, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7108232975006104, + "num_tokens": 134596226.0, + "step": 5283 + }, + { + "epoch": 0.5802767406105864, + "grad_norm": 2.276059865951538, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.706579327583313, + "num_tokens": 134620465.0, + "step": 5284 + }, + { + "epoch": 0.5803865583132001, + "grad_norm": 2.173980236053467, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6920777559280396, + "num_tokens": 134647674.0, + "step": 5285 + }, + { + "epoch": 0.5804963760158137, + "grad_norm": 2.152594566345215, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7109746932983398, + "num_tokens": 134673310.0, + "step": 5286 + }, + { + "epoch": 0.5806061937184274, + "grad_norm": 2.286827564239502, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.6934619545936584, + "num_tokens": 134698700.0, + "step": 5287 + }, + { + "epoch": 0.5807160114210411, + "grad_norm": 2.1116676330566406, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.701565146446228, + "num_tokens": 134725800.0, + "step": 5288 + }, + { + "epoch": 0.5808258291236548, + "grad_norm": 2.2876107692718506, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7280594110488892, + "num_tokens": 134750107.0, + "step": 5289 + }, + { + "epoch": 0.5809356468262684, + "grad_norm": 2.0488340854644775, + "learning_rate": 1e-06, + "loss": 1.0679, + "mean_token_accuracy": 0.6720389723777771, + "num_tokens": 134778733.0, + "step": 5290 + }, + { + "epoch": 0.5810454645288821, + "grad_norm": 2.0954437255859375, + "learning_rate": 1e-06, + "loss": 0.9913, + "mean_token_accuracy": 0.6966239809989929, + "num_tokens": 134805909.0, + "step": 5291 + }, + { + "epoch": 0.5811552822314957, + "grad_norm": 1.93262779712677, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.731657862663269, + "num_tokens": 134834789.0, + "step": 5292 + }, + { + "epoch": 0.5812650999341094, + "grad_norm": 2.406893014907837, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7087295651435852, + "num_tokens": 134857017.0, + "step": 5293 + }, + { + "epoch": 0.581374917636723, + "grad_norm": 2.103639841079712, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7226814031600952, + "num_tokens": 134883700.0, + "step": 5294 + }, + { + "epoch": 0.5814847353393368, + "grad_norm": 2.14095139503479, + "learning_rate": 1e-06, + "loss": 1.0441, + "mean_token_accuracy": 0.6983044147491455, + "num_tokens": 134912193.0, + "step": 5295 + }, + { + "epoch": 0.5815945530419504, + "grad_norm": 2.0722951889038086, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7037770748138428, + "num_tokens": 134938424.0, + "step": 5296 + }, + { + "epoch": 0.581704370744564, + "grad_norm": 2.0869953632354736, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.712510883808136, + "num_tokens": 134966591.0, + "step": 5297 + }, + { + "epoch": 0.5818141884471777, + "grad_norm": 2.3908212184906006, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7152752876281738, + "num_tokens": 134988480.0, + "step": 5298 + }, + { + "epoch": 0.5819240061497913, + "grad_norm": 2.273221254348755, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6951370239257812, + "num_tokens": 135014097.0, + "step": 5299 + }, + { + "epoch": 0.582033823852405, + "grad_norm": 2.2754154205322266, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6958433389663696, + "num_tokens": 135038953.0, + "step": 5300 + }, + { + "epoch": 0.5821436415550186, + "grad_norm": 1.9495621919631958, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7306339740753174, + "num_tokens": 135068058.0, + "step": 5301 + }, + { + "epoch": 0.5822534592576323, + "grad_norm": 2.0430455207824707, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.6955781579017639, + "num_tokens": 135097122.0, + "step": 5302 + }, + { + "epoch": 0.582363276960246, + "grad_norm": 2.146043300628662, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.6975170373916626, + "num_tokens": 135122963.0, + "step": 5303 + }, + { + "epoch": 0.5824730946628597, + "grad_norm": 2.1188905239105225, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7292917370796204, + "num_tokens": 135148673.0, + "step": 5304 + }, + { + "epoch": 0.5825829123654733, + "grad_norm": 2.149742364883423, + "learning_rate": 1e-06, + "loss": 1.0684, + "mean_token_accuracy": 0.6856454610824585, + "num_tokens": 135177009.0, + "step": 5305 + }, + { + "epoch": 0.582692730068087, + "grad_norm": 2.183393716812134, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.6975897550582886, + "num_tokens": 135201728.0, + "step": 5306 + }, + { + "epoch": 0.5828025477707006, + "grad_norm": 2.1177680492401123, + "learning_rate": 1e-06, + "loss": 1.0546, + "mean_token_accuracy": 0.6782275438308716, + "num_tokens": 135228196.0, + "step": 5307 + }, + { + "epoch": 0.5829123654733143, + "grad_norm": 2.0591907501220703, + "learning_rate": 1e-06, + "loss": 1.0858, + "mean_token_accuracy": 0.6758893728256226, + "num_tokens": 135258568.0, + "step": 5308 + }, + { + "epoch": 0.5830221831759279, + "grad_norm": 2.125187873840332, + "learning_rate": 1e-06, + "loss": 1.0836, + "mean_token_accuracy": 0.6788020133972168, + "num_tokens": 135287058.0, + "step": 5309 + }, + { + "epoch": 0.5831320008785417, + "grad_norm": 2.4180328845977783, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7164466381072998, + "num_tokens": 135307408.0, + "step": 5310 + }, + { + "epoch": 0.5832418185811553, + "grad_norm": 2.137451171875, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.7013596892356873, + "num_tokens": 135333757.0, + "step": 5311 + }, + { + "epoch": 0.583351636283769, + "grad_norm": 2.023049831390381, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7047384977340698, + "num_tokens": 135362766.0, + "step": 5312 + }, + { + "epoch": 0.5834614539863826, + "grad_norm": 2.366886854171753, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7075086832046509, + "num_tokens": 135385504.0, + "step": 5313 + }, + { + "epoch": 0.5835712716889963, + "grad_norm": 2.188632011413574, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.7008882761001587, + "num_tokens": 135411205.0, + "step": 5314 + }, + { + "epoch": 0.5836810893916099, + "grad_norm": 2.0466620922088623, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6883071660995483, + "num_tokens": 135439154.0, + "step": 5315 + }, + { + "epoch": 0.5837909070942235, + "grad_norm": 2.492403984069824, + "learning_rate": 1e-06, + "loss": 1.0607, + "mean_token_accuracy": 0.6838688850402832, + "num_tokens": 135459734.0, + "step": 5316 + }, + { + "epoch": 0.5839007247968373, + "grad_norm": 2.3937766551971436, + "learning_rate": 1e-06, + "loss": 0.9791, + "mean_token_accuracy": 0.7047932147979736, + "num_tokens": 135480754.0, + "step": 5317 + }, + { + "epoch": 0.584010542499451, + "grad_norm": 2.2006676197052, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7042586207389832, + "num_tokens": 135505815.0, + "step": 5318 + }, + { + "epoch": 0.5841203602020646, + "grad_norm": 2.100130081176758, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7227756977081299, + "num_tokens": 135531749.0, + "step": 5319 + }, + { + "epoch": 0.5842301779046782, + "grad_norm": 2.2398085594177246, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.7005107402801514, + "num_tokens": 135555177.0, + "step": 5320 + }, + { + "epoch": 0.5843399956072919, + "grad_norm": 2.0665318965911865, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6963257789611816, + "num_tokens": 135583316.0, + "step": 5321 + }, + { + "epoch": 0.5844498133099055, + "grad_norm": 2.3251476287841797, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6926274299621582, + "num_tokens": 135609160.0, + "step": 5322 + }, + { + "epoch": 0.5845596310125192, + "grad_norm": 1.8174282312393188, + "learning_rate": 1e-06, + "loss": 1.0742, + "mean_token_accuracy": 0.6801630258560181, + "num_tokens": 135644594.0, + "step": 5323 + }, + { + "epoch": 0.5846694487151329, + "grad_norm": 1.9608333110809326, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6867499351501465, + "num_tokens": 135676767.0, + "step": 5324 + }, + { + "epoch": 0.5847792664177466, + "grad_norm": 1.9789098501205444, + "learning_rate": 1e-06, + "loss": 1.0397, + "mean_token_accuracy": 0.691835343837738, + "num_tokens": 135708650.0, + "step": 5325 + }, + { + "epoch": 0.5848890841203602, + "grad_norm": 2.318526029586792, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7151607871055603, + "num_tokens": 135731009.0, + "step": 5326 + }, + { + "epoch": 0.5849989018229739, + "grad_norm": 2.3346498012542725, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7106422781944275, + "num_tokens": 135754478.0, + "step": 5327 + }, + { + "epoch": 0.5851087195255875, + "grad_norm": 2.2664926052093506, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7101876735687256, + "num_tokens": 135778208.0, + "step": 5328 + }, + { + "epoch": 0.5852185372282012, + "grad_norm": 2.3981049060821533, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7251983880996704, + "num_tokens": 135798751.0, + "step": 5329 + }, + { + "epoch": 0.5853283549308148, + "grad_norm": 2.421823740005493, + "learning_rate": 1e-06, + "loss": 1.0295, + "mean_token_accuracy": 0.6864519119262695, + "num_tokens": 135821604.0, + "step": 5330 + }, + { + "epoch": 0.5854381726334285, + "grad_norm": 2.230031728744507, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7267236113548279, + "num_tokens": 135844485.0, + "step": 5331 + }, + { + "epoch": 0.5855479903360422, + "grad_norm": 2.984510898590088, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7371755838394165, + "num_tokens": 135867951.0, + "step": 5332 + }, + { + "epoch": 0.5856578080386559, + "grad_norm": 2.0313568115234375, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.7013014554977417, + "num_tokens": 135896046.0, + "step": 5333 + }, + { + "epoch": 0.5857676257412695, + "grad_norm": 2.4194464683532715, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7280508279800415, + "num_tokens": 135915204.0, + "step": 5334 + }, + { + "epoch": 0.5858774434438832, + "grad_norm": 2.2237932682037354, + "learning_rate": 1e-06, + "loss": 1.0274, + "mean_token_accuracy": 0.6938686370849609, + "num_tokens": 135942709.0, + "step": 5335 + }, + { + "epoch": 0.5859872611464968, + "grad_norm": 2.253941774368286, + "learning_rate": 1e-06, + "loss": 1.0295, + "mean_token_accuracy": 0.6855382919311523, + "num_tokens": 135969243.0, + "step": 5336 + }, + { + "epoch": 0.5860970788491104, + "grad_norm": 1.9963644742965698, + "learning_rate": 1e-06, + "loss": 1.0476, + "mean_token_accuracy": 0.6787502765655518, + "num_tokens": 135999733.0, + "step": 5337 + }, + { + "epoch": 0.5862068965517241, + "grad_norm": 2.2970340251922607, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7340509295463562, + "num_tokens": 136022414.0, + "step": 5338 + }, + { + "epoch": 0.5863167142543378, + "grad_norm": 2.171905755996704, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6969300508499146, + "num_tokens": 136049483.0, + "step": 5339 + }, + { + "epoch": 0.5864265319569515, + "grad_norm": 1.9432240724563599, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7073734998703003, + "num_tokens": 136078716.0, + "step": 5340 + }, + { + "epoch": 0.5865363496595651, + "grad_norm": 2.207857131958008, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7037911415100098, + "num_tokens": 136104478.0, + "step": 5341 + }, + { + "epoch": 0.5866461673621788, + "grad_norm": 2.311283588409424, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6896244287490845, + "num_tokens": 136127001.0, + "step": 5342 + }, + { + "epoch": 0.5867559850647924, + "grad_norm": 2.1318883895874023, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.699813961982727, + "num_tokens": 136155869.0, + "step": 5343 + }, + { + "epoch": 0.5868658027674061, + "grad_norm": 2.021803855895996, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6937084197998047, + "num_tokens": 136187126.0, + "step": 5344 + }, + { + "epoch": 0.5869756204700197, + "grad_norm": 2.19395112991333, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7304142713546753, + "num_tokens": 136211405.0, + "step": 5345 + }, + { + "epoch": 0.5870854381726335, + "grad_norm": 2.41237473487854, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7234072089195251, + "num_tokens": 136231545.0, + "step": 5346 + }, + { + "epoch": 0.5871952558752471, + "grad_norm": 2.5062055587768555, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.7039506435394287, + "num_tokens": 136252630.0, + "step": 5347 + }, + { + "epoch": 0.5873050735778608, + "grad_norm": 2.070559024810791, + "learning_rate": 1e-06, + "loss": 1.0764, + "mean_token_accuracy": 0.6783366799354553, + "num_tokens": 136283337.0, + "step": 5348 + }, + { + "epoch": 0.5874148912804744, + "grad_norm": 2.065784215927124, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6971886157989502, + "num_tokens": 136312434.0, + "step": 5349 + }, + { + "epoch": 0.5875247089830881, + "grad_norm": 2.020286798477173, + "learning_rate": 1e-06, + "loss": 1.1092, + "mean_token_accuracy": 0.6670249700546265, + "num_tokens": 136342708.0, + "step": 5350 + }, + { + "epoch": 0.5876345266857017, + "grad_norm": 2.039306640625, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.712725818157196, + "num_tokens": 136371030.0, + "step": 5351 + }, + { + "epoch": 0.5877443443883154, + "grad_norm": 2.310633897781372, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7175614833831787, + "num_tokens": 136392302.0, + "step": 5352 + }, + { + "epoch": 0.5878541620909291, + "grad_norm": 2.1512539386749268, + "learning_rate": 1e-06, + "loss": 1.0902, + "mean_token_accuracy": 0.6933924555778503, + "num_tokens": 136418194.0, + "step": 5353 + }, + { + "epoch": 0.5879639797935428, + "grad_norm": 2.087163209915161, + "learning_rate": 1e-06, + "loss": 1.07, + "mean_token_accuracy": 0.6763004660606384, + "num_tokens": 136446262.0, + "step": 5354 + }, + { + "epoch": 0.5880737974961564, + "grad_norm": 2.146820306777954, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.703345775604248, + "num_tokens": 136471555.0, + "step": 5355 + }, + { + "epoch": 0.58818361519877, + "grad_norm": 2.1947519779205322, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6968305110931396, + "num_tokens": 136496208.0, + "step": 5356 + }, + { + "epoch": 0.5882934329013837, + "grad_norm": 2.588622808456421, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7191005349159241, + "num_tokens": 136515484.0, + "step": 5357 + }, + { + "epoch": 0.5884032506039973, + "grad_norm": 2.3347694873809814, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.7040713429450989, + "num_tokens": 136539680.0, + "step": 5358 + }, + { + "epoch": 0.588513068306611, + "grad_norm": 2.6087141036987305, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7038097381591797, + "num_tokens": 136560029.0, + "step": 5359 + }, + { + "epoch": 0.5886228860092246, + "grad_norm": 2.084761619567871, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.696681797504425, + "num_tokens": 136588862.0, + "step": 5360 + }, + { + "epoch": 0.5887327037118384, + "grad_norm": 2.082667589187622, + "learning_rate": 1e-06, + "loss": 0.8313, + "mean_token_accuracy": 0.7453302145004272, + "num_tokens": 136615374.0, + "step": 5361 + }, + { + "epoch": 0.588842521414452, + "grad_norm": 2.3593218326568604, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7015398740768433, + "num_tokens": 136637529.0, + "step": 5362 + }, + { + "epoch": 0.5889523391170657, + "grad_norm": 2.099308729171753, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7143359780311584, + "num_tokens": 136663906.0, + "step": 5363 + }, + { + "epoch": 0.5890621568196793, + "grad_norm": 2.2667057514190674, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7217971682548523, + "num_tokens": 136685225.0, + "step": 5364 + }, + { + "epoch": 0.589171974522293, + "grad_norm": 2.3881642818450928, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7279298305511475, + "num_tokens": 136706403.0, + "step": 5365 + }, + { + "epoch": 0.5892817922249066, + "grad_norm": 2.4393157958984375, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7362611293792725, + "num_tokens": 136726932.0, + "step": 5366 + }, + { + "epoch": 0.5893916099275203, + "grad_norm": 2.50927734375, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7148127555847168, + "num_tokens": 136747014.0, + "step": 5367 + }, + { + "epoch": 0.589501427630134, + "grad_norm": 2.1935300827026367, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7128013968467712, + "num_tokens": 136772270.0, + "step": 5368 + }, + { + "epoch": 0.5896112453327477, + "grad_norm": 2.204859733581543, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.6906311511993408, + "num_tokens": 136797327.0, + "step": 5369 + }, + { + "epoch": 0.5897210630353613, + "grad_norm": 2.579982042312622, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7122292518615723, + "num_tokens": 136817285.0, + "step": 5370 + }, + { + "epoch": 0.589830880737975, + "grad_norm": 2.038688898086548, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.702938437461853, + "num_tokens": 136845907.0, + "step": 5371 + }, + { + "epoch": 0.5899406984405886, + "grad_norm": 2.9509105682373047, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7237557172775269, + "num_tokens": 136860702.0, + "step": 5372 + }, + { + "epoch": 0.5900505161432023, + "grad_norm": 2.0821797847747803, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6967772245407104, + "num_tokens": 136890558.0, + "step": 5373 + }, + { + "epoch": 0.5901603338458159, + "grad_norm": 2.0967109203338623, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7239639163017273, + "num_tokens": 136916651.0, + "step": 5374 + }, + { + "epoch": 0.5902701515484297, + "grad_norm": 2.1454200744628906, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.6958699226379395, + "num_tokens": 136944694.0, + "step": 5375 + }, + { + "epoch": 0.5903799692510433, + "grad_norm": 2.3160157203674316, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6977723240852356, + "num_tokens": 136970487.0, + "step": 5376 + }, + { + "epoch": 0.590489786953657, + "grad_norm": 2.038041591644287, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7074865698814392, + "num_tokens": 136997889.0, + "step": 5377 + }, + { + "epoch": 0.5905996046562706, + "grad_norm": 2.0953383445739746, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7049120664596558, + "num_tokens": 137025209.0, + "step": 5378 + }, + { + "epoch": 0.5907094223588842, + "grad_norm": 2.137096643447876, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.7013669013977051, + "num_tokens": 137052945.0, + "step": 5379 + }, + { + "epoch": 0.5908192400614979, + "grad_norm": 2.0441701412200928, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7147266268730164, + "num_tokens": 137082046.0, + "step": 5380 + }, + { + "epoch": 0.5909290577641115, + "grad_norm": 1.871334433555603, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.7030578851699829, + "num_tokens": 137113885.0, + "step": 5381 + }, + { + "epoch": 0.5910388754667253, + "grad_norm": 2.1309733390808105, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7016396522521973, + "num_tokens": 137141406.0, + "step": 5382 + }, + { + "epoch": 0.5911486931693389, + "grad_norm": 2.167053461074829, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6954936981201172, + "num_tokens": 137168729.0, + "step": 5383 + }, + { + "epoch": 0.5912585108719526, + "grad_norm": 2.4546608924865723, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7092568874359131, + "num_tokens": 137189776.0, + "step": 5384 + }, + { + "epoch": 0.5913683285745662, + "grad_norm": 2.13490629196167, + "learning_rate": 1e-06, + "loss": 1.0434, + "mean_token_accuracy": 0.6851176023483276, + "num_tokens": 137218748.0, + "step": 5385 + }, + { + "epoch": 0.5914781462771799, + "grad_norm": 2.1805648803710938, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.6960550546646118, + "num_tokens": 137245542.0, + "step": 5386 + }, + { + "epoch": 0.5915879639797935, + "grad_norm": 2.211374521255493, + "learning_rate": 1e-06, + "loss": 1.0609, + "mean_token_accuracy": 0.6782288551330566, + "num_tokens": 137269397.0, + "step": 5387 + }, + { + "epoch": 0.5916977816824072, + "grad_norm": 2.022737741470337, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6942899823188782, + "num_tokens": 137299294.0, + "step": 5388 + }, + { + "epoch": 0.5918075993850208, + "grad_norm": 2.312981605529785, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.6971168518066406, + "num_tokens": 137323115.0, + "step": 5389 + }, + { + "epoch": 0.5919174170876346, + "grad_norm": 2.341646194458008, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7067981362342834, + "num_tokens": 137346886.0, + "step": 5390 + }, + { + "epoch": 0.5920272347902482, + "grad_norm": 2.195497989654541, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7156878709793091, + "num_tokens": 137371840.0, + "step": 5391 + }, + { + "epoch": 0.5921370524928619, + "grad_norm": 2.1512374877929688, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7110099196434021, + "num_tokens": 137398654.0, + "step": 5392 + }, + { + "epoch": 0.5922468701954755, + "grad_norm": 1.9274438619613647, + "learning_rate": 1e-06, + "loss": 1.0454, + "mean_token_accuracy": 0.6863268613815308, + "num_tokens": 137431095.0, + "step": 5393 + }, + { + "epoch": 0.5923566878980892, + "grad_norm": 2.163194417953491, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7005102038383484, + "num_tokens": 137456663.0, + "step": 5394 + }, + { + "epoch": 0.5924665056007028, + "grad_norm": 2.0837085247039795, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7083279490470886, + "num_tokens": 137484886.0, + "step": 5395 + }, + { + "epoch": 0.5925763233033164, + "grad_norm": 2.4822723865509033, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.6915797591209412, + "num_tokens": 137506459.0, + "step": 5396 + }, + { + "epoch": 0.5926861410059302, + "grad_norm": 2.3367042541503906, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7157270908355713, + "num_tokens": 137529681.0, + "step": 5397 + }, + { + "epoch": 0.5927959587085438, + "grad_norm": 2.182460308074951, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6866176128387451, + "num_tokens": 137555151.0, + "step": 5398 + }, + { + "epoch": 0.5929057764111575, + "grad_norm": 2.1288647651672363, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7113690972328186, + "num_tokens": 137582100.0, + "step": 5399 + }, + { + "epoch": 0.5930155941137711, + "grad_norm": 2.427471876144409, + "learning_rate": 1e-06, + "loss": 1.0471, + "mean_token_accuracy": 0.6906104683876038, + "num_tokens": 137603371.0, + "step": 5400 + }, + { + "epoch": 0.5931254118163848, + "grad_norm": 2.3864974975585938, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7183576822280884, + "num_tokens": 137626305.0, + "step": 5401 + }, + { + "epoch": 0.5932352295189984, + "grad_norm": 2.0832974910736084, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7060776948928833, + "num_tokens": 137653285.0, + "step": 5402 + }, + { + "epoch": 0.5933450472216121, + "grad_norm": 2.4020142555236816, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7116529941558838, + "num_tokens": 137674670.0, + "step": 5403 + }, + { + "epoch": 0.5934548649242258, + "grad_norm": 1.9996103048324585, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7211796045303345, + "num_tokens": 137704384.0, + "step": 5404 + }, + { + "epoch": 0.5935646826268395, + "grad_norm": 2.1854395866394043, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6987371444702148, + "num_tokens": 137731322.0, + "step": 5405 + }, + { + "epoch": 0.5936745003294531, + "grad_norm": 2.4232428073883057, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.7016506195068359, + "num_tokens": 137753491.0, + "step": 5406 + }, + { + "epoch": 0.5937843180320668, + "grad_norm": 2.3484904766082764, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7215522527694702, + "num_tokens": 137776001.0, + "step": 5407 + }, + { + "epoch": 0.5938941357346804, + "grad_norm": 2.248178482055664, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7072933912277222, + "num_tokens": 137799883.0, + "step": 5408 + }, + { + "epoch": 0.5940039534372941, + "grad_norm": 2.0737171173095703, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7007662057876587, + "num_tokens": 137827011.0, + "step": 5409 + }, + { + "epoch": 0.5941137711399077, + "grad_norm": 2.2758185863494873, + "learning_rate": 1e-06, + "loss": 1.045, + "mean_token_accuracy": 0.6810721158981323, + "num_tokens": 137851149.0, + "step": 5410 + }, + { + "epoch": 0.5942235888425215, + "grad_norm": 2.371720314025879, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7231309413909912, + "num_tokens": 137870259.0, + "step": 5411 + }, + { + "epoch": 0.5943334065451351, + "grad_norm": 1.987105369567871, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7080795764923096, + "num_tokens": 137897411.0, + "step": 5412 + }, + { + "epoch": 0.5944432242477488, + "grad_norm": 2.10007381439209, + "learning_rate": 1e-06, + "loss": 1.0392, + "mean_token_accuracy": 0.6862173080444336, + "num_tokens": 137924931.0, + "step": 5413 + }, + { + "epoch": 0.5945530419503624, + "grad_norm": 1.9790289402008057, + "learning_rate": 1e-06, + "loss": 1.0369, + "mean_token_accuracy": 0.6905953288078308, + "num_tokens": 137955446.0, + "step": 5414 + }, + { + "epoch": 0.594662859652976, + "grad_norm": 2.3241794109344482, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7133904695510864, + "num_tokens": 137978560.0, + "step": 5415 + }, + { + "epoch": 0.5947726773555897, + "grad_norm": 2.3867578506469727, + "learning_rate": 1e-06, + "loss": 1.0556, + "mean_token_accuracy": 0.6922435760498047, + "num_tokens": 138003010.0, + "step": 5416 + }, + { + "epoch": 0.5948824950582033, + "grad_norm": 2.078742742538452, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.7191777229309082, + "num_tokens": 138031184.0, + "step": 5417 + }, + { + "epoch": 0.5949923127608171, + "grad_norm": 2.2476749420166016, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7085005044937134, + "num_tokens": 138054477.0, + "step": 5418 + }, + { + "epoch": 0.5951021304634307, + "grad_norm": 1.972209095954895, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7396343946456909, + "num_tokens": 138083716.0, + "step": 5419 + }, + { + "epoch": 0.5952119481660444, + "grad_norm": 2.1152708530426025, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7094159722328186, + "num_tokens": 138109635.0, + "step": 5420 + }, + { + "epoch": 0.595321765868658, + "grad_norm": 1.9925726652145386, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.7011082768440247, + "num_tokens": 138139187.0, + "step": 5421 + }, + { + "epoch": 0.5954315835712717, + "grad_norm": 2.191291093826294, + "learning_rate": 1e-06, + "loss": 1.0732, + "mean_token_accuracy": 0.6835615038871765, + "num_tokens": 138165299.0, + "step": 5422 + }, + { + "epoch": 0.5955414012738853, + "grad_norm": 2.5346076488494873, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6981072425842285, + "num_tokens": 138184780.0, + "step": 5423 + }, + { + "epoch": 0.595651218976499, + "grad_norm": 2.18269944190979, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7171869277954102, + "num_tokens": 138209929.0, + "step": 5424 + }, + { + "epoch": 0.5957610366791126, + "grad_norm": 2.4465224742889404, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6982172131538391, + "num_tokens": 138231648.0, + "step": 5425 + }, + { + "epoch": 0.5958708543817264, + "grad_norm": 2.0603528022766113, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6938055753707886, + "num_tokens": 138259863.0, + "step": 5426 + }, + { + "epoch": 0.59598067208434, + "grad_norm": 2.229640483856201, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.711788535118103, + "num_tokens": 138284045.0, + "step": 5427 + }, + { + "epoch": 0.5960904897869537, + "grad_norm": 2.108581066131592, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7045824527740479, + "num_tokens": 138313151.0, + "step": 5428 + }, + { + "epoch": 0.5962003074895673, + "grad_norm": 2.1256532669067383, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6876291036605835, + "num_tokens": 138340693.0, + "step": 5429 + }, + { + "epoch": 0.596310125192181, + "grad_norm": 2.4359238147735596, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7128403186798096, + "num_tokens": 138363441.0, + "step": 5430 + }, + { + "epoch": 0.5964199428947946, + "grad_norm": 2.3875157833099365, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7040566205978394, + "num_tokens": 138385538.0, + "step": 5431 + }, + { + "epoch": 0.5965297605974083, + "grad_norm": 2.14373779296875, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7201796770095825, + "num_tokens": 138413714.0, + "step": 5432 + }, + { + "epoch": 0.596639578300022, + "grad_norm": 2.245570421218872, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.7003278732299805, + "num_tokens": 138438627.0, + "step": 5433 + }, + { + "epoch": 0.5967493960026357, + "grad_norm": 2.143381357192993, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7162458300590515, + "num_tokens": 138465173.0, + "step": 5434 + }, + { + "epoch": 0.5968592137052493, + "grad_norm": 1.979607343673706, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7228950262069702, + "num_tokens": 138491690.0, + "step": 5435 + }, + { + "epoch": 0.596969031407863, + "grad_norm": 2.290527105331421, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.6852047443389893, + "num_tokens": 138517112.0, + "step": 5436 + }, + { + "epoch": 0.5970788491104766, + "grad_norm": 2.243077278137207, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7125300168991089, + "num_tokens": 138541073.0, + "step": 5437 + }, + { + "epoch": 0.5971886668130902, + "grad_norm": 2.205597400665283, + "learning_rate": 1e-06, + "loss": 1.04, + "mean_token_accuracy": 0.6918653845787048, + "num_tokens": 138566007.0, + "step": 5438 + }, + { + "epoch": 0.5972984845157039, + "grad_norm": 2.208007335662842, + "learning_rate": 1e-06, + "loss": 1.1151, + "mean_token_accuracy": 0.6665032505989075, + "num_tokens": 138593574.0, + "step": 5439 + }, + { + "epoch": 0.5974083022183176, + "grad_norm": 2.230290412902832, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.6981136798858643, + "num_tokens": 138618984.0, + "step": 5440 + }, + { + "epoch": 0.5975181199209313, + "grad_norm": 2.1349754333496094, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7080971002578735, + "num_tokens": 138644762.0, + "step": 5441 + }, + { + "epoch": 0.5976279376235449, + "grad_norm": 2.2841830253601074, + "learning_rate": 1e-06, + "loss": 1.0425, + "mean_token_accuracy": 0.6888890266418457, + "num_tokens": 138668707.0, + "step": 5442 + }, + { + "epoch": 0.5977377553261586, + "grad_norm": 2.0719099044799805, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6958389282226562, + "num_tokens": 138695042.0, + "step": 5443 + }, + { + "epoch": 0.5978475730287722, + "grad_norm": 2.3541674613952637, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.70295250415802, + "num_tokens": 138717256.0, + "step": 5444 + }, + { + "epoch": 0.5979573907313859, + "grad_norm": 2.217207670211792, + "learning_rate": 1e-06, + "loss": 1.0367, + "mean_token_accuracy": 0.6858927607536316, + "num_tokens": 138742864.0, + "step": 5445 + }, + { + "epoch": 0.5980672084339995, + "grad_norm": 1.9192817211151123, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6903901100158691, + "num_tokens": 138776741.0, + "step": 5446 + }, + { + "epoch": 0.5981770261366133, + "grad_norm": 2.2290446758270264, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6782360672950745, + "num_tokens": 138802015.0, + "step": 5447 + }, + { + "epoch": 0.5982868438392269, + "grad_norm": 2.1730616092681885, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7008622288703918, + "num_tokens": 138826509.0, + "step": 5448 + }, + { + "epoch": 0.5983966615418406, + "grad_norm": 2.09710431098938, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6925860643386841, + "num_tokens": 138855140.0, + "step": 5449 + }, + { + "epoch": 0.5985064792444542, + "grad_norm": 2.4643023014068604, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.715095043182373, + "num_tokens": 138875657.0, + "step": 5450 + }, + { + "epoch": 0.5986162969470679, + "grad_norm": 2.375734329223633, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.6954895257949829, + "num_tokens": 138897649.0, + "step": 5451 + }, + { + "epoch": 0.5987261146496815, + "grad_norm": 2.3950607776641846, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.718506932258606, + "num_tokens": 138918132.0, + "step": 5452 + }, + { + "epoch": 0.5988359323522952, + "grad_norm": 2.0063343048095703, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6988728046417236, + "num_tokens": 138949697.0, + "step": 5453 + }, + { + "epoch": 0.5989457500549088, + "grad_norm": 1.913760781288147, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.697867214679718, + "num_tokens": 138980234.0, + "step": 5454 + }, + { + "epoch": 0.5990555677575226, + "grad_norm": 2.1556010246276855, + "learning_rate": 1e-06, + "loss": 1.039, + "mean_token_accuracy": 0.6886658668518066, + "num_tokens": 139007514.0, + "step": 5455 + }, + { + "epoch": 0.5991653854601362, + "grad_norm": 2.0889902114868164, + "learning_rate": 1e-06, + "loss": 1.0733, + "mean_token_accuracy": 0.6797865033149719, + "num_tokens": 139034821.0, + "step": 5456 + }, + { + "epoch": 0.5992752031627498, + "grad_norm": 2.060903549194336, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.7078696489334106, + "num_tokens": 139062437.0, + "step": 5457 + }, + { + "epoch": 0.5993850208653635, + "grad_norm": 2.344165325164795, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7080782651901245, + "num_tokens": 139085805.0, + "step": 5458 + }, + { + "epoch": 0.5994948385679771, + "grad_norm": 2.0377819538116455, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.7021464109420776, + "num_tokens": 139115149.0, + "step": 5459 + }, + { + "epoch": 0.5996046562705908, + "grad_norm": 2.152240514755249, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7266896963119507, + "num_tokens": 139139401.0, + "step": 5460 + }, + { + "epoch": 0.5997144739732044, + "grad_norm": 1.8918464183807373, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.6965217590332031, + "num_tokens": 139170264.0, + "step": 5461 + }, + { + "epoch": 0.5998242916758182, + "grad_norm": 2.223977565765381, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6883572936058044, + "num_tokens": 139198005.0, + "step": 5462 + }, + { + "epoch": 0.5999341093784318, + "grad_norm": 2.4125845432281494, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7101326584815979, + "num_tokens": 139220552.0, + "step": 5463 + }, + { + "epoch": 0.6000439270810455, + "grad_norm": 2.384686231613159, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.7004395723342896, + "num_tokens": 139241811.0, + "step": 5464 + }, + { + "epoch": 0.6001537447836591, + "grad_norm": 2.445594549179077, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6942543983459473, + "num_tokens": 139267814.0, + "step": 5465 + }, + { + "epoch": 0.6002635624862728, + "grad_norm": 2.122908592224121, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7231428027153015, + "num_tokens": 139294253.0, + "step": 5466 + }, + { + "epoch": 0.6003733801888864, + "grad_norm": 1.9884902238845825, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7241202592849731, + "num_tokens": 139325385.0, + "step": 5467 + }, + { + "epoch": 0.6004831978915001, + "grad_norm": 2.5770177841186523, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7200746536254883, + "num_tokens": 139344389.0, + "step": 5468 + }, + { + "epoch": 0.6005930155941138, + "grad_norm": 2.336074113845825, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6863858103752136, + "num_tokens": 139369480.0, + "step": 5469 + }, + { + "epoch": 0.6007028332967275, + "grad_norm": 2.1132261753082275, + "learning_rate": 1e-06, + "loss": 1.0932, + "mean_token_accuracy": 0.6728322505950928, + "num_tokens": 139396654.0, + "step": 5470 + }, + { + "epoch": 0.6008126509993411, + "grad_norm": 2.352780342102051, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6973713636398315, + "num_tokens": 139420788.0, + "step": 5471 + }, + { + "epoch": 0.6009224687019548, + "grad_norm": 2.118086338043213, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7121546268463135, + "num_tokens": 139446711.0, + "step": 5472 + }, + { + "epoch": 0.6010322864045684, + "grad_norm": 2.358258008956909, + "learning_rate": 1e-06, + "loss": 1.0588, + "mean_token_accuracy": 0.6830257177352905, + "num_tokens": 139473577.0, + "step": 5473 + }, + { + "epoch": 0.601142104107182, + "grad_norm": 2.1031715869903564, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7114112973213196, + "num_tokens": 139498752.0, + "step": 5474 + }, + { + "epoch": 0.6012519218097957, + "grad_norm": 2.483384132385254, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6955568194389343, + "num_tokens": 139520723.0, + "step": 5475 + }, + { + "epoch": 0.6013617395124095, + "grad_norm": 2.497535467147827, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.714285135269165, + "num_tokens": 139541827.0, + "step": 5476 + }, + { + "epoch": 0.6014715572150231, + "grad_norm": 2.375328779220581, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.706886351108551, + "num_tokens": 139564202.0, + "step": 5477 + }, + { + "epoch": 0.6015813749176367, + "grad_norm": 2.300980567932129, + "learning_rate": 1e-06, + "loss": 1.0562, + "mean_token_accuracy": 0.6827763319015503, + "num_tokens": 139587082.0, + "step": 5478 + }, + { + "epoch": 0.6016911926202504, + "grad_norm": 2.0654563903808594, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6931103467941284, + "num_tokens": 139616026.0, + "step": 5479 + }, + { + "epoch": 0.601801010322864, + "grad_norm": 2.3757760524749756, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6910489797592163, + "num_tokens": 139637063.0, + "step": 5480 + }, + { + "epoch": 0.6019108280254777, + "grad_norm": 2.171539068222046, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7099965214729309, + "num_tokens": 139661714.0, + "step": 5481 + }, + { + "epoch": 0.6020206457280913, + "grad_norm": 2.3019444942474365, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7190138697624207, + "num_tokens": 139683698.0, + "step": 5482 + }, + { + "epoch": 0.602130463430705, + "grad_norm": 2.4549124240875244, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7137841582298279, + "num_tokens": 139705250.0, + "step": 5483 + }, + { + "epoch": 0.6022402811333187, + "grad_norm": 2.882922410964966, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7250652313232422, + "num_tokens": 139720830.0, + "step": 5484 + }, + { + "epoch": 0.6023500988359324, + "grad_norm": 2.265446186065674, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6957500576972961, + "num_tokens": 139747776.0, + "step": 5485 + }, + { + "epoch": 0.602459916538546, + "grad_norm": 2.213026762008667, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7013756632804871, + "num_tokens": 139772078.0, + "step": 5486 + }, + { + "epoch": 0.6025697342411597, + "grad_norm": 2.332923173904419, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7143723964691162, + "num_tokens": 139794143.0, + "step": 5487 + }, + { + "epoch": 0.6026795519437733, + "grad_norm": 2.689954996109009, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7394998669624329, + "num_tokens": 139811493.0, + "step": 5488 + }, + { + "epoch": 0.602789369646387, + "grad_norm": 2.1616034507751465, + "learning_rate": 1e-06, + "loss": 1.0196, + "mean_token_accuracy": 0.6983193159103394, + "num_tokens": 139838168.0, + "step": 5489 + }, + { + "epoch": 0.6028991873490006, + "grad_norm": 2.3155288696289062, + "learning_rate": 1e-06, + "loss": 1.0751, + "mean_token_accuracy": 0.6831259727478027, + "num_tokens": 139865749.0, + "step": 5490 + }, + { + "epoch": 0.6030090050516144, + "grad_norm": 2.4312140941619873, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7315343022346497, + "num_tokens": 139886587.0, + "step": 5491 + }, + { + "epoch": 0.603118822754228, + "grad_norm": 2.1794915199279785, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7057161331176758, + "num_tokens": 139912670.0, + "step": 5492 + }, + { + "epoch": 0.6032286404568417, + "grad_norm": 2.249894380569458, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.706696629524231, + "num_tokens": 139936413.0, + "step": 5493 + }, + { + "epoch": 0.6033384581594553, + "grad_norm": 2.0600485801696777, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7187812924385071, + "num_tokens": 139964175.0, + "step": 5494 + }, + { + "epoch": 0.603448275862069, + "grad_norm": 1.982672929763794, + "learning_rate": 1e-06, + "loss": 1.0787, + "mean_token_accuracy": 0.684860110282898, + "num_tokens": 139996875.0, + "step": 5495 + }, + { + "epoch": 0.6035580935646826, + "grad_norm": 2.176943302154541, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6852678656578064, + "num_tokens": 140022241.0, + "step": 5496 + }, + { + "epoch": 0.6036679112672962, + "grad_norm": 2.302485466003418, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7168571949005127, + "num_tokens": 140044398.0, + "step": 5497 + }, + { + "epoch": 0.60377772896991, + "grad_norm": 2.403425693511963, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.722980260848999, + "num_tokens": 140065273.0, + "step": 5498 + }, + { + "epoch": 0.6038875466725236, + "grad_norm": 2.1872267723083496, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7011091709136963, + "num_tokens": 140091426.0, + "step": 5499 + }, + { + "epoch": 0.6039973643751373, + "grad_norm": 2.1763994693756104, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.6871458888053894, + "num_tokens": 140117918.0, + "step": 5500 + }, + { + "epoch": 0.6041071820777509, + "grad_norm": 2.0401859283447266, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7113003730773926, + "num_tokens": 140144770.0, + "step": 5501 + }, + { + "epoch": 0.6042169997803646, + "grad_norm": 1.9911435842514038, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7126471996307373, + "num_tokens": 140174038.0, + "step": 5502 + }, + { + "epoch": 0.6043268174829782, + "grad_norm": 2.1864593029022217, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7178525924682617, + "num_tokens": 140197895.0, + "step": 5503 + }, + { + "epoch": 0.6044366351855919, + "grad_norm": 2.188223361968994, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7077963352203369, + "num_tokens": 140222515.0, + "step": 5504 + }, + { + "epoch": 0.6045464528882056, + "grad_norm": 1.9721438884735107, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6908172369003296, + "num_tokens": 140256484.0, + "step": 5505 + }, + { + "epoch": 0.6046562705908193, + "grad_norm": 2.2743732929229736, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.6946209669113159, + "num_tokens": 140282979.0, + "step": 5506 + }, + { + "epoch": 0.6047660882934329, + "grad_norm": 2.1210715770721436, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.7016081809997559, + "num_tokens": 140309831.0, + "step": 5507 + }, + { + "epoch": 0.6048759059960466, + "grad_norm": 2.2094814777374268, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7200137972831726, + "num_tokens": 140334835.0, + "step": 5508 + }, + { + "epoch": 0.6049857236986602, + "grad_norm": 2.248659372329712, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7169073820114136, + "num_tokens": 140359011.0, + "step": 5509 + }, + { + "epoch": 0.6050955414012739, + "grad_norm": 2.066352128982544, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.6964011192321777, + "num_tokens": 140386764.0, + "step": 5510 + }, + { + "epoch": 0.6052053591038875, + "grad_norm": 2.028165817260742, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7230581045150757, + "num_tokens": 140416682.0, + "step": 5511 + }, + { + "epoch": 0.6053151768065012, + "grad_norm": 2.3352839946746826, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7081775665283203, + "num_tokens": 140437997.0, + "step": 5512 + }, + { + "epoch": 0.6054249945091149, + "grad_norm": 2.1347808837890625, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6974567174911499, + "num_tokens": 140466157.0, + "step": 5513 + }, + { + "epoch": 0.6055348122117286, + "grad_norm": 2.593168258666992, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7130746841430664, + "num_tokens": 140486420.0, + "step": 5514 + }, + { + "epoch": 0.6056446299143422, + "grad_norm": 2.170175552368164, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.6850294470787048, + "num_tokens": 140515603.0, + "step": 5515 + }, + { + "epoch": 0.6057544476169558, + "grad_norm": 2.4365086555480957, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7107349634170532, + "num_tokens": 140536580.0, + "step": 5516 + }, + { + "epoch": 0.6058642653195695, + "grad_norm": 2.1695282459259033, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.6997452974319458, + "num_tokens": 140563338.0, + "step": 5517 + }, + { + "epoch": 0.6059740830221831, + "grad_norm": 2.176382064819336, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7069987654685974, + "num_tokens": 140591448.0, + "step": 5518 + }, + { + "epoch": 0.6060839007247968, + "grad_norm": 2.100940465927124, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6907133460044861, + "num_tokens": 140619636.0, + "step": 5519 + }, + { + "epoch": 0.6061937184274105, + "grad_norm": 2.270200490951538, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6985234022140503, + "num_tokens": 140648178.0, + "step": 5520 + }, + { + "epoch": 0.6063035361300242, + "grad_norm": 2.412076711654663, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.6961528062820435, + "num_tokens": 140668489.0, + "step": 5521 + }, + { + "epoch": 0.6064133538326378, + "grad_norm": 2.2201988697052, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6935158967971802, + "num_tokens": 140692960.0, + "step": 5522 + }, + { + "epoch": 0.6065231715352515, + "grad_norm": 2.5396180152893066, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7048056721687317, + "num_tokens": 140712620.0, + "step": 5523 + }, + { + "epoch": 0.6066329892378651, + "grad_norm": 1.994044303894043, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.707067608833313, + "num_tokens": 140741958.0, + "step": 5524 + }, + { + "epoch": 0.6067428069404788, + "grad_norm": 2.0106112957000732, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7065351605415344, + "num_tokens": 140773664.0, + "step": 5525 + }, + { + "epoch": 0.6068526246430924, + "grad_norm": 2.174478769302368, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6879990100860596, + "num_tokens": 140799890.0, + "step": 5526 + }, + { + "epoch": 0.6069624423457062, + "grad_norm": 2.1842844486236572, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7075890898704529, + "num_tokens": 140825329.0, + "step": 5527 + }, + { + "epoch": 0.6070722600483198, + "grad_norm": 2.4138588905334473, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6975168585777283, + "num_tokens": 140848383.0, + "step": 5528 + }, + { + "epoch": 0.6071820777509335, + "grad_norm": 2.4259767532348633, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7162262201309204, + "num_tokens": 140869144.0, + "step": 5529 + }, + { + "epoch": 0.6072918954535471, + "grad_norm": 1.9605817794799805, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.697795033454895, + "num_tokens": 140899875.0, + "step": 5530 + }, + { + "epoch": 0.6074017131561608, + "grad_norm": 2.315068006515503, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.709419846534729, + "num_tokens": 140921302.0, + "step": 5531 + }, + { + "epoch": 0.6075115308587744, + "grad_norm": 2.2285361289978027, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.7011346817016602, + "num_tokens": 140945786.0, + "step": 5532 + }, + { + "epoch": 0.607621348561388, + "grad_norm": 2.160039186477661, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.6978929042816162, + "num_tokens": 140972988.0, + "step": 5533 + }, + { + "epoch": 0.6077311662640018, + "grad_norm": 2.2818708419799805, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7082752585411072, + "num_tokens": 140995249.0, + "step": 5534 + }, + { + "epoch": 0.6078409839666155, + "grad_norm": 2.1815412044525146, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.7014960646629333, + "num_tokens": 141021522.0, + "step": 5535 + }, + { + "epoch": 0.6079508016692291, + "grad_norm": 2.190089225769043, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7078278064727783, + "num_tokens": 141049377.0, + "step": 5536 + }, + { + "epoch": 0.6080606193718427, + "grad_norm": 2.0001678466796875, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6907861232757568, + "num_tokens": 141077790.0, + "step": 5537 + }, + { + "epoch": 0.6081704370744564, + "grad_norm": 2.302434206008911, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7254981994628906, + "num_tokens": 141100061.0, + "step": 5538 + }, + { + "epoch": 0.60828025477707, + "grad_norm": 2.1568028926849365, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7200299501419067, + "num_tokens": 141129772.0, + "step": 5539 + }, + { + "epoch": 0.6083900724796837, + "grad_norm": 2.3965179920196533, + "learning_rate": 1e-06, + "loss": 1.0718, + "mean_token_accuracy": 0.6786283850669861, + "num_tokens": 141153384.0, + "step": 5540 + }, + { + "epoch": 0.6084998901822973, + "grad_norm": 1.9170255661010742, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7031134366989136, + "num_tokens": 141184300.0, + "step": 5541 + }, + { + "epoch": 0.6086097078849111, + "grad_norm": 2.0428662300109863, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6936272382736206, + "num_tokens": 141214568.0, + "step": 5542 + }, + { + "epoch": 0.6087195255875247, + "grad_norm": 2.097299098968506, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7060071229934692, + "num_tokens": 141243551.0, + "step": 5543 + }, + { + "epoch": 0.6088293432901384, + "grad_norm": 2.3630502223968506, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7153447866439819, + "num_tokens": 141266291.0, + "step": 5544 + }, + { + "epoch": 0.608939160992752, + "grad_norm": 1.942949891090393, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.703520655632019, + "num_tokens": 141297103.0, + "step": 5545 + }, + { + "epoch": 0.6090489786953657, + "grad_norm": 2.1480140686035156, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6878670454025269, + "num_tokens": 141323945.0, + "step": 5546 + }, + { + "epoch": 0.6091587963979793, + "grad_norm": 2.0945701599121094, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.7010088562965393, + "num_tokens": 141351558.0, + "step": 5547 + }, + { + "epoch": 0.609268614100593, + "grad_norm": 2.292160987854004, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.716866672039032, + "num_tokens": 141373952.0, + "step": 5548 + }, + { + "epoch": 0.6093784318032067, + "grad_norm": 2.7675857543945312, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.733308732509613, + "num_tokens": 141390035.0, + "step": 5549 + }, + { + "epoch": 0.6094882495058204, + "grad_norm": 2.262885808944702, + "learning_rate": 1e-06, + "loss": 1.0581, + "mean_token_accuracy": 0.6871997117996216, + "num_tokens": 141416062.0, + "step": 5550 + }, + { + "epoch": 0.609598067208434, + "grad_norm": 1.9737603664398193, + "learning_rate": 1e-06, + "loss": 1.1006, + "mean_token_accuracy": 0.6703625321388245, + "num_tokens": 141448261.0, + "step": 5551 + }, + { + "epoch": 0.6097078849110477, + "grad_norm": 1.9378985166549683, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6971055269241333, + "num_tokens": 141482257.0, + "step": 5552 + }, + { + "epoch": 0.6098177026136613, + "grad_norm": 2.365112781524658, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.7018377780914307, + "num_tokens": 141505951.0, + "step": 5553 + }, + { + "epoch": 0.609927520316275, + "grad_norm": 2.0188915729522705, + "learning_rate": 1e-06, + "loss": 1.0033, + "mean_token_accuracy": 0.6943620443344116, + "num_tokens": 141535791.0, + "step": 5554 + }, + { + "epoch": 0.6100373380188886, + "grad_norm": 2.236233949661255, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.6875461935997009, + "num_tokens": 141562143.0, + "step": 5555 + }, + { + "epoch": 0.6101471557215024, + "grad_norm": 2.361699104309082, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7086020708084106, + "num_tokens": 141584912.0, + "step": 5556 + }, + { + "epoch": 0.610256973424116, + "grad_norm": 2.484919548034668, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7366456389427185, + "num_tokens": 141603108.0, + "step": 5557 + }, + { + "epoch": 0.6103667911267296, + "grad_norm": 2.249619245529175, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.6949096918106079, + "num_tokens": 141629916.0, + "step": 5558 + }, + { + "epoch": 0.6104766088293433, + "grad_norm": 2.071268081665039, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6925473213195801, + "num_tokens": 141659541.0, + "step": 5559 + }, + { + "epoch": 0.6105864265319569, + "grad_norm": 2.5039615631103516, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.7102612257003784, + "num_tokens": 141682444.0, + "step": 5560 + }, + { + "epoch": 0.6106962442345706, + "grad_norm": 2.0726795196533203, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7091038227081299, + "num_tokens": 141710039.0, + "step": 5561 + }, + { + "epoch": 0.6108060619371842, + "grad_norm": 2.2601232528686523, + "learning_rate": 1e-06, + "loss": 1.0143, + "mean_token_accuracy": 0.6976659297943115, + "num_tokens": 141735729.0, + "step": 5562 + }, + { + "epoch": 0.610915879639798, + "grad_norm": 2.219809055328369, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6818291544914246, + "num_tokens": 141762019.0, + "step": 5563 + }, + { + "epoch": 0.6110256973424116, + "grad_norm": 2.3779895305633545, + "learning_rate": 1e-06, + "loss": 1.0325, + "mean_token_accuracy": 0.6926953792572021, + "num_tokens": 141785342.0, + "step": 5564 + }, + { + "epoch": 0.6111355150450253, + "grad_norm": 2.172312021255493, + "learning_rate": 1e-06, + "loss": 1.0108, + "mean_token_accuracy": 0.6917276382446289, + "num_tokens": 141810841.0, + "step": 5565 + }, + { + "epoch": 0.6112453327476389, + "grad_norm": 2.2075772285461426, + "learning_rate": 1e-06, + "loss": 1.0442, + "mean_token_accuracy": 0.6821259260177612, + "num_tokens": 141837377.0, + "step": 5566 + }, + { + "epoch": 0.6113551504502526, + "grad_norm": 2.0664496421813965, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7206273674964905, + "num_tokens": 141864334.0, + "step": 5567 + }, + { + "epoch": 0.6114649681528662, + "grad_norm": 2.351942777633667, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7178393602371216, + "num_tokens": 141887780.0, + "step": 5568 + }, + { + "epoch": 0.6115747858554799, + "grad_norm": 2.383734941482544, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7161420583724976, + "num_tokens": 141908948.0, + "step": 5569 + }, + { + "epoch": 0.6116846035580935, + "grad_norm": 1.9169690608978271, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7050769329071045, + "num_tokens": 141941560.0, + "step": 5570 + }, + { + "epoch": 0.6117944212607073, + "grad_norm": 1.8419032096862793, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.6980124711990356, + "num_tokens": 141976095.0, + "step": 5571 + }, + { + "epoch": 0.6119042389633209, + "grad_norm": 2.0686914920806885, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7297215461730957, + "num_tokens": 142004250.0, + "step": 5572 + }, + { + "epoch": 0.6120140566659346, + "grad_norm": 2.2264902591705322, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7134965062141418, + "num_tokens": 142028777.0, + "step": 5573 + }, + { + "epoch": 0.6121238743685482, + "grad_norm": 2.1402344703674316, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7033365964889526, + "num_tokens": 142055078.0, + "step": 5574 + }, + { + "epoch": 0.6122336920711619, + "grad_norm": 2.349656820297241, + "learning_rate": 1e-06, + "loss": 1.0137, + "mean_token_accuracy": 0.6919878721237183, + "num_tokens": 142079029.0, + "step": 5575 + }, + { + "epoch": 0.6123435097737755, + "grad_norm": 2.6098527908325195, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7106865644454956, + "num_tokens": 142098594.0, + "step": 5576 + }, + { + "epoch": 0.6124533274763891, + "grad_norm": 2.518767833709717, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.7002638578414917, + "num_tokens": 142119074.0, + "step": 5577 + }, + { + "epoch": 0.6125631451790029, + "grad_norm": 2.066253662109375, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7245063781738281, + "num_tokens": 142145399.0, + "step": 5578 + }, + { + "epoch": 0.6126729628816165, + "grad_norm": 2.0650312900543213, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7157739400863647, + "num_tokens": 142174802.0, + "step": 5579 + }, + { + "epoch": 0.6127827805842302, + "grad_norm": 2.2185473442077637, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6995328664779663, + "num_tokens": 142201295.0, + "step": 5580 + }, + { + "epoch": 0.6128925982868438, + "grad_norm": 1.9155769348144531, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6865518689155579, + "num_tokens": 142233800.0, + "step": 5581 + }, + { + "epoch": 0.6130024159894575, + "grad_norm": 2.140988826751709, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.6989396214485168, + "num_tokens": 142259548.0, + "step": 5582 + }, + { + "epoch": 0.6131122336920711, + "grad_norm": 2.242743968963623, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7209682464599609, + "num_tokens": 142283925.0, + "step": 5583 + }, + { + "epoch": 0.6132220513946848, + "grad_norm": 2.3258001804351807, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6906403303146362, + "num_tokens": 142306756.0, + "step": 5584 + }, + { + "epoch": 0.6133318690972985, + "grad_norm": 2.1759345531463623, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6915943622589111, + "num_tokens": 142334300.0, + "step": 5585 + }, + { + "epoch": 0.6134416867999122, + "grad_norm": 2.1605234146118164, + "learning_rate": 1e-06, + "loss": 1.0436, + "mean_token_accuracy": 0.6855208873748779, + "num_tokens": 142360561.0, + "step": 5586 + }, + { + "epoch": 0.6135515045025258, + "grad_norm": 2.0966415405273438, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7128651142120361, + "num_tokens": 142387874.0, + "step": 5587 + }, + { + "epoch": 0.6136613222051395, + "grad_norm": 2.0324783325195312, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.7057268619537354, + "num_tokens": 142417186.0, + "step": 5588 + }, + { + "epoch": 0.6137711399077531, + "grad_norm": 2.1467225551605225, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7074463963508606, + "num_tokens": 142442473.0, + "step": 5589 + }, + { + "epoch": 0.6138809576103668, + "grad_norm": 2.235095500946045, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7291721701622009, + "num_tokens": 142466597.0, + "step": 5590 + }, + { + "epoch": 0.6139907753129804, + "grad_norm": 2.3385603427886963, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7123425006866455, + "num_tokens": 142488065.0, + "step": 5591 + }, + { + "epoch": 0.6141005930155942, + "grad_norm": 2.2054991722106934, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.709539532661438, + "num_tokens": 142512715.0, + "step": 5592 + }, + { + "epoch": 0.6142104107182078, + "grad_norm": 2.145944595336914, + "learning_rate": 1e-06, + "loss": 1.0314, + "mean_token_accuracy": 0.6942781209945679, + "num_tokens": 142538866.0, + "step": 5593 + }, + { + "epoch": 0.6143202284208215, + "grad_norm": 2.40620493888855, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.719612181186676, + "num_tokens": 142560028.0, + "step": 5594 + }, + { + "epoch": 0.6144300461234351, + "grad_norm": 2.340237855911255, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6967999339103699, + "num_tokens": 142584620.0, + "step": 5595 + }, + { + "epoch": 0.6145398638260487, + "grad_norm": 2.1785950660705566, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6893120408058167, + "num_tokens": 142611005.0, + "step": 5596 + }, + { + "epoch": 0.6146496815286624, + "grad_norm": 2.2716357707977295, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7188133597373962, + "num_tokens": 142633276.0, + "step": 5597 + }, + { + "epoch": 0.614759499231276, + "grad_norm": 2.262587547302246, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7328786849975586, + "num_tokens": 142658392.0, + "step": 5598 + }, + { + "epoch": 0.6148693169338898, + "grad_norm": 2.593937397003174, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7104705572128296, + "num_tokens": 142676435.0, + "step": 5599 + }, + { + "epoch": 0.6149791346365034, + "grad_norm": 2.0389280319213867, + "learning_rate": 1e-06, + "loss": 1.0631, + "mean_token_accuracy": 0.6898400783538818, + "num_tokens": 142705988.0, + "step": 5600 + }, + { + "epoch": 0.6150889523391171, + "grad_norm": 2.0194222927093506, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7216475605964661, + "num_tokens": 142733368.0, + "step": 5601 + }, + { + "epoch": 0.6151987700417307, + "grad_norm": 2.3220369815826416, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7152585387229919, + "num_tokens": 142756587.0, + "step": 5602 + }, + { + "epoch": 0.6153085877443444, + "grad_norm": 2.598020315170288, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7180643677711487, + "num_tokens": 142774990.0, + "step": 5603 + }, + { + "epoch": 0.615418405446958, + "grad_norm": 2.50313663482666, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6926705837249756, + "num_tokens": 142797142.0, + "step": 5604 + }, + { + "epoch": 0.6155282231495717, + "grad_norm": 2.1629903316497803, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6834799647331238, + "num_tokens": 142825025.0, + "step": 5605 + }, + { + "epoch": 0.6156380408521853, + "grad_norm": 2.0175652503967285, + "learning_rate": 1e-06, + "loss": 1.0576, + "mean_token_accuracy": 0.6866899728775024, + "num_tokens": 142854871.0, + "step": 5606 + }, + { + "epoch": 0.6157478585547991, + "grad_norm": 2.2993850708007812, + "learning_rate": 1e-06, + "loss": 1.0163, + "mean_token_accuracy": 0.6920313835144043, + "num_tokens": 142878459.0, + "step": 5607 + }, + { + "epoch": 0.6158576762574127, + "grad_norm": 2.3625900745391846, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7100505828857422, + "num_tokens": 142900243.0, + "step": 5608 + }, + { + "epoch": 0.6159674939600264, + "grad_norm": 2.556516647338867, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7085026502609253, + "num_tokens": 142918414.0, + "step": 5609 + }, + { + "epoch": 0.61607731166264, + "grad_norm": 2.155726909637451, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7188338041305542, + "num_tokens": 142944213.0, + "step": 5610 + }, + { + "epoch": 0.6161871293652537, + "grad_norm": 2.302210569381714, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7046279907226562, + "num_tokens": 142969380.0, + "step": 5611 + }, + { + "epoch": 0.6162969470678673, + "grad_norm": 2.350466251373291, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6947985291481018, + "num_tokens": 142992623.0, + "step": 5612 + }, + { + "epoch": 0.616406764770481, + "grad_norm": 2.315281391143799, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.729710578918457, + "num_tokens": 143014962.0, + "step": 5613 + }, + { + "epoch": 0.6165165824730947, + "grad_norm": 2.1345770359039307, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7222793102264404, + "num_tokens": 143039455.0, + "step": 5614 + }, + { + "epoch": 0.6166264001757084, + "grad_norm": 2.291539430618286, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7033931016921997, + "num_tokens": 143064685.0, + "step": 5615 + }, + { + "epoch": 0.616736217878322, + "grad_norm": 2.115337371826172, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6912091970443726, + "num_tokens": 143091196.0, + "step": 5616 + }, + { + "epoch": 0.6168460355809356, + "grad_norm": 2.053438425064087, + "learning_rate": 1e-06, + "loss": 1.0693, + "mean_token_accuracy": 0.677111029624939, + "num_tokens": 143120627.0, + "step": 5617 + }, + { + "epoch": 0.6169558532835493, + "grad_norm": 2.1619160175323486, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7103955149650574, + "num_tokens": 143144657.0, + "step": 5618 + }, + { + "epoch": 0.6170656709861629, + "grad_norm": 1.9436653852462769, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.6949117183685303, + "num_tokens": 143177728.0, + "step": 5619 + }, + { + "epoch": 0.6171754886887766, + "grad_norm": 2.364797353744507, + "learning_rate": 1e-06, + "loss": 1.0428, + "mean_token_accuracy": 0.684333086013794, + "num_tokens": 143203130.0, + "step": 5620 + }, + { + "epoch": 0.6172853063913903, + "grad_norm": 1.9779131412506104, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.707956850528717, + "num_tokens": 143232763.0, + "step": 5621 + }, + { + "epoch": 0.617395124094004, + "grad_norm": 2.059701681137085, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7025138139724731, + "num_tokens": 143261156.0, + "step": 5622 + }, + { + "epoch": 0.6175049417966176, + "grad_norm": 2.1099932193756104, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7088605165481567, + "num_tokens": 143288305.0, + "step": 5623 + }, + { + "epoch": 0.6176147594992313, + "grad_norm": 2.2065088748931885, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7114928960800171, + "num_tokens": 143313319.0, + "step": 5624 + }, + { + "epoch": 0.6177245772018449, + "grad_norm": 2.02247953414917, + "learning_rate": 1e-06, + "loss": 1.0731, + "mean_token_accuracy": 0.6770888566970825, + "num_tokens": 143343121.0, + "step": 5625 + }, + { + "epoch": 0.6178343949044586, + "grad_norm": 2.3952856063842773, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7063243389129639, + "num_tokens": 143363021.0, + "step": 5626 + }, + { + "epoch": 0.6179442126070722, + "grad_norm": 2.1975269317626953, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.6977739334106445, + "num_tokens": 143389634.0, + "step": 5627 + }, + { + "epoch": 0.618054030309686, + "grad_norm": 2.3003475666046143, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7259299755096436, + "num_tokens": 143414073.0, + "step": 5628 + }, + { + "epoch": 0.6181638480122996, + "grad_norm": 2.327561616897583, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7159342765808105, + "num_tokens": 143439139.0, + "step": 5629 + }, + { + "epoch": 0.6182736657149133, + "grad_norm": 2.14284086227417, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7043335437774658, + "num_tokens": 143465602.0, + "step": 5630 + }, + { + "epoch": 0.6183834834175269, + "grad_norm": 2.2421343326568604, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7067865133285522, + "num_tokens": 143489720.0, + "step": 5631 + }, + { + "epoch": 0.6184933011201406, + "grad_norm": 1.9843040704727173, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.6903477907180786, + "num_tokens": 143520847.0, + "step": 5632 + }, + { + "epoch": 0.6186031188227542, + "grad_norm": 1.9094592332839966, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7090310454368591, + "num_tokens": 143554700.0, + "step": 5633 + }, + { + "epoch": 0.6187129365253679, + "grad_norm": 2.238913059234619, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.7026989459991455, + "num_tokens": 143578498.0, + "step": 5634 + }, + { + "epoch": 0.6188227542279815, + "grad_norm": 2.1507575511932373, + "learning_rate": 1e-06, + "loss": 1.0671, + "mean_token_accuracy": 0.688010573387146, + "num_tokens": 143604617.0, + "step": 5635 + }, + { + "epoch": 0.6189325719305953, + "grad_norm": 2.352888345718384, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7119840383529663, + "num_tokens": 143626944.0, + "step": 5636 + }, + { + "epoch": 0.6190423896332089, + "grad_norm": 2.0936033725738525, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.72182297706604, + "num_tokens": 143653625.0, + "step": 5637 + }, + { + "epoch": 0.6191522073358225, + "grad_norm": 2.193474292755127, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.6994365453720093, + "num_tokens": 143679382.0, + "step": 5638 + }, + { + "epoch": 0.6192620250384362, + "grad_norm": 2.352179765701294, + "learning_rate": 1e-06, + "loss": 1.0387, + "mean_token_accuracy": 0.6842137575149536, + "num_tokens": 143708735.0, + "step": 5639 + }, + { + "epoch": 0.6193718427410498, + "grad_norm": 2.2593820095062256, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.705875813961029, + "num_tokens": 143734067.0, + "step": 5640 + }, + { + "epoch": 0.6194816604436635, + "grad_norm": 2.099102020263672, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7057912945747375, + "num_tokens": 143760451.0, + "step": 5641 + }, + { + "epoch": 0.6195914781462771, + "grad_norm": 2.4903626441955566, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6981709599494934, + "num_tokens": 143784071.0, + "step": 5642 + }, + { + "epoch": 0.6197012958488909, + "grad_norm": 2.7029857635498047, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7117706537246704, + "num_tokens": 143802444.0, + "step": 5643 + }, + { + "epoch": 0.6198111135515045, + "grad_norm": 2.174565553665161, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.697053849697113, + "num_tokens": 143829592.0, + "step": 5644 + }, + { + "epoch": 0.6199209312541182, + "grad_norm": 2.276137590408325, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6860764026641846, + "num_tokens": 143853910.0, + "step": 5645 + }, + { + "epoch": 0.6200307489567318, + "grad_norm": 2.366255283355713, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7026934623718262, + "num_tokens": 143877722.0, + "step": 5646 + }, + { + "epoch": 0.6201405666593455, + "grad_norm": 2.619828939437866, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7316714525222778, + "num_tokens": 143896110.0, + "step": 5647 + }, + { + "epoch": 0.6202503843619591, + "grad_norm": 2.122143268585205, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6998755931854248, + "num_tokens": 143922743.0, + "step": 5648 + }, + { + "epoch": 0.6203602020645728, + "grad_norm": 2.407010555267334, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7118289470672607, + "num_tokens": 143944728.0, + "step": 5649 + }, + { + "epoch": 0.6204700197671865, + "grad_norm": 2.521634817123413, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7429078221321106, + "num_tokens": 143965034.0, + "step": 5650 + }, + { + "epoch": 0.6205798374698002, + "grad_norm": 2.2388224601745605, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7224660515785217, + "num_tokens": 143988541.0, + "step": 5651 + }, + { + "epoch": 0.6206896551724138, + "grad_norm": 2.208617687225342, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7205063104629517, + "num_tokens": 144012139.0, + "step": 5652 + }, + { + "epoch": 0.6207994728750275, + "grad_norm": 2.0878891944885254, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.7009992599487305, + "num_tokens": 144041192.0, + "step": 5653 + }, + { + "epoch": 0.6209092905776411, + "grad_norm": 2.18235182762146, + "learning_rate": 1e-06, + "loss": 1.0306, + "mean_token_accuracy": 0.6879208087921143, + "num_tokens": 144067319.0, + "step": 5654 + }, + { + "epoch": 0.6210191082802548, + "grad_norm": 2.0372326374053955, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7071213722229004, + "num_tokens": 144095333.0, + "step": 5655 + }, + { + "epoch": 0.6211289259828684, + "grad_norm": 2.250792980194092, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.7053754329681396, + "num_tokens": 144118814.0, + "step": 5656 + }, + { + "epoch": 0.6212387436854822, + "grad_norm": 2.3761227130889893, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6934857368469238, + "num_tokens": 144141905.0, + "step": 5657 + }, + { + "epoch": 0.6213485613880958, + "grad_norm": 2.494157075881958, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7117758989334106, + "num_tokens": 144162531.0, + "step": 5658 + }, + { + "epoch": 0.6214583790907094, + "grad_norm": 2.064832925796509, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7149670124053955, + "num_tokens": 144190028.0, + "step": 5659 + }, + { + "epoch": 0.6215681967933231, + "grad_norm": 2.478316307067871, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.7460850477218628, + "num_tokens": 144209114.0, + "step": 5660 + }, + { + "epoch": 0.6216780144959367, + "grad_norm": 2.351719856262207, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7314745187759399, + "num_tokens": 144230736.0, + "step": 5661 + }, + { + "epoch": 0.6217878321985504, + "grad_norm": 2.6149895191192627, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.722831130027771, + "num_tokens": 144248782.0, + "step": 5662 + }, + { + "epoch": 0.621897649901164, + "grad_norm": 2.242244243621826, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7135632038116455, + "num_tokens": 144272194.0, + "step": 5663 + }, + { + "epoch": 0.6220074676037777, + "grad_norm": 1.866190791130066, + "learning_rate": 1e-06, + "loss": 1.0335, + "mean_token_accuracy": 0.6898748874664307, + "num_tokens": 144305533.0, + "step": 5664 + }, + { + "epoch": 0.6221172853063914, + "grad_norm": 2.4646518230438232, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6963580250740051, + "num_tokens": 144328483.0, + "step": 5665 + }, + { + "epoch": 0.6222271030090051, + "grad_norm": 2.273569107055664, + "learning_rate": 1e-06, + "loss": 1.0225, + "mean_token_accuracy": 0.7076359987258911, + "num_tokens": 144352874.0, + "step": 5666 + }, + { + "epoch": 0.6223369207116187, + "grad_norm": 2.0497641563415527, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.722277045249939, + "num_tokens": 144379500.0, + "step": 5667 + }, + { + "epoch": 0.6224467384142324, + "grad_norm": 2.407576322555542, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7112959623336792, + "num_tokens": 144400789.0, + "step": 5668 + }, + { + "epoch": 0.622556556116846, + "grad_norm": 2.05778431892395, + "learning_rate": 1e-06, + "loss": 1.0733, + "mean_token_accuracy": 0.6871352791786194, + "num_tokens": 144430214.0, + "step": 5669 + }, + { + "epoch": 0.6226663738194597, + "grad_norm": 2.1421408653259277, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7224634885787964, + "num_tokens": 144455632.0, + "step": 5670 + }, + { + "epoch": 0.6227761915220733, + "grad_norm": 2.219940423965454, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7264620065689087, + "num_tokens": 144479336.0, + "step": 5671 + }, + { + "epoch": 0.6228860092246871, + "grad_norm": 2.2179887294769287, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7081941962242126, + "num_tokens": 144502155.0, + "step": 5672 + }, + { + "epoch": 0.6229958269273007, + "grad_norm": 2.367777109146118, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6915444135665894, + "num_tokens": 144525851.0, + "step": 5673 + }, + { + "epoch": 0.6231056446299144, + "grad_norm": 2.6529488563537598, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7292223572731018, + "num_tokens": 144544545.0, + "step": 5674 + }, + { + "epoch": 0.623215462332528, + "grad_norm": 2.171537399291992, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.6993533968925476, + "num_tokens": 144569209.0, + "step": 5675 + }, + { + "epoch": 0.6233252800351416, + "grad_norm": 2.3240034580230713, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6811892986297607, + "num_tokens": 144594774.0, + "step": 5676 + }, + { + "epoch": 0.6234350977377553, + "grad_norm": 1.9896522760391235, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7077112197875977, + "num_tokens": 144626082.0, + "step": 5677 + }, + { + "epoch": 0.6235449154403689, + "grad_norm": 2.1823019981384277, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7200006246566772, + "num_tokens": 144651209.0, + "step": 5678 + }, + { + "epoch": 0.6236547331429827, + "grad_norm": 2.323892831802368, + "learning_rate": 1e-06, + "loss": 1.0494, + "mean_token_accuracy": 0.6984842419624329, + "num_tokens": 144676024.0, + "step": 5679 + }, + { + "epoch": 0.6237645508455963, + "grad_norm": 2.3371598720550537, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7016963362693787, + "num_tokens": 144699226.0, + "step": 5680 + }, + { + "epoch": 0.62387436854821, + "grad_norm": 2.136746883392334, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7115715146064758, + "num_tokens": 144726918.0, + "step": 5681 + }, + { + "epoch": 0.6239841862508236, + "grad_norm": 2.587484121322632, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7202510237693787, + "num_tokens": 144747628.0, + "step": 5682 + }, + { + "epoch": 0.6240940039534373, + "grad_norm": 2.011855363845825, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7070253491401672, + "num_tokens": 144776485.0, + "step": 5683 + }, + { + "epoch": 0.6242038216560509, + "grad_norm": 1.924848198890686, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7053002119064331, + "num_tokens": 144807994.0, + "step": 5684 + }, + { + "epoch": 0.6243136393586646, + "grad_norm": 2.0802719593048096, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7164150476455688, + "num_tokens": 144833832.0, + "step": 5685 + }, + { + "epoch": 0.6244234570612783, + "grad_norm": 2.286850929260254, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7194024324417114, + "num_tokens": 144856379.0, + "step": 5686 + }, + { + "epoch": 0.624533274763892, + "grad_norm": 2.351027727127075, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7052811980247498, + "num_tokens": 144878890.0, + "step": 5687 + }, + { + "epoch": 0.6246430924665056, + "grad_norm": 2.3812146186828613, + "learning_rate": 1e-06, + "loss": 1.0311, + "mean_token_accuracy": 0.6828894019126892, + "num_tokens": 144901923.0, + "step": 5688 + }, + { + "epoch": 0.6247529101691193, + "grad_norm": 2.3599815368652344, + "learning_rate": 1e-06, + "loss": 1.0571, + "mean_token_accuracy": 0.6794494390487671, + "num_tokens": 144927472.0, + "step": 5689 + }, + { + "epoch": 0.6248627278717329, + "grad_norm": 2.1371445655822754, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7039594650268555, + "num_tokens": 144952015.0, + "step": 5690 + }, + { + "epoch": 0.6249725455743466, + "grad_norm": 2.30903697013855, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.6889580488204956, + "num_tokens": 144976967.0, + "step": 5691 + }, + { + "epoch": 0.6250823632769602, + "grad_norm": 2.2136752605438232, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7092918157577515, + "num_tokens": 145002486.0, + "step": 5692 + }, + { + "epoch": 0.6251921809795739, + "grad_norm": 2.049518346786499, + "learning_rate": 1e-06, + "loss": 1.0567, + "mean_token_accuracy": 0.6978312730789185, + "num_tokens": 145032987.0, + "step": 5693 + }, + { + "epoch": 0.6253019986821876, + "grad_norm": 2.022063970565796, + "learning_rate": 1e-06, + "loss": 1.0331, + "mean_token_accuracy": 0.6947914958000183, + "num_tokens": 145063179.0, + "step": 5694 + }, + { + "epoch": 0.6254118163848013, + "grad_norm": 2.281360149383545, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7275307774543762, + "num_tokens": 145087774.0, + "step": 5695 + }, + { + "epoch": 0.6255216340874149, + "grad_norm": 2.2563018798828125, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7049573659896851, + "num_tokens": 145110236.0, + "step": 5696 + }, + { + "epoch": 0.6256314517900285, + "grad_norm": 2.1883084774017334, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7064101696014404, + "num_tokens": 145135963.0, + "step": 5697 + }, + { + "epoch": 0.6257412694926422, + "grad_norm": 2.347496509552002, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.6979056596755981, + "num_tokens": 145159481.0, + "step": 5698 + }, + { + "epoch": 0.6258510871952558, + "grad_norm": 2.556992769241333, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7150817513465881, + "num_tokens": 145179713.0, + "step": 5699 + }, + { + "epoch": 0.6259609048978695, + "grad_norm": 2.0971198081970215, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7038260698318481, + "num_tokens": 145207296.0, + "step": 5700 + }, + { + "epoch": 0.6260707226004832, + "grad_norm": 2.172128200531006, + "learning_rate": 1e-06, + "loss": 0.9829, + "mean_token_accuracy": 0.6957036256790161, + "num_tokens": 145234228.0, + "step": 5701 + }, + { + "epoch": 0.6261805403030969, + "grad_norm": 2.2671267986297607, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6875087022781372, + "num_tokens": 145260849.0, + "step": 5702 + }, + { + "epoch": 0.6262903580057105, + "grad_norm": 1.8950519561767578, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6848200559616089, + "num_tokens": 145294087.0, + "step": 5703 + }, + { + "epoch": 0.6264001757083242, + "grad_norm": 2.17387318611145, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6915873885154724, + "num_tokens": 145325318.0, + "step": 5704 + }, + { + "epoch": 0.6265099934109378, + "grad_norm": 2.3239996433258057, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7055370807647705, + "num_tokens": 145350598.0, + "step": 5705 + }, + { + "epoch": 0.6266198111135515, + "grad_norm": 2.323256731033325, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7141473293304443, + "num_tokens": 145373346.0, + "step": 5706 + }, + { + "epoch": 0.6267296288161651, + "grad_norm": 2.118326187133789, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7351179122924805, + "num_tokens": 145398124.0, + "step": 5707 + }, + { + "epoch": 0.6268394465187789, + "grad_norm": 2.2305338382720947, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6956861019134521, + "num_tokens": 145423059.0, + "step": 5708 + }, + { + "epoch": 0.6269492642213925, + "grad_norm": 2.141702175140381, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6934831738471985, + "num_tokens": 145448898.0, + "step": 5709 + }, + { + "epoch": 0.6270590819240062, + "grad_norm": 2.196239471435547, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7073763608932495, + "num_tokens": 145474326.0, + "step": 5710 + }, + { + "epoch": 0.6271688996266198, + "grad_norm": 2.2002503871917725, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.6983394622802734, + "num_tokens": 145497381.0, + "step": 5711 + }, + { + "epoch": 0.6272787173292335, + "grad_norm": 2.1538379192352295, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.7029963731765747, + "num_tokens": 145522729.0, + "step": 5712 + }, + { + "epoch": 0.6273885350318471, + "grad_norm": 2.277031898498535, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7079298496246338, + "num_tokens": 145545902.0, + "step": 5713 + }, + { + "epoch": 0.6274983527344608, + "grad_norm": 2.101586103439331, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.728614330291748, + "num_tokens": 145573211.0, + "step": 5714 + }, + { + "epoch": 0.6276081704370745, + "grad_norm": 2.3974294662475586, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.6971648931503296, + "num_tokens": 145595109.0, + "step": 5715 + }, + { + "epoch": 0.6277179881396882, + "grad_norm": 2.1332414150238037, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7222422957420349, + "num_tokens": 145619786.0, + "step": 5716 + }, + { + "epoch": 0.6278278058423018, + "grad_norm": 2.3523504734039307, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6970539689064026, + "num_tokens": 145646880.0, + "step": 5717 + }, + { + "epoch": 0.6279376235449154, + "grad_norm": 2.1875765323638916, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7096834182739258, + "num_tokens": 145672951.0, + "step": 5718 + }, + { + "epoch": 0.6280474412475291, + "grad_norm": 2.589109182357788, + "learning_rate": 1e-06, + "loss": 1.0066, + "mean_token_accuracy": 0.6994860172271729, + "num_tokens": 145694102.0, + "step": 5719 + }, + { + "epoch": 0.6281572589501427, + "grad_norm": 2.1826953887939453, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6939328908920288, + "num_tokens": 145720537.0, + "step": 5720 + }, + { + "epoch": 0.6282670766527564, + "grad_norm": 2.307833671569824, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.692605197429657, + "num_tokens": 145743511.0, + "step": 5721 + }, + { + "epoch": 0.62837689435537, + "grad_norm": 2.0926461219787598, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7305632829666138, + "num_tokens": 145766763.0, + "step": 5722 + }, + { + "epoch": 0.6284867120579838, + "grad_norm": 2.3528034687042236, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6988070607185364, + "num_tokens": 145789977.0, + "step": 5723 + }, + { + "epoch": 0.6285965297605974, + "grad_norm": 2.0010883808135986, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6916689872741699, + "num_tokens": 145820500.0, + "step": 5724 + }, + { + "epoch": 0.6287063474632111, + "grad_norm": 2.4789109230041504, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7144449353218079, + "num_tokens": 145843522.0, + "step": 5725 + }, + { + "epoch": 0.6288161651658247, + "grad_norm": 2.3814306259155273, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.695691704750061, + "num_tokens": 145868492.0, + "step": 5726 + }, + { + "epoch": 0.6289259828684384, + "grad_norm": 2.32218861579895, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7141790986061096, + "num_tokens": 145890528.0, + "step": 5727 + }, + { + "epoch": 0.629035800571052, + "grad_norm": 1.995998501777649, + "learning_rate": 1e-06, + "loss": 1.0606, + "mean_token_accuracy": 0.6735684275627136, + "num_tokens": 145922329.0, + "step": 5728 + }, + { + "epoch": 0.6291456182736657, + "grad_norm": 2.20792293548584, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.692456841468811, + "num_tokens": 145947157.0, + "step": 5729 + }, + { + "epoch": 0.6292554359762794, + "grad_norm": 2.4288599491119385, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7185366153717041, + "num_tokens": 145967988.0, + "step": 5730 + }, + { + "epoch": 0.6293652536788931, + "grad_norm": 2.3722007274627686, + "learning_rate": 1e-06, + "loss": 1.0673, + "mean_token_accuracy": 0.6785464286804199, + "num_tokens": 145992769.0, + "step": 5731 + }, + { + "epoch": 0.6294750713815067, + "grad_norm": 2.2474374771118164, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7171932458877563, + "num_tokens": 146016469.0, + "step": 5732 + }, + { + "epoch": 0.6295848890841204, + "grad_norm": 2.534341812133789, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7118890285491943, + "num_tokens": 146037729.0, + "step": 5733 + }, + { + "epoch": 0.629694706786734, + "grad_norm": 2.2020411491394043, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7063878774642944, + "num_tokens": 146064542.0, + "step": 5734 + }, + { + "epoch": 0.6298045244893477, + "grad_norm": 2.241697072982788, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7068666219711304, + "num_tokens": 146089721.0, + "step": 5735 + }, + { + "epoch": 0.6299143421919613, + "grad_norm": 1.9489898681640625, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6950188875198364, + "num_tokens": 146121213.0, + "step": 5736 + }, + { + "epoch": 0.630024159894575, + "grad_norm": 2.392788887023926, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7058055996894836, + "num_tokens": 146142545.0, + "step": 5737 + }, + { + "epoch": 0.6301339775971887, + "grad_norm": 2.370918035507202, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7174044847488403, + "num_tokens": 146164141.0, + "step": 5738 + }, + { + "epoch": 0.6302437952998023, + "grad_norm": 2.0415756702423096, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7140148878097534, + "num_tokens": 146191088.0, + "step": 5739 + }, + { + "epoch": 0.630353613002416, + "grad_norm": 2.3853561878204346, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7351889610290527, + "num_tokens": 146211423.0, + "step": 5740 + }, + { + "epoch": 0.6304634307050296, + "grad_norm": 2.1684577465057373, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7118204832077026, + "num_tokens": 146236660.0, + "step": 5741 + }, + { + "epoch": 0.6305732484076433, + "grad_norm": 2.246307849884033, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.6969620585441589, + "num_tokens": 146260066.0, + "step": 5742 + }, + { + "epoch": 0.6306830661102569, + "grad_norm": 2.72743558883667, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7107174396514893, + "num_tokens": 146276888.0, + "step": 5743 + }, + { + "epoch": 0.6307928838128707, + "grad_norm": 1.8685551881790161, + "learning_rate": 1e-06, + "loss": 1.0418, + "mean_token_accuracy": 0.6851693391799927, + "num_tokens": 146308787.0, + "step": 5744 + }, + { + "epoch": 0.6309027015154843, + "grad_norm": 2.1700875759124756, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7029160261154175, + "num_tokens": 146335131.0, + "step": 5745 + }, + { + "epoch": 0.631012519218098, + "grad_norm": 2.120795726776123, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.7042006254196167, + "num_tokens": 146360420.0, + "step": 5746 + }, + { + "epoch": 0.6311223369207116, + "grad_norm": 2.057440996170044, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7041350603103638, + "num_tokens": 146387257.0, + "step": 5747 + }, + { + "epoch": 0.6312321546233253, + "grad_norm": 2.369138479232788, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.69817054271698, + "num_tokens": 146409627.0, + "step": 5748 + }, + { + "epoch": 0.6313419723259389, + "grad_norm": 2.1146934032440186, + "learning_rate": 1e-06, + "loss": 1.0637, + "mean_token_accuracy": 0.6826274394989014, + "num_tokens": 146438676.0, + "step": 5749 + }, + { + "epoch": 0.6314517900285526, + "grad_norm": 2.1650314331054688, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7148367166519165, + "num_tokens": 146465004.0, + "step": 5750 + }, + { + "epoch": 0.6315616077311663, + "grad_norm": 2.2044875621795654, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7092299461364746, + "num_tokens": 146488807.0, + "step": 5751 + }, + { + "epoch": 0.63167142543378, + "grad_norm": 2.036874532699585, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.716788649559021, + "num_tokens": 146518766.0, + "step": 5752 + }, + { + "epoch": 0.6317812431363936, + "grad_norm": 2.562220811843872, + "learning_rate": 1e-06, + "loss": 0.7369, + "mean_token_accuracy": 0.765059232711792, + "num_tokens": 146535764.0, + "step": 5753 + }, + { + "epoch": 0.6318910608390073, + "grad_norm": 2.4026846885681152, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7135248184204102, + "num_tokens": 146558611.0, + "step": 5754 + }, + { + "epoch": 0.6320008785416209, + "grad_norm": 2.277702569961548, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.6895118951797485, + "num_tokens": 146581916.0, + "step": 5755 + }, + { + "epoch": 0.6321106962442345, + "grad_norm": 2.5128021240234375, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.697310745716095, + "num_tokens": 146602360.0, + "step": 5756 + }, + { + "epoch": 0.6322205139468482, + "grad_norm": 2.18983793258667, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7018202543258667, + "num_tokens": 146627066.0, + "step": 5757 + }, + { + "epoch": 0.6323303316494618, + "grad_norm": 2.289436101913452, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.7064049243927002, + "num_tokens": 146652068.0, + "step": 5758 + }, + { + "epoch": 0.6324401493520756, + "grad_norm": 2.1688449382781982, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7130312323570251, + "num_tokens": 146674924.0, + "step": 5759 + }, + { + "epoch": 0.6325499670546892, + "grad_norm": 2.135188341140747, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.701088547706604, + "num_tokens": 146701171.0, + "step": 5760 + }, + { + "epoch": 0.6326597847573029, + "grad_norm": 2.088123083114624, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7035472989082336, + "num_tokens": 146727199.0, + "step": 5761 + }, + { + "epoch": 0.6327696024599165, + "grad_norm": 2.5525143146514893, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7303178906440735, + "num_tokens": 146745594.0, + "step": 5762 + }, + { + "epoch": 0.6328794201625302, + "grad_norm": 2.214073419570923, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7055806517601013, + "num_tokens": 146769411.0, + "step": 5763 + }, + { + "epoch": 0.6329892378651438, + "grad_norm": 2.444857597351074, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7331554889678955, + "num_tokens": 146789642.0, + "step": 5764 + }, + { + "epoch": 0.6330990555677575, + "grad_norm": 2.165710687637329, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.7144606113433838, + "num_tokens": 146814099.0, + "step": 5765 + }, + { + "epoch": 0.6332088732703712, + "grad_norm": 2.0426411628723145, + "learning_rate": 1e-06, + "loss": 1.081, + "mean_token_accuracy": 0.6772348284721375, + "num_tokens": 146841946.0, + "step": 5766 + }, + { + "epoch": 0.6333186909729849, + "grad_norm": 2.159862756729126, + "learning_rate": 1e-06, + "loss": 1.0671, + "mean_token_accuracy": 0.6797354817390442, + "num_tokens": 146869043.0, + "step": 5767 + }, + { + "epoch": 0.6334285086755985, + "grad_norm": 1.9782637357711792, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7077153921127319, + "num_tokens": 146897632.0, + "step": 5768 + }, + { + "epoch": 0.6335383263782122, + "grad_norm": 1.9964958429336548, + "learning_rate": 1e-06, + "loss": 0.9771, + "mean_token_accuracy": 0.7038236856460571, + "num_tokens": 146927301.0, + "step": 5769 + }, + { + "epoch": 0.6336481440808258, + "grad_norm": 2.2242207527160645, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6909801959991455, + "num_tokens": 146951571.0, + "step": 5770 + }, + { + "epoch": 0.6337579617834395, + "grad_norm": 2.322396755218506, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7054998874664307, + "num_tokens": 146974912.0, + "step": 5771 + }, + { + "epoch": 0.6338677794860531, + "grad_norm": 2.211188316345215, + "learning_rate": 1e-06, + "loss": 1.055, + "mean_token_accuracy": 0.6839674115180969, + "num_tokens": 146999754.0, + "step": 5772 + }, + { + "epoch": 0.6339775971886669, + "grad_norm": 2.448369264602661, + "learning_rate": 1e-06, + "loss": 1.0516, + "mean_token_accuracy": 0.6963121294975281, + "num_tokens": 147020769.0, + "step": 5773 + }, + { + "epoch": 0.6340874148912805, + "grad_norm": 2.203307628631592, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7179271578788757, + "num_tokens": 147043987.0, + "step": 5774 + }, + { + "epoch": 0.6341972325938942, + "grad_norm": 2.2823588848114014, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7009086608886719, + "num_tokens": 147067443.0, + "step": 5775 + }, + { + "epoch": 0.6343070502965078, + "grad_norm": 2.117140531539917, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7052534818649292, + "num_tokens": 147095231.0, + "step": 5776 + }, + { + "epoch": 0.6344168679991214, + "grad_norm": 2.271826982498169, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7059345245361328, + "num_tokens": 147117753.0, + "step": 5777 + }, + { + "epoch": 0.6345266857017351, + "grad_norm": 2.5150279998779297, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7276429533958435, + "num_tokens": 147137232.0, + "step": 5778 + }, + { + "epoch": 0.6346365034043487, + "grad_norm": 2.3710854053497314, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.71189945936203, + "num_tokens": 147159315.0, + "step": 5779 + }, + { + "epoch": 0.6347463211069625, + "grad_norm": 2.2525713443756104, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.7003774642944336, + "num_tokens": 147183706.0, + "step": 5780 + }, + { + "epoch": 0.6348561388095761, + "grad_norm": 2.4088668823242188, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7173891663551331, + "num_tokens": 147204928.0, + "step": 5781 + }, + { + "epoch": 0.6349659565121898, + "grad_norm": 2.2616918087005615, + "learning_rate": 1e-06, + "loss": 1.0278, + "mean_token_accuracy": 0.6925010085105896, + "num_tokens": 147229923.0, + "step": 5782 + }, + { + "epoch": 0.6350757742148034, + "grad_norm": 2.4288575649261475, + "learning_rate": 1e-06, + "loss": 1.0702, + "mean_token_accuracy": 0.6822161674499512, + "num_tokens": 147252019.0, + "step": 5783 + }, + { + "epoch": 0.6351855919174171, + "grad_norm": 2.050692558288574, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6934598088264465, + "num_tokens": 147280327.0, + "step": 5784 + }, + { + "epoch": 0.6352954096200307, + "grad_norm": 2.2309234142303467, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7198212146759033, + "num_tokens": 147302777.0, + "step": 5785 + }, + { + "epoch": 0.6354052273226444, + "grad_norm": 2.091627359390259, + "learning_rate": 1e-06, + "loss": 1.0261, + "mean_token_accuracy": 0.6923023462295532, + "num_tokens": 147332042.0, + "step": 5786 + }, + { + "epoch": 0.635515045025258, + "grad_norm": 1.8705854415893555, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6966499090194702, + "num_tokens": 147367912.0, + "step": 5787 + }, + { + "epoch": 0.6356248627278718, + "grad_norm": 2.093657970428467, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.702931821346283, + "num_tokens": 147395517.0, + "step": 5788 + }, + { + "epoch": 0.6357346804304854, + "grad_norm": 2.2667765617370605, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7071758508682251, + "num_tokens": 147420090.0, + "step": 5789 + }, + { + "epoch": 0.6358444981330991, + "grad_norm": 2.076629161834717, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7081927061080933, + "num_tokens": 147448679.0, + "step": 5790 + }, + { + "epoch": 0.6359543158357127, + "grad_norm": 2.3026630878448486, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7323240637779236, + "num_tokens": 147470458.0, + "step": 5791 + }, + { + "epoch": 0.6360641335383264, + "grad_norm": 2.280006170272827, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7198630571365356, + "num_tokens": 147494095.0, + "step": 5792 + }, + { + "epoch": 0.63617395124094, + "grad_norm": 2.2128853797912598, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7104553580284119, + "num_tokens": 147518474.0, + "step": 5793 + }, + { + "epoch": 0.6362837689435537, + "grad_norm": 2.1743085384368896, + "learning_rate": 1e-06, + "loss": 1.0888, + "mean_token_accuracy": 0.6730743050575256, + "num_tokens": 147546572.0, + "step": 5794 + }, + { + "epoch": 0.6363935866461674, + "grad_norm": 2.214606285095215, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.702122151851654, + "num_tokens": 147571262.0, + "step": 5795 + }, + { + "epoch": 0.636503404348781, + "grad_norm": 1.942367672920227, + "learning_rate": 1e-06, + "loss": 1.0176, + "mean_token_accuracy": 0.6961885690689087, + "num_tokens": 147605916.0, + "step": 5796 + }, + { + "epoch": 0.6366132220513947, + "grad_norm": 1.9713218212127686, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7130832672119141, + "num_tokens": 147638261.0, + "step": 5797 + }, + { + "epoch": 0.6367230397540083, + "grad_norm": 2.1324405670166016, + "learning_rate": 1e-06, + "loss": 1.0313, + "mean_token_accuracy": 0.6849474310874939, + "num_tokens": 147667811.0, + "step": 5798 + }, + { + "epoch": 0.636832857456622, + "grad_norm": 2.3956618309020996, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.709311842918396, + "num_tokens": 147692779.0, + "step": 5799 + }, + { + "epoch": 0.6369426751592356, + "grad_norm": 1.932134747505188, + "learning_rate": 1e-06, + "loss": 1.0517, + "mean_token_accuracy": 0.678932249546051, + "num_tokens": 147724109.0, + "step": 5800 + }, + { + "epoch": 0.6370524928618493, + "grad_norm": 2.144275426864624, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7117078304290771, + "num_tokens": 147749999.0, + "step": 5801 + }, + { + "epoch": 0.637162310564463, + "grad_norm": 2.2562618255615234, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7024209499359131, + "num_tokens": 147773734.0, + "step": 5802 + }, + { + "epoch": 0.6372721282670767, + "grad_norm": 2.1924984455108643, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7016510367393494, + "num_tokens": 147797853.0, + "step": 5803 + }, + { + "epoch": 0.6373819459696903, + "grad_norm": 2.47615909576416, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7195898294448853, + "num_tokens": 147816770.0, + "step": 5804 + }, + { + "epoch": 0.637491763672304, + "grad_norm": 2.198660135269165, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.695853590965271, + "num_tokens": 147842334.0, + "step": 5805 + }, + { + "epoch": 0.6376015813749176, + "grad_norm": 2.06485915184021, + "learning_rate": 1e-06, + "loss": 1.0391, + "mean_token_accuracy": 0.6884844303131104, + "num_tokens": 147871873.0, + "step": 5806 + }, + { + "epoch": 0.6377113990775313, + "grad_norm": 2.4470627307891846, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.703255832195282, + "num_tokens": 147893504.0, + "step": 5807 + }, + { + "epoch": 0.6378212167801449, + "grad_norm": 2.528578281402588, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7316666841506958, + "num_tokens": 147912781.0, + "step": 5808 + }, + { + "epoch": 0.6379310344827587, + "grad_norm": 2.448216676712036, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7197563052177429, + "num_tokens": 147933917.0, + "step": 5809 + }, + { + "epoch": 0.6380408521853723, + "grad_norm": 2.2931301593780518, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6949347257614136, + "num_tokens": 147958813.0, + "step": 5810 + }, + { + "epoch": 0.638150669887986, + "grad_norm": 2.53468918800354, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7095887660980225, + "num_tokens": 147978126.0, + "step": 5811 + }, + { + "epoch": 0.6382604875905996, + "grad_norm": 2.3537890911102295, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.7037287950515747, + "num_tokens": 148005453.0, + "step": 5812 + }, + { + "epoch": 0.6383703052932133, + "grad_norm": 2.3529374599456787, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7166325449943542, + "num_tokens": 148028629.0, + "step": 5813 + }, + { + "epoch": 0.6384801229958269, + "grad_norm": 2.1645476818084717, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6950450539588928, + "num_tokens": 148055336.0, + "step": 5814 + }, + { + "epoch": 0.6385899406984406, + "grad_norm": 2.099497079849243, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7036800384521484, + "num_tokens": 148083484.0, + "step": 5815 + }, + { + "epoch": 0.6386997584010542, + "grad_norm": 2.175088405609131, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.7014453411102295, + "num_tokens": 148109834.0, + "step": 5816 + }, + { + "epoch": 0.638809576103668, + "grad_norm": 2.223787307739258, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.716458797454834, + "num_tokens": 148133574.0, + "step": 5817 + }, + { + "epoch": 0.6389193938062816, + "grad_norm": 2.349782705307007, + "learning_rate": 1e-06, + "loss": 1.0877, + "mean_token_accuracy": 0.6833713054656982, + "num_tokens": 148156166.0, + "step": 5818 + }, + { + "epoch": 0.6390292115088952, + "grad_norm": 1.9967602491378784, + "learning_rate": 1e-06, + "loss": 1.0475, + "mean_token_accuracy": 0.6832135915756226, + "num_tokens": 148188221.0, + "step": 5819 + }, + { + "epoch": 0.6391390292115089, + "grad_norm": 2.1807401180267334, + "learning_rate": 1e-06, + "loss": 1.0105, + "mean_token_accuracy": 0.705049991607666, + "num_tokens": 148215416.0, + "step": 5820 + }, + { + "epoch": 0.6392488469141225, + "grad_norm": 2.226827621459961, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.7078787684440613, + "num_tokens": 148239015.0, + "step": 5821 + }, + { + "epoch": 0.6393586646167362, + "grad_norm": 2.261744976043701, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.6963984966278076, + "num_tokens": 148264150.0, + "step": 5822 + }, + { + "epoch": 0.6394684823193498, + "grad_norm": 2.5278472900390625, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6868563890457153, + "num_tokens": 148284977.0, + "step": 5823 + }, + { + "epoch": 0.6395783000219636, + "grad_norm": 2.2646398544311523, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6930967569351196, + "num_tokens": 148310577.0, + "step": 5824 + }, + { + "epoch": 0.6396881177245772, + "grad_norm": 2.523021697998047, + "learning_rate": 1e-06, + "loss": 1.041, + "mean_token_accuracy": 0.6927450895309448, + "num_tokens": 148337456.0, + "step": 5825 + }, + { + "epoch": 0.6397979354271909, + "grad_norm": 2.227659225463867, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.709989070892334, + "num_tokens": 148362193.0, + "step": 5826 + }, + { + "epoch": 0.6399077531298045, + "grad_norm": 2.1954922676086426, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.68637615442276, + "num_tokens": 148389847.0, + "step": 5827 + }, + { + "epoch": 0.6400175708324182, + "grad_norm": 2.334618330001831, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7194538712501526, + "num_tokens": 148414299.0, + "step": 5828 + }, + { + "epoch": 0.6401273885350318, + "grad_norm": 2.3675577640533447, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7162331342697144, + "num_tokens": 148435435.0, + "step": 5829 + }, + { + "epoch": 0.6402372062376455, + "grad_norm": 2.19696307182312, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7183600664138794, + "num_tokens": 148460067.0, + "step": 5830 + }, + { + "epoch": 0.6403470239402592, + "grad_norm": 2.168992519378662, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.6990265846252441, + "num_tokens": 148486300.0, + "step": 5831 + }, + { + "epoch": 0.6404568416428729, + "grad_norm": 2.0685014724731445, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7491883039474487, + "num_tokens": 148511430.0, + "step": 5832 + }, + { + "epoch": 0.6405666593454865, + "grad_norm": 2.2081217765808105, + "learning_rate": 1e-06, + "loss": 1.0692, + "mean_token_accuracy": 0.684260368347168, + "num_tokens": 148540161.0, + "step": 5833 + }, + { + "epoch": 0.6406764770481002, + "grad_norm": 2.388798713684082, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.6954860687255859, + "num_tokens": 148564963.0, + "step": 5834 + }, + { + "epoch": 0.6407862947507138, + "grad_norm": 2.3749380111694336, + "learning_rate": 1e-06, + "loss": 1.0521, + "mean_token_accuracy": 0.6844996809959412, + "num_tokens": 148588718.0, + "step": 5835 + }, + { + "epoch": 0.6408961124533274, + "grad_norm": 2.2880611419677734, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7128154635429382, + "num_tokens": 148611584.0, + "step": 5836 + }, + { + "epoch": 0.6410059301559411, + "grad_norm": 2.1022744178771973, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7245921492576599, + "num_tokens": 148637151.0, + "step": 5837 + }, + { + "epoch": 0.6411157478585549, + "grad_norm": 1.8920300006866455, + "learning_rate": 1e-06, + "loss": 1.0523, + "mean_token_accuracy": 0.6875991821289062, + "num_tokens": 148670599.0, + "step": 5838 + }, + { + "epoch": 0.6412255655611685, + "grad_norm": 2.3576266765594482, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7039492130279541, + "num_tokens": 148692540.0, + "step": 5839 + }, + { + "epoch": 0.6413353832637821, + "grad_norm": 2.147937059402466, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7079576253890991, + "num_tokens": 148717757.0, + "step": 5840 + }, + { + "epoch": 0.6414452009663958, + "grad_norm": 2.018547296524048, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7245112061500549, + "num_tokens": 148745983.0, + "step": 5841 + }, + { + "epoch": 0.6415550186690094, + "grad_norm": 2.0164051055908203, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.6952512860298157, + "num_tokens": 148775219.0, + "step": 5842 + }, + { + "epoch": 0.6416648363716231, + "grad_norm": 2.5781137943267822, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6865894794464111, + "num_tokens": 148796617.0, + "step": 5843 + }, + { + "epoch": 0.6417746540742367, + "grad_norm": 2.164550542831421, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6973763704299927, + "num_tokens": 148822986.0, + "step": 5844 + }, + { + "epoch": 0.6418844717768504, + "grad_norm": 2.169764518737793, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6966722011566162, + "num_tokens": 148849145.0, + "step": 5845 + }, + { + "epoch": 0.6419942894794641, + "grad_norm": 2.490743637084961, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6979819536209106, + "num_tokens": 148871823.0, + "step": 5846 + }, + { + "epoch": 0.6421041071820778, + "grad_norm": 2.0516226291656494, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7313019633293152, + "num_tokens": 148898075.0, + "step": 5847 + }, + { + "epoch": 0.6422139248846914, + "grad_norm": 2.3730461597442627, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7057559490203857, + "num_tokens": 148921746.0, + "step": 5848 + }, + { + "epoch": 0.6423237425873051, + "grad_norm": 2.0841355323791504, + "learning_rate": 1e-06, + "loss": 1.0842, + "mean_token_accuracy": 0.6742076873779297, + "num_tokens": 148956331.0, + "step": 5849 + }, + { + "epoch": 0.6424335602899187, + "grad_norm": 2.049487829208374, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6911484003067017, + "num_tokens": 148984496.0, + "step": 5850 + }, + { + "epoch": 0.6425433779925324, + "grad_norm": 2.406203031539917, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7076589465141296, + "num_tokens": 149005669.0, + "step": 5851 + }, + { + "epoch": 0.642653195695146, + "grad_norm": 1.9604905843734741, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7176457047462463, + "num_tokens": 149033712.0, + "step": 5852 + }, + { + "epoch": 0.6427630133977598, + "grad_norm": 2.109022617340088, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7069224715232849, + "num_tokens": 149060093.0, + "step": 5853 + }, + { + "epoch": 0.6428728311003734, + "grad_norm": 2.294490337371826, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7134265899658203, + "num_tokens": 149082350.0, + "step": 5854 + }, + { + "epoch": 0.6429826488029871, + "grad_norm": 2.1414735317230225, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7132149934768677, + "num_tokens": 149108913.0, + "step": 5855 + }, + { + "epoch": 0.6430924665056007, + "grad_norm": 2.085212230682373, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.7008395195007324, + "num_tokens": 149137408.0, + "step": 5856 + }, + { + "epoch": 0.6432022842082143, + "grad_norm": 1.996980905532837, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7126453518867493, + "num_tokens": 149164430.0, + "step": 5857 + }, + { + "epoch": 0.643312101910828, + "grad_norm": 2.001493453979492, + "learning_rate": 1e-06, + "loss": 1.0272, + "mean_token_accuracy": 0.6886695623397827, + "num_tokens": 149193302.0, + "step": 5858 + }, + { + "epoch": 0.6434219196134416, + "grad_norm": 2.103532552719116, + "learning_rate": 1e-06, + "loss": 1.0406, + "mean_token_accuracy": 0.6916686296463013, + "num_tokens": 149220978.0, + "step": 5859 + }, + { + "epoch": 0.6435317373160554, + "grad_norm": 2.660198211669922, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6902718544006348, + "num_tokens": 149241224.0, + "step": 5860 + }, + { + "epoch": 0.643641555018669, + "grad_norm": 2.268984079360962, + "learning_rate": 1e-06, + "loss": 1.048, + "mean_token_accuracy": 0.6899574995040894, + "num_tokens": 149266342.0, + "step": 5861 + }, + { + "epoch": 0.6437513727212827, + "grad_norm": 2.0467352867126465, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6917640566825867, + "num_tokens": 149296850.0, + "step": 5862 + }, + { + "epoch": 0.6438611904238963, + "grad_norm": 2.2726454734802246, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7110394835472107, + "num_tokens": 149322050.0, + "step": 5863 + }, + { + "epoch": 0.64397100812651, + "grad_norm": 2.1132116317749023, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7079368829727173, + "num_tokens": 149350205.0, + "step": 5864 + }, + { + "epoch": 0.6440808258291236, + "grad_norm": 1.9493433237075806, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.714644193649292, + "num_tokens": 149380523.0, + "step": 5865 + }, + { + "epoch": 0.6441906435317373, + "grad_norm": 2.231675386428833, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.6926274299621582, + "num_tokens": 149405962.0, + "step": 5866 + }, + { + "epoch": 0.644300461234351, + "grad_norm": 1.9708479642868042, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.7046213150024414, + "num_tokens": 149436377.0, + "step": 5867 + }, + { + "epoch": 0.6444102789369647, + "grad_norm": 2.359325885772705, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7227585911750793, + "num_tokens": 149458940.0, + "step": 5868 + }, + { + "epoch": 0.6445200966395783, + "grad_norm": 2.268566370010376, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7054641246795654, + "num_tokens": 149481831.0, + "step": 5869 + }, + { + "epoch": 0.644629914342192, + "grad_norm": 2.227569341659546, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.6928633451461792, + "num_tokens": 149507153.0, + "step": 5870 + }, + { + "epoch": 0.6447397320448056, + "grad_norm": 2.2131173610687256, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.708276093006134, + "num_tokens": 149532462.0, + "step": 5871 + }, + { + "epoch": 0.6448495497474193, + "grad_norm": 2.2793681621551514, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7149554491043091, + "num_tokens": 149554657.0, + "step": 5872 + }, + { + "epoch": 0.6449593674500329, + "grad_norm": 2.4190149307250977, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7177063822746277, + "num_tokens": 149576646.0, + "step": 5873 + }, + { + "epoch": 0.6450691851526466, + "grad_norm": 2.2525298595428467, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7100117206573486, + "num_tokens": 149602337.0, + "step": 5874 + }, + { + "epoch": 0.6451790028552603, + "grad_norm": 2.1409707069396973, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7181089520454407, + "num_tokens": 149629739.0, + "step": 5875 + }, + { + "epoch": 0.645288820557874, + "grad_norm": 2.1999783515930176, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.729644775390625, + "num_tokens": 149653100.0, + "step": 5876 + }, + { + "epoch": 0.6453986382604876, + "grad_norm": 2.1298985481262207, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7028083801269531, + "num_tokens": 149679927.0, + "step": 5877 + }, + { + "epoch": 0.6455084559631012, + "grad_norm": 2.515763759613037, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.740176260471344, + "num_tokens": 149699213.0, + "step": 5878 + }, + { + "epoch": 0.6456182736657149, + "grad_norm": 1.8909521102905273, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6912370920181274, + "num_tokens": 149730533.0, + "step": 5879 + }, + { + "epoch": 0.6457280913683285, + "grad_norm": 2.432060956954956, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7432360649108887, + "num_tokens": 149750389.0, + "step": 5880 + }, + { + "epoch": 0.6458379090709422, + "grad_norm": 2.0583689212799072, + "learning_rate": 1e-06, + "loss": 1.0294, + "mean_token_accuracy": 0.6838797926902771, + "num_tokens": 149778479.0, + "step": 5881 + }, + { + "epoch": 0.6459477267735559, + "grad_norm": 2.1568753719329834, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6857438683509827, + "num_tokens": 149806352.0, + "step": 5882 + }, + { + "epoch": 0.6460575444761696, + "grad_norm": 2.1723875999450684, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6957623958587646, + "num_tokens": 149831739.0, + "step": 5883 + }, + { + "epoch": 0.6461673621787832, + "grad_norm": 2.1110596656799316, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7048941850662231, + "num_tokens": 149860570.0, + "step": 5884 + }, + { + "epoch": 0.6462771798813969, + "grad_norm": 2.212372064590454, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7337254285812378, + "num_tokens": 149884898.0, + "step": 5885 + }, + { + "epoch": 0.6463869975840105, + "grad_norm": 2.080580949783325, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7254458665847778, + "num_tokens": 149910492.0, + "step": 5886 + }, + { + "epoch": 0.6464968152866242, + "grad_norm": 2.1687843799591064, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.6982216238975525, + "num_tokens": 149938832.0, + "step": 5887 + }, + { + "epoch": 0.6466066329892378, + "grad_norm": 2.2149879932403564, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.700821042060852, + "num_tokens": 149962929.0, + "step": 5888 + }, + { + "epoch": 0.6467164506918516, + "grad_norm": 2.2094104290008545, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7042470574378967, + "num_tokens": 149987789.0, + "step": 5889 + }, + { + "epoch": 0.6468262683944652, + "grad_norm": 2.4244697093963623, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.6894545555114746, + "num_tokens": 150010217.0, + "step": 5890 + }, + { + "epoch": 0.6469360860970789, + "grad_norm": 2.2619242668151855, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7094513177871704, + "num_tokens": 150033165.0, + "step": 5891 + }, + { + "epoch": 0.6470459037996925, + "grad_norm": 1.9597744941711426, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7029114961624146, + "num_tokens": 150063268.0, + "step": 5892 + }, + { + "epoch": 0.6471557215023062, + "grad_norm": 2.0402677059173584, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6942926645278931, + "num_tokens": 150092137.0, + "step": 5893 + }, + { + "epoch": 0.6472655392049198, + "grad_norm": 2.4593868255615234, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7168816328048706, + "num_tokens": 150111662.0, + "step": 5894 + }, + { + "epoch": 0.6473753569075335, + "grad_norm": 2.0131442546844482, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7176631689071655, + "num_tokens": 150140192.0, + "step": 5895 + }, + { + "epoch": 0.6474851746101472, + "grad_norm": 2.093510627746582, + "learning_rate": 1e-06, + "loss": 1.0771, + "mean_token_accuracy": 0.6811352968215942, + "num_tokens": 150168040.0, + "step": 5896 + }, + { + "epoch": 0.6475949923127609, + "grad_norm": 2.400869369506836, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7281196117401123, + "num_tokens": 150190189.0, + "step": 5897 + }, + { + "epoch": 0.6477048100153745, + "grad_norm": 2.153454542160034, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.69659423828125, + "num_tokens": 150216049.0, + "step": 5898 + }, + { + "epoch": 0.6478146277179881, + "grad_norm": 2.027986526489258, + "learning_rate": 1e-06, + "loss": 1.0741, + "mean_token_accuracy": 0.6781284809112549, + "num_tokens": 150246370.0, + "step": 5899 + }, + { + "epoch": 0.6479244454206018, + "grad_norm": 1.9927048683166504, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7091172933578491, + "num_tokens": 150276581.0, + "step": 5900 + }, + { + "epoch": 0.6480342631232154, + "grad_norm": 2.4369149208068848, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7174263000488281, + "num_tokens": 150296907.0, + "step": 5901 + }, + { + "epoch": 0.6481440808258291, + "grad_norm": 2.129100799560547, + "learning_rate": 1e-06, + "loss": 0.9991, + "mean_token_accuracy": 0.6979763507843018, + "num_tokens": 150323650.0, + "step": 5902 + }, + { + "epoch": 0.6482538985284428, + "grad_norm": 2.2305307388305664, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7105772495269775, + "num_tokens": 150348704.0, + "step": 5903 + }, + { + "epoch": 0.6483637162310565, + "grad_norm": 1.872496485710144, + "learning_rate": 1e-06, + "loss": 1.0595, + "mean_token_accuracy": 0.6790354251861572, + "num_tokens": 150385838.0, + "step": 5904 + }, + { + "epoch": 0.6484735339336701, + "grad_norm": 2.098062515258789, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6982802152633667, + "num_tokens": 150412449.0, + "step": 5905 + }, + { + "epoch": 0.6485833516362838, + "grad_norm": 2.4837965965270996, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7185254693031311, + "num_tokens": 150431727.0, + "step": 5906 + }, + { + "epoch": 0.6486931693388974, + "grad_norm": 2.2735884189605713, + "learning_rate": 1e-06, + "loss": 1.0441, + "mean_token_accuracy": 0.6919832825660706, + "num_tokens": 150457983.0, + "step": 5907 + }, + { + "epoch": 0.6488029870415111, + "grad_norm": 2.1886932849884033, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7149237394332886, + "num_tokens": 150481762.0, + "step": 5908 + }, + { + "epoch": 0.6489128047441247, + "grad_norm": 2.1569137573242188, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7087913751602173, + "num_tokens": 150507946.0, + "step": 5909 + }, + { + "epoch": 0.6490226224467384, + "grad_norm": 2.175736904144287, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7245508432388306, + "num_tokens": 150533174.0, + "step": 5910 + }, + { + "epoch": 0.6491324401493521, + "grad_norm": 2.175614595413208, + "learning_rate": 1e-06, + "loss": 1.0389, + "mean_token_accuracy": 0.6818598508834839, + "num_tokens": 150559998.0, + "step": 5911 + }, + { + "epoch": 0.6492422578519658, + "grad_norm": 2.236654043197632, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.6946338415145874, + "num_tokens": 150583972.0, + "step": 5912 + }, + { + "epoch": 0.6493520755545794, + "grad_norm": 1.7947471141815186, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7058671712875366, + "num_tokens": 150617752.0, + "step": 5913 + }, + { + "epoch": 0.6494618932571931, + "grad_norm": 1.926334261894226, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7162469029426575, + "num_tokens": 150647767.0, + "step": 5914 + }, + { + "epoch": 0.6495717109598067, + "grad_norm": 2.1957030296325684, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7261801362037659, + "num_tokens": 150673878.0, + "step": 5915 + }, + { + "epoch": 0.6496815286624203, + "grad_norm": 2.190523862838745, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7028173208236694, + "num_tokens": 150698198.0, + "step": 5916 + }, + { + "epoch": 0.649791346365034, + "grad_norm": 1.9904465675354004, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6981000900268555, + "num_tokens": 150727594.0, + "step": 5917 + }, + { + "epoch": 0.6499011640676478, + "grad_norm": 2.157862901687622, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6985589265823364, + "num_tokens": 150755629.0, + "step": 5918 + }, + { + "epoch": 0.6500109817702614, + "grad_norm": 2.497972011566162, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7109240889549255, + "num_tokens": 150776550.0, + "step": 5919 + }, + { + "epoch": 0.650120799472875, + "grad_norm": 2.15293288230896, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7047009468078613, + "num_tokens": 150802304.0, + "step": 5920 + }, + { + "epoch": 0.6502306171754887, + "grad_norm": 2.272768974304199, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7123318910598755, + "num_tokens": 150827034.0, + "step": 5921 + }, + { + "epoch": 0.6503404348781023, + "grad_norm": 2.146719455718994, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6924697160720825, + "num_tokens": 150853404.0, + "step": 5922 + }, + { + "epoch": 0.650450252580716, + "grad_norm": 2.048037052154541, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7017557621002197, + "num_tokens": 150882246.0, + "step": 5923 + }, + { + "epoch": 0.6505600702833296, + "grad_norm": 2.2974376678466797, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7339379787445068, + "num_tokens": 150903520.0, + "step": 5924 + }, + { + "epoch": 0.6506698879859434, + "grad_norm": 2.143785238265991, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6988775730133057, + "num_tokens": 150929040.0, + "step": 5925 + }, + { + "epoch": 0.650779705688557, + "grad_norm": 2.2858328819274902, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.696465015411377, + "num_tokens": 150954380.0, + "step": 5926 + }, + { + "epoch": 0.6508895233911707, + "grad_norm": 1.940305471420288, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7071666717529297, + "num_tokens": 150984429.0, + "step": 5927 + }, + { + "epoch": 0.6509993410937843, + "grad_norm": 2.111670732498169, + "learning_rate": 1e-06, + "loss": 1.0545, + "mean_token_accuracy": 0.6876468658447266, + "num_tokens": 151012480.0, + "step": 5928 + }, + { + "epoch": 0.651109158796398, + "grad_norm": 2.004377841949463, + "learning_rate": 1e-06, + "loss": 1.0497, + "mean_token_accuracy": 0.682711124420166, + "num_tokens": 151042119.0, + "step": 5929 + }, + { + "epoch": 0.6512189764990116, + "grad_norm": 2.084679126739502, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6931244134902954, + "num_tokens": 151069018.0, + "step": 5930 + }, + { + "epoch": 0.6513287942016253, + "grad_norm": 2.1850638389587402, + "learning_rate": 1e-06, + "loss": 1.0974, + "mean_token_accuracy": 0.6689781546592712, + "num_tokens": 151096657.0, + "step": 5931 + }, + { + "epoch": 0.651438611904239, + "grad_norm": 1.9415113925933838, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.697626531124115, + "num_tokens": 151127259.0, + "step": 5932 + }, + { + "epoch": 0.6515484296068527, + "grad_norm": 2.534313678741455, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7139467000961304, + "num_tokens": 151145658.0, + "step": 5933 + }, + { + "epoch": 0.6516582473094663, + "grad_norm": 1.9569487571716309, + "learning_rate": 1e-06, + "loss": 0.9784, + "mean_token_accuracy": 0.7064282894134521, + "num_tokens": 151174467.0, + "step": 5934 + }, + { + "epoch": 0.65176806501208, + "grad_norm": 2.1408092975616455, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.703161895275116, + "num_tokens": 151200337.0, + "step": 5935 + }, + { + "epoch": 0.6518778827146936, + "grad_norm": 2.5351831912994385, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7309579253196716, + "num_tokens": 151218854.0, + "step": 5936 + }, + { + "epoch": 0.6519877004173072, + "grad_norm": 2.1966893672943115, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.708367645740509, + "num_tokens": 151242306.0, + "step": 5937 + }, + { + "epoch": 0.6520975181199209, + "grad_norm": 2.169508695602417, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.67649906873703, + "num_tokens": 151268711.0, + "step": 5938 + }, + { + "epoch": 0.6522073358225345, + "grad_norm": 2.647064208984375, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7224299907684326, + "num_tokens": 151286427.0, + "step": 5939 + }, + { + "epoch": 0.6523171535251483, + "grad_norm": 2.215704917907715, + "learning_rate": 1e-06, + "loss": 1.0083, + "mean_token_accuracy": 0.6982178688049316, + "num_tokens": 151312589.0, + "step": 5940 + }, + { + "epoch": 0.6524269712277619, + "grad_norm": 2.246400833129883, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7075876593589783, + "num_tokens": 151339017.0, + "step": 5941 + }, + { + "epoch": 0.6525367889303756, + "grad_norm": 2.500084400177002, + "learning_rate": 1e-06, + "loss": 1.0377, + "mean_token_accuracy": 0.682722806930542, + "num_tokens": 151360927.0, + "step": 5942 + }, + { + "epoch": 0.6526466066329892, + "grad_norm": 2.2041051387786865, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6889723539352417, + "num_tokens": 151388518.0, + "step": 5943 + }, + { + "epoch": 0.6527564243356029, + "grad_norm": 2.094740629196167, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7070257663726807, + "num_tokens": 151416526.0, + "step": 5944 + }, + { + "epoch": 0.6528662420382165, + "grad_norm": 2.016688585281372, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7077150344848633, + "num_tokens": 151443337.0, + "step": 5945 + }, + { + "epoch": 0.6529760597408302, + "grad_norm": 2.1202194690704346, + "learning_rate": 1e-06, + "loss": 1.0634, + "mean_token_accuracy": 0.6778713464736938, + "num_tokens": 151470390.0, + "step": 5946 + }, + { + "epoch": 0.6530858774434439, + "grad_norm": 2.1778926849365234, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7079455256462097, + "num_tokens": 151495299.0, + "step": 5947 + }, + { + "epoch": 0.6531956951460576, + "grad_norm": 2.3556392192840576, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.714445948600769, + "num_tokens": 151517669.0, + "step": 5948 + }, + { + "epoch": 0.6533055128486712, + "grad_norm": 2.157433032989502, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.6967881321907043, + "num_tokens": 151542424.0, + "step": 5949 + }, + { + "epoch": 0.6534153305512849, + "grad_norm": 2.4156172275543213, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7049693465232849, + "num_tokens": 151566936.0, + "step": 5950 + }, + { + "epoch": 0.6535251482538985, + "grad_norm": 2.5111048221588135, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7180140614509583, + "num_tokens": 151587830.0, + "step": 5951 + }, + { + "epoch": 0.6536349659565122, + "grad_norm": 2.090536117553711, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7165933847427368, + "num_tokens": 151614969.0, + "step": 5952 + }, + { + "epoch": 0.6537447836591258, + "grad_norm": 2.1807758808135986, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.722370445728302, + "num_tokens": 151640494.0, + "step": 5953 + }, + { + "epoch": 0.6538546013617396, + "grad_norm": 2.1486895084381104, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7068285942077637, + "num_tokens": 151666563.0, + "step": 5954 + }, + { + "epoch": 0.6539644190643532, + "grad_norm": 2.2157554626464844, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7079429030418396, + "num_tokens": 151692823.0, + "step": 5955 + }, + { + "epoch": 0.6540742367669669, + "grad_norm": 2.499218702316284, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7154110074043274, + "num_tokens": 151713077.0, + "step": 5956 + }, + { + "epoch": 0.6541840544695805, + "grad_norm": 2.153576135635376, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7311152219772339, + "num_tokens": 151736985.0, + "step": 5957 + }, + { + "epoch": 0.6542938721721941, + "grad_norm": 2.0579755306243896, + "learning_rate": 1e-06, + "loss": 1.0894, + "mean_token_accuracy": 0.6823103427886963, + "num_tokens": 151764186.0, + "step": 5958 + }, + { + "epoch": 0.6544036898748078, + "grad_norm": 2.206023931503296, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7027535438537598, + "num_tokens": 151789493.0, + "step": 5959 + }, + { + "epoch": 0.6545135075774214, + "grad_norm": 2.522639036178589, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7336701154708862, + "num_tokens": 151808077.0, + "step": 5960 + }, + { + "epoch": 0.6546233252800352, + "grad_norm": 2.169175624847412, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.699007511138916, + "num_tokens": 151833376.0, + "step": 5961 + }, + { + "epoch": 0.6547331429826488, + "grad_norm": 2.1027331352233887, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6962721943855286, + "num_tokens": 151860795.0, + "step": 5962 + }, + { + "epoch": 0.6548429606852625, + "grad_norm": 2.166433572769165, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7168447971343994, + "num_tokens": 151887276.0, + "step": 5963 + }, + { + "epoch": 0.6549527783878761, + "grad_norm": 2.197064161300659, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6887304782867432, + "num_tokens": 151911600.0, + "step": 5964 + }, + { + "epoch": 0.6550625960904898, + "grad_norm": 1.9214125871658325, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.702511727809906, + "num_tokens": 151944399.0, + "step": 5965 + }, + { + "epoch": 0.6551724137931034, + "grad_norm": 2.1322782039642334, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7163997292518616, + "num_tokens": 151970479.0, + "step": 5966 + }, + { + "epoch": 0.6552822314957171, + "grad_norm": 2.007068395614624, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7058712244033813, + "num_tokens": 151997246.0, + "step": 5967 + }, + { + "epoch": 0.6553920491983307, + "grad_norm": 2.2071218490600586, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.7041764259338379, + "num_tokens": 152022401.0, + "step": 5968 + }, + { + "epoch": 0.6555018669009445, + "grad_norm": 2.338338851928711, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7315754890441895, + "num_tokens": 152044542.0, + "step": 5969 + }, + { + "epoch": 0.6556116846035581, + "grad_norm": 2.4914779663085938, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7124626636505127, + "num_tokens": 152062755.0, + "step": 5970 + }, + { + "epoch": 0.6557215023061718, + "grad_norm": 2.4279518127441406, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7085933685302734, + "num_tokens": 152085734.0, + "step": 5971 + }, + { + "epoch": 0.6558313200087854, + "grad_norm": 2.0622057914733887, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7305923104286194, + "num_tokens": 152111636.0, + "step": 5972 + }, + { + "epoch": 0.6559411377113991, + "grad_norm": 2.2252259254455566, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.700384795665741, + "num_tokens": 152137163.0, + "step": 5973 + }, + { + "epoch": 0.6560509554140127, + "grad_norm": 2.189159870147705, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.699074923992157, + "num_tokens": 152162404.0, + "step": 5974 + }, + { + "epoch": 0.6561607731166264, + "grad_norm": 2.1919312477111816, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7111555337905884, + "num_tokens": 152187039.0, + "step": 5975 + }, + { + "epoch": 0.6562705908192401, + "grad_norm": 2.3211727142333984, + "learning_rate": 1e-06, + "loss": 1.0442, + "mean_token_accuracy": 0.6854383945465088, + "num_tokens": 152210960.0, + "step": 5976 + }, + { + "epoch": 0.6563804085218538, + "grad_norm": 2.073641061782837, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7254725098609924, + "num_tokens": 152236111.0, + "step": 5977 + }, + { + "epoch": 0.6564902262244674, + "grad_norm": 2.0415561199188232, + "learning_rate": 1e-06, + "loss": 1.0365, + "mean_token_accuracy": 0.6856458187103271, + "num_tokens": 152264281.0, + "step": 5978 + }, + { + "epoch": 0.656600043927081, + "grad_norm": 2.0806076526641846, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7050668001174927, + "num_tokens": 152291776.0, + "step": 5979 + }, + { + "epoch": 0.6567098616296947, + "grad_norm": 2.1342971324920654, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7251785397529602, + "num_tokens": 152316626.0, + "step": 5980 + }, + { + "epoch": 0.6568196793323083, + "grad_norm": 2.4791576862335205, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6961760520935059, + "num_tokens": 152337310.0, + "step": 5981 + }, + { + "epoch": 0.656929497034922, + "grad_norm": 2.177469253540039, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7201405763626099, + "num_tokens": 152361484.0, + "step": 5982 + }, + { + "epoch": 0.6570393147375357, + "grad_norm": 2.0542120933532715, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7081292271614075, + "num_tokens": 152389441.0, + "step": 5983 + }, + { + "epoch": 0.6571491324401494, + "grad_norm": 2.3837180137634277, + "learning_rate": 1e-06, + "loss": 1.081, + "mean_token_accuracy": 0.6873551607131958, + "num_tokens": 152411403.0, + "step": 5984 + }, + { + "epoch": 0.657258950142763, + "grad_norm": 2.334969997406006, + "learning_rate": 1e-06, + "loss": 1.0782, + "mean_token_accuracy": 0.6753844618797302, + "num_tokens": 152435435.0, + "step": 5985 + }, + { + "epoch": 0.6573687678453767, + "grad_norm": 2.621278762817383, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7347390651702881, + "num_tokens": 152453969.0, + "step": 5986 + }, + { + "epoch": 0.6574785855479903, + "grad_norm": 2.1647865772247314, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7252113819122314, + "num_tokens": 152479515.0, + "step": 5987 + }, + { + "epoch": 0.657588403250604, + "grad_norm": 2.2286977767944336, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7250493764877319, + "num_tokens": 152503908.0, + "step": 5988 + }, + { + "epoch": 0.6576982209532176, + "grad_norm": 2.353201150894165, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7089653015136719, + "num_tokens": 152525645.0, + "step": 5989 + }, + { + "epoch": 0.6578080386558314, + "grad_norm": 2.4270613193511963, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7089487910270691, + "num_tokens": 152546660.0, + "step": 5990 + }, + { + "epoch": 0.657917856358445, + "grad_norm": 2.395860195159912, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6993330717086792, + "num_tokens": 152570314.0, + "step": 5991 + }, + { + "epoch": 0.6580276740610587, + "grad_norm": 2.1813035011291504, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.696628212928772, + "num_tokens": 152595457.0, + "step": 5992 + }, + { + "epoch": 0.6581374917636723, + "grad_norm": 2.1470375061035156, + "learning_rate": 1e-06, + "loss": 0.8278, + "mean_token_accuracy": 0.7548568248748779, + "num_tokens": 152619899.0, + "step": 5993 + }, + { + "epoch": 0.658247309466286, + "grad_norm": 2.46647572517395, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7088291645050049, + "num_tokens": 152639344.0, + "step": 5994 + }, + { + "epoch": 0.6583571271688996, + "grad_norm": 2.1357572078704834, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7036664485931396, + "num_tokens": 152665276.0, + "step": 5995 + }, + { + "epoch": 0.6584669448715132, + "grad_norm": 2.0911917686462402, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7018663287162781, + "num_tokens": 152693045.0, + "step": 5996 + }, + { + "epoch": 0.6585767625741269, + "grad_norm": 2.1617162227630615, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7269325256347656, + "num_tokens": 152718457.0, + "step": 5997 + }, + { + "epoch": 0.6586865802767407, + "grad_norm": 2.3729870319366455, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.7014238238334656, + "num_tokens": 152740366.0, + "step": 5998 + }, + { + "epoch": 0.6587963979793543, + "grad_norm": 1.9039995670318604, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7109648585319519, + "num_tokens": 152772400.0, + "step": 5999 + }, + { + "epoch": 0.6589062156819679, + "grad_norm": 1.9927037954330444, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7200876474380493, + "num_tokens": 152799039.0, + "step": 6000 + }, + { + "epoch": 0.6590160333845816, + "grad_norm": 2.137235641479492, + "learning_rate": 1e-06, + "loss": 1.0669, + "mean_token_accuracy": 0.6837934255599976, + "num_tokens": 152825149.0, + "step": 6001 + }, + { + "epoch": 0.6591258510871952, + "grad_norm": 2.295180559158325, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7074444890022278, + "num_tokens": 152847842.0, + "step": 6002 + }, + { + "epoch": 0.6592356687898089, + "grad_norm": 2.485663890838623, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.713645339012146, + "num_tokens": 152869000.0, + "step": 6003 + }, + { + "epoch": 0.6593454864924225, + "grad_norm": 2.4047253131866455, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7039835453033447, + "num_tokens": 152889765.0, + "step": 6004 + }, + { + "epoch": 0.6594553041950363, + "grad_norm": 2.237917423248291, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7160897254943848, + "num_tokens": 152913129.0, + "step": 6005 + }, + { + "epoch": 0.6595651218976499, + "grad_norm": 2.306004524230957, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7276349067687988, + "num_tokens": 152934754.0, + "step": 6006 + }, + { + "epoch": 0.6596749396002636, + "grad_norm": 2.1867151260375977, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7042587995529175, + "num_tokens": 152959976.0, + "step": 6007 + }, + { + "epoch": 0.6597847573028772, + "grad_norm": 2.114616870880127, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6961629986763, + "num_tokens": 152987805.0, + "step": 6008 + }, + { + "epoch": 0.6598945750054909, + "grad_norm": 2.4735281467437744, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7088990807533264, + "num_tokens": 153009065.0, + "step": 6009 + }, + { + "epoch": 0.6600043927081045, + "grad_norm": 2.347442626953125, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6911585330963135, + "num_tokens": 153032376.0, + "step": 6010 + }, + { + "epoch": 0.6601142104107182, + "grad_norm": 2.0791447162628174, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.72372967004776, + "num_tokens": 153057989.0, + "step": 6011 + }, + { + "epoch": 0.6602240281133319, + "grad_norm": 2.245753049850464, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.6890642046928406, + "num_tokens": 153082229.0, + "step": 6012 + }, + { + "epoch": 0.6603338458159456, + "grad_norm": 2.3846311569213867, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7084458470344543, + "num_tokens": 153105666.0, + "step": 6013 + }, + { + "epoch": 0.6604436635185592, + "grad_norm": 2.027315616607666, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6888988614082336, + "num_tokens": 153137922.0, + "step": 6014 + }, + { + "epoch": 0.6605534812211729, + "grad_norm": 2.2434914112091064, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6995449662208557, + "num_tokens": 153163533.0, + "step": 6015 + }, + { + "epoch": 0.6606632989237865, + "grad_norm": 2.266599655151367, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6938616037368774, + "num_tokens": 153189040.0, + "step": 6016 + }, + { + "epoch": 0.6607731166264001, + "grad_norm": 2.3587028980255127, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.70362389087677, + "num_tokens": 153211034.0, + "step": 6017 + }, + { + "epoch": 0.6608829343290138, + "grad_norm": 2.1413443088531494, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7108198404312134, + "num_tokens": 153235307.0, + "step": 6018 + }, + { + "epoch": 0.6609927520316276, + "grad_norm": 2.3273022174835205, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7186490893363953, + "num_tokens": 153257165.0, + "step": 6019 + }, + { + "epoch": 0.6611025697342412, + "grad_norm": 1.9505201578140259, + "learning_rate": 1e-06, + "loss": 1.106, + "mean_token_accuracy": 0.6715434789657593, + "num_tokens": 153291824.0, + "step": 6020 + }, + { + "epoch": 0.6612123874368548, + "grad_norm": 1.9976725578308105, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7063285112380981, + "num_tokens": 153319986.0, + "step": 6021 + }, + { + "epoch": 0.6613222051394685, + "grad_norm": 2.2273974418640137, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7247839570045471, + "num_tokens": 153343690.0, + "step": 6022 + }, + { + "epoch": 0.6614320228420821, + "grad_norm": 2.265629291534424, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7092833518981934, + "num_tokens": 153367646.0, + "step": 6023 + }, + { + "epoch": 0.6615418405446958, + "grad_norm": 2.165318250656128, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7014034986495972, + "num_tokens": 153394521.0, + "step": 6024 + }, + { + "epoch": 0.6616516582473094, + "grad_norm": 2.23982310295105, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7278358936309814, + "num_tokens": 153417506.0, + "step": 6025 + }, + { + "epoch": 0.6617614759499231, + "grad_norm": 2.133452892303467, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7182856798171997, + "num_tokens": 153443010.0, + "step": 6026 + }, + { + "epoch": 0.6618712936525368, + "grad_norm": 2.2029714584350586, + "learning_rate": 1e-06, + "loss": 1.0116, + "mean_token_accuracy": 0.7044597268104553, + "num_tokens": 153468269.0, + "step": 6027 + }, + { + "epoch": 0.6619811113551505, + "grad_norm": 2.402517795562744, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6950194835662842, + "num_tokens": 153490890.0, + "step": 6028 + }, + { + "epoch": 0.6620909290577641, + "grad_norm": 2.196552276611328, + "learning_rate": 1e-06, + "loss": 1.0918, + "mean_token_accuracy": 0.6751686334609985, + "num_tokens": 153518305.0, + "step": 6029 + }, + { + "epoch": 0.6622007467603778, + "grad_norm": 2.1335608959198, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.6991211175918579, + "num_tokens": 153544806.0, + "step": 6030 + }, + { + "epoch": 0.6623105644629914, + "grad_norm": 2.4711432456970215, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7186772227287292, + "num_tokens": 153564490.0, + "step": 6031 + }, + { + "epoch": 0.6624203821656051, + "grad_norm": 2.220215320587158, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7159988880157471, + "num_tokens": 153588324.0, + "step": 6032 + }, + { + "epoch": 0.6625301998682187, + "grad_norm": 2.247575044631958, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6902035474777222, + "num_tokens": 153611785.0, + "step": 6033 + }, + { + "epoch": 0.6626400175708325, + "grad_norm": 1.8385084867477417, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.711289644241333, + "num_tokens": 153645596.0, + "step": 6034 + }, + { + "epoch": 0.6627498352734461, + "grad_norm": 2.271136999130249, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6987805366516113, + "num_tokens": 153670164.0, + "step": 6035 + }, + { + "epoch": 0.6628596529760598, + "grad_norm": 2.118600606918335, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7279320955276489, + "num_tokens": 153696248.0, + "step": 6036 + }, + { + "epoch": 0.6629694706786734, + "grad_norm": 2.2203011512756348, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.6953914165496826, + "num_tokens": 153722042.0, + "step": 6037 + }, + { + "epoch": 0.663079288381287, + "grad_norm": 1.9706830978393555, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7262604236602783, + "num_tokens": 153750204.0, + "step": 6038 + }, + { + "epoch": 0.6631891060839007, + "grad_norm": 2.345419406890869, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7359553575515747, + "num_tokens": 153771742.0, + "step": 6039 + }, + { + "epoch": 0.6632989237865143, + "grad_norm": 2.136082172393799, + "learning_rate": 1e-06, + "loss": 1.0469, + "mean_token_accuracy": 0.6839342713356018, + "num_tokens": 153800646.0, + "step": 6040 + }, + { + "epoch": 0.6634087414891281, + "grad_norm": 2.547966957092285, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7197160124778748, + "num_tokens": 153819495.0, + "step": 6041 + }, + { + "epoch": 0.6635185591917417, + "grad_norm": 2.1061794757843018, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7092689275741577, + "num_tokens": 153849146.0, + "step": 6042 + }, + { + "epoch": 0.6636283768943554, + "grad_norm": 2.6003575325012207, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7258533239364624, + "num_tokens": 153867748.0, + "step": 6043 + }, + { + "epoch": 0.663738194596969, + "grad_norm": 2.4716286659240723, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7067852020263672, + "num_tokens": 153889119.0, + "step": 6044 + }, + { + "epoch": 0.6638480122995827, + "grad_norm": 2.4662458896636963, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7240382432937622, + "num_tokens": 153909840.0, + "step": 6045 + }, + { + "epoch": 0.6639578300021963, + "grad_norm": 2.154825210571289, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.7091755867004395, + "num_tokens": 153937055.0, + "step": 6046 + }, + { + "epoch": 0.66406764770481, + "grad_norm": 2.2848122119903564, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7183963656425476, + "num_tokens": 153960433.0, + "step": 6047 + }, + { + "epoch": 0.6641774654074237, + "grad_norm": 2.0788826942443848, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7025768756866455, + "num_tokens": 153988030.0, + "step": 6048 + }, + { + "epoch": 0.6642872831100374, + "grad_norm": 2.865602970123291, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.722504734992981, + "num_tokens": 154005067.0, + "step": 6049 + }, + { + "epoch": 0.664397100812651, + "grad_norm": 2.1452009677886963, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6929277181625366, + "num_tokens": 154030931.0, + "step": 6050 + }, + { + "epoch": 0.6645069185152647, + "grad_norm": 2.19224214553833, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7137576341629028, + "num_tokens": 154056476.0, + "step": 6051 + }, + { + "epoch": 0.6646167362178783, + "grad_norm": 2.006129026412964, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7027632594108582, + "num_tokens": 154085536.0, + "step": 6052 + }, + { + "epoch": 0.664726553920492, + "grad_norm": 2.0996432304382324, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7148160338401794, + "num_tokens": 154111395.0, + "step": 6053 + }, + { + "epoch": 0.6648363716231056, + "grad_norm": 2.14947772026062, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7228633165359497, + "num_tokens": 154134430.0, + "step": 6054 + }, + { + "epoch": 0.6649461893257194, + "grad_norm": 2.231599807739258, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6834120750427246, + "num_tokens": 154161195.0, + "step": 6055 + }, + { + "epoch": 0.665056007028333, + "grad_norm": 1.857946515083313, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.699763834476471, + "num_tokens": 154196484.0, + "step": 6056 + }, + { + "epoch": 0.6651658247309467, + "grad_norm": 2.5321648120880127, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7252542972564697, + "num_tokens": 154215054.0, + "step": 6057 + }, + { + "epoch": 0.6652756424335603, + "grad_norm": 2.2887721061706543, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7247928380966187, + "num_tokens": 154236090.0, + "step": 6058 + }, + { + "epoch": 0.665385460136174, + "grad_norm": 2.224702835083008, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7095443606376648, + "num_tokens": 154259466.0, + "step": 6059 + }, + { + "epoch": 0.6654952778387876, + "grad_norm": 2.2349789142608643, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7272221446037292, + "num_tokens": 154281672.0, + "step": 6060 + }, + { + "epoch": 0.6656050955414012, + "grad_norm": 2.0591752529144287, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7241418957710266, + "num_tokens": 154308075.0, + "step": 6061 + }, + { + "epoch": 0.6657149132440149, + "grad_norm": 1.9696264266967773, + "learning_rate": 1e-06, + "loss": 1.0512, + "mean_token_accuracy": 0.6785675287246704, + "num_tokens": 154338555.0, + "step": 6062 + }, + { + "epoch": 0.6658247309466286, + "grad_norm": 2.1624364852905273, + "learning_rate": 1e-06, + "loss": 1.0409, + "mean_token_accuracy": 0.6920747756958008, + "num_tokens": 154365777.0, + "step": 6063 + }, + { + "epoch": 0.6659345486492423, + "grad_norm": 2.3304128646850586, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7132288813591003, + "num_tokens": 154388596.0, + "step": 6064 + }, + { + "epoch": 0.6660443663518559, + "grad_norm": 2.429482936859131, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7307846546173096, + "num_tokens": 154410621.0, + "step": 6065 + }, + { + "epoch": 0.6661541840544696, + "grad_norm": 2.1751949787139893, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7263180017471313, + "num_tokens": 154434749.0, + "step": 6066 + }, + { + "epoch": 0.6662640017570832, + "grad_norm": 2.3087687492370605, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7096554040908813, + "num_tokens": 154458069.0, + "step": 6067 + }, + { + "epoch": 0.6663738194596969, + "grad_norm": 2.440972328186035, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7389373779296875, + "num_tokens": 154478464.0, + "step": 6068 + }, + { + "epoch": 0.6664836371623105, + "grad_norm": 2.3969974517822266, + "learning_rate": 1e-06, + "loss": 1.0432, + "mean_token_accuracy": 0.6856494545936584, + "num_tokens": 154500810.0, + "step": 6069 + }, + { + "epoch": 0.6665934548649243, + "grad_norm": 1.9369821548461914, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.6974065899848938, + "num_tokens": 154530052.0, + "step": 6070 + }, + { + "epoch": 0.6667032725675379, + "grad_norm": 2.1850085258483887, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7279265522956848, + "num_tokens": 154554194.0, + "step": 6071 + }, + { + "epoch": 0.6668130902701516, + "grad_norm": 1.98985755443573, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.6930633783340454, + "num_tokens": 154585093.0, + "step": 6072 + }, + { + "epoch": 0.6669229079727652, + "grad_norm": 1.9061362743377686, + "learning_rate": 1e-06, + "loss": 1.1041, + "mean_token_accuracy": 0.6674430966377258, + "num_tokens": 154619436.0, + "step": 6073 + }, + { + "epoch": 0.6670327256753789, + "grad_norm": 2.1290998458862305, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.708010196685791, + "num_tokens": 154647508.0, + "step": 6074 + }, + { + "epoch": 0.6671425433779925, + "grad_norm": 2.130063056945801, + "learning_rate": 1e-06, + "loss": 1.0303, + "mean_token_accuracy": 0.6902375221252441, + "num_tokens": 154675064.0, + "step": 6075 + }, + { + "epoch": 0.6672523610806061, + "grad_norm": 2.2250537872314453, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.712283730506897, + "num_tokens": 154699845.0, + "step": 6076 + }, + { + "epoch": 0.6673621787832199, + "grad_norm": 2.087958335876465, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7137494087219238, + "num_tokens": 154728038.0, + "step": 6077 + }, + { + "epoch": 0.6674719964858336, + "grad_norm": 2.406630039215088, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6956491470336914, + "num_tokens": 154750840.0, + "step": 6078 + }, + { + "epoch": 0.6675818141884472, + "grad_norm": 2.2307002544403076, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7047331929206848, + "num_tokens": 154774657.0, + "step": 6079 + }, + { + "epoch": 0.6676916318910608, + "grad_norm": 2.142162322998047, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6900488138198853, + "num_tokens": 154801140.0, + "step": 6080 + }, + { + "epoch": 0.6678014495936745, + "grad_norm": 2.2164041996002197, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7068397998809814, + "num_tokens": 154825420.0, + "step": 6081 + }, + { + "epoch": 0.6679112672962881, + "grad_norm": 2.3068063259124756, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.6972221732139587, + "num_tokens": 154849811.0, + "step": 6082 + }, + { + "epoch": 0.6680210849989018, + "grad_norm": 2.430649757385254, + "learning_rate": 1e-06, + "loss": 1.0171, + "mean_token_accuracy": 0.6947578191757202, + "num_tokens": 154872625.0, + "step": 6083 + }, + { + "epoch": 0.6681309027015155, + "grad_norm": 2.247270107269287, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7140519022941589, + "num_tokens": 154898159.0, + "step": 6084 + }, + { + "epoch": 0.6682407204041292, + "grad_norm": 2.149075746536255, + "learning_rate": 1e-06, + "loss": 1.038, + "mean_token_accuracy": 0.6903015971183777, + "num_tokens": 154925555.0, + "step": 6085 + }, + { + "epoch": 0.6683505381067428, + "grad_norm": 2.2735354900360107, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7150013446807861, + "num_tokens": 154950245.0, + "step": 6086 + }, + { + "epoch": 0.6684603558093565, + "grad_norm": 1.896074652671814, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.7021325826644897, + "num_tokens": 154982075.0, + "step": 6087 + }, + { + "epoch": 0.6685701735119701, + "grad_norm": 2.0491995811462402, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7141861915588379, + "num_tokens": 155009731.0, + "step": 6088 + }, + { + "epoch": 0.6686799912145838, + "grad_norm": 2.288769483566284, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7180254459381104, + "num_tokens": 155032707.0, + "step": 6089 + }, + { + "epoch": 0.6687898089171974, + "grad_norm": 1.9880679845809937, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6990567445755005, + "num_tokens": 155063768.0, + "step": 6090 + }, + { + "epoch": 0.6688996266198111, + "grad_norm": 2.0020663738250732, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6937898993492126, + "num_tokens": 155094630.0, + "step": 6091 + }, + { + "epoch": 0.6690094443224248, + "grad_norm": 2.202633857727051, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7120472192764282, + "num_tokens": 155119930.0, + "step": 6092 + }, + { + "epoch": 0.6691192620250385, + "grad_norm": 2.330949068069458, + "learning_rate": 1e-06, + "loss": 1.0479, + "mean_token_accuracy": 0.686287522315979, + "num_tokens": 155143661.0, + "step": 6093 + }, + { + "epoch": 0.6692290797276521, + "grad_norm": 2.0168540477752686, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7218185067176819, + "num_tokens": 155171189.0, + "step": 6094 + }, + { + "epoch": 0.6693388974302658, + "grad_norm": 2.438537359237671, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7316335439682007, + "num_tokens": 155190750.0, + "step": 6095 + }, + { + "epoch": 0.6694487151328794, + "grad_norm": 2.125173568725586, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6904665231704712, + "num_tokens": 155220489.0, + "step": 6096 + }, + { + "epoch": 0.669558532835493, + "grad_norm": 2.063009262084961, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7108865976333618, + "num_tokens": 155248086.0, + "step": 6097 + }, + { + "epoch": 0.6696683505381067, + "grad_norm": 1.990029215812683, + "learning_rate": 1e-06, + "loss": 1.0455, + "mean_token_accuracy": 0.6961370706558228, + "num_tokens": 155277925.0, + "step": 6098 + }, + { + "epoch": 0.6697781682407205, + "grad_norm": 1.9890564680099487, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.688153862953186, + "num_tokens": 155308218.0, + "step": 6099 + }, + { + "epoch": 0.6698879859433341, + "grad_norm": 2.216503620147705, + "learning_rate": 1e-06, + "loss": 1.0777, + "mean_token_accuracy": 0.6692368984222412, + "num_tokens": 155335289.0, + "step": 6100 + }, + { + "epoch": 0.6699978036459477, + "grad_norm": 2.0616536140441895, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6971909999847412, + "num_tokens": 155362501.0, + "step": 6101 + }, + { + "epoch": 0.6701076213485614, + "grad_norm": 2.329611301422119, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.721944272518158, + "num_tokens": 155384892.0, + "step": 6102 + }, + { + "epoch": 0.670217439051175, + "grad_norm": 2.364722728729248, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7173677682876587, + "num_tokens": 155406849.0, + "step": 6103 + }, + { + "epoch": 0.6703272567537887, + "grad_norm": 2.1757678985595703, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7040430307388306, + "num_tokens": 155431977.0, + "step": 6104 + }, + { + "epoch": 0.6704370744564023, + "grad_norm": 2.542539358139038, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7392157316207886, + "num_tokens": 155451797.0, + "step": 6105 + }, + { + "epoch": 0.6705468921590161, + "grad_norm": 2.424347400665283, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7136365175247192, + "num_tokens": 155471957.0, + "step": 6106 + }, + { + "epoch": 0.6706567098616297, + "grad_norm": 2.41558575630188, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7209771275520325, + "num_tokens": 155493647.0, + "step": 6107 + }, + { + "epoch": 0.6707665275642434, + "grad_norm": 2.3263840675354004, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7040699124336243, + "num_tokens": 155516378.0, + "step": 6108 + }, + { + "epoch": 0.670876345266857, + "grad_norm": 2.2826240062713623, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6944842338562012, + "num_tokens": 155540711.0, + "step": 6109 + }, + { + "epoch": 0.6709861629694707, + "grad_norm": 2.424107789993286, + "learning_rate": 1e-06, + "loss": 1.0704, + "mean_token_accuracy": 0.6834650039672852, + "num_tokens": 155562896.0, + "step": 6110 + }, + { + "epoch": 0.6710959806720843, + "grad_norm": 2.0713539123535156, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.6974659562110901, + "num_tokens": 155591348.0, + "step": 6111 + }, + { + "epoch": 0.671205798374698, + "grad_norm": 2.1898322105407715, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.6992558836936951, + "num_tokens": 155619106.0, + "step": 6112 + }, + { + "epoch": 0.6713156160773117, + "grad_norm": 2.184030771255493, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6940735578536987, + "num_tokens": 155645443.0, + "step": 6113 + }, + { + "epoch": 0.6714254337799254, + "grad_norm": 2.335460662841797, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7181074023246765, + "num_tokens": 155667001.0, + "step": 6114 + }, + { + "epoch": 0.671535251482539, + "grad_norm": 2.4634222984313965, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7190127372741699, + "num_tokens": 155687567.0, + "step": 6115 + }, + { + "epoch": 0.6716450691851527, + "grad_norm": 2.1443164348602295, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7208620309829712, + "num_tokens": 155712523.0, + "step": 6116 + }, + { + "epoch": 0.6717548868877663, + "grad_norm": 2.494619369506836, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.720451831817627, + "num_tokens": 155731917.0, + "step": 6117 + }, + { + "epoch": 0.67186470459038, + "grad_norm": 2.0980215072631836, + "learning_rate": 1e-06, + "loss": 1.0522, + "mean_token_accuracy": 0.6980270743370056, + "num_tokens": 155761040.0, + "step": 6118 + }, + { + "epoch": 0.6719745222929936, + "grad_norm": 2.1406116485595703, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7053552865982056, + "num_tokens": 155785845.0, + "step": 6119 + }, + { + "epoch": 0.6720843399956072, + "grad_norm": 1.9117339849472046, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.710435152053833, + "num_tokens": 155817660.0, + "step": 6120 + }, + { + "epoch": 0.672194157698221, + "grad_norm": 2.096979856491089, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.71095210313797, + "num_tokens": 155843523.0, + "step": 6121 + }, + { + "epoch": 0.6723039754008346, + "grad_norm": 2.2226109504699707, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.71787428855896, + "num_tokens": 155867096.0, + "step": 6122 + }, + { + "epoch": 0.6724137931034483, + "grad_norm": 2.282381057739258, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.698482096195221, + "num_tokens": 155892337.0, + "step": 6123 + }, + { + "epoch": 0.6725236108060619, + "grad_norm": 2.424211025238037, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7380184531211853, + "num_tokens": 155910939.0, + "step": 6124 + }, + { + "epoch": 0.6726334285086756, + "grad_norm": 2.1011111736297607, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7031238079071045, + "num_tokens": 155938324.0, + "step": 6125 + }, + { + "epoch": 0.6727432462112892, + "grad_norm": 2.5398571491241455, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7266998291015625, + "num_tokens": 155956930.0, + "step": 6126 + }, + { + "epoch": 0.6728530639139029, + "grad_norm": 2.231703758239746, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7270634174346924, + "num_tokens": 155980265.0, + "step": 6127 + }, + { + "epoch": 0.6729628816165166, + "grad_norm": 2.3907556533813477, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7320379614830017, + "num_tokens": 156001527.0, + "step": 6128 + }, + { + "epoch": 0.6730726993191303, + "grad_norm": 2.0601720809936523, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7076827883720398, + "num_tokens": 156029386.0, + "step": 6129 + }, + { + "epoch": 0.6731825170217439, + "grad_norm": 2.3460018634796143, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.6964734196662903, + "num_tokens": 156054357.0, + "step": 6130 + }, + { + "epoch": 0.6732923347243576, + "grad_norm": 2.1950855255126953, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7435694932937622, + "num_tokens": 156077074.0, + "step": 6131 + }, + { + "epoch": 0.6734021524269712, + "grad_norm": 2.3168325424194336, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6981664896011353, + "num_tokens": 156100431.0, + "step": 6132 + }, + { + "epoch": 0.6735119701295849, + "grad_norm": 2.6200170516967773, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7202459573745728, + "num_tokens": 156119128.0, + "step": 6133 + }, + { + "epoch": 0.6736217878321985, + "grad_norm": 2.1068503856658936, + "learning_rate": 1e-06, + "loss": 1.0376, + "mean_token_accuracy": 0.6928458213806152, + "num_tokens": 156147941.0, + "step": 6134 + }, + { + "epoch": 0.6737316055348123, + "grad_norm": 2.0718095302581787, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7194316387176514, + "num_tokens": 156175356.0, + "step": 6135 + }, + { + "epoch": 0.6738414232374259, + "grad_norm": 2.3388869762420654, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7069599628448486, + "num_tokens": 156198875.0, + "step": 6136 + }, + { + "epoch": 0.6739512409400396, + "grad_norm": 2.4592740535736084, + "learning_rate": 1e-06, + "loss": 1.0224, + "mean_token_accuracy": 0.6855199337005615, + "num_tokens": 156220048.0, + "step": 6137 + }, + { + "epoch": 0.6740610586426532, + "grad_norm": 2.4031453132629395, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7219099998474121, + "num_tokens": 156241155.0, + "step": 6138 + }, + { + "epoch": 0.6741708763452668, + "grad_norm": 2.265153408050537, + "learning_rate": 1e-06, + "loss": 1.0822, + "mean_token_accuracy": 0.6788753867149353, + "num_tokens": 156265007.0, + "step": 6139 + }, + { + "epoch": 0.6742806940478805, + "grad_norm": 2.2437551021575928, + "learning_rate": 1e-06, + "loss": 1.0276, + "mean_token_accuracy": 0.6920266151428223, + "num_tokens": 156290256.0, + "step": 6140 + }, + { + "epoch": 0.6743905117504941, + "grad_norm": 2.146214485168457, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7309791445732117, + "num_tokens": 156316134.0, + "step": 6141 + }, + { + "epoch": 0.6745003294531079, + "grad_norm": 2.136887311935425, + "learning_rate": 1e-06, + "loss": 1.0501, + "mean_token_accuracy": 0.6823259592056274, + "num_tokens": 156345855.0, + "step": 6142 + }, + { + "epoch": 0.6746101471557215, + "grad_norm": 2.182039260864258, + "learning_rate": 1e-06, + "loss": 1.032, + "mean_token_accuracy": 0.6849527359008789, + "num_tokens": 156371341.0, + "step": 6143 + }, + { + "epoch": 0.6747199648583352, + "grad_norm": 2.440941572189331, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.7036634683609009, + "num_tokens": 156393612.0, + "step": 6144 + }, + { + "epoch": 0.6748297825609488, + "grad_norm": 2.4594833850860596, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7192840576171875, + "num_tokens": 156413152.0, + "step": 6145 + }, + { + "epoch": 0.6749396002635625, + "grad_norm": 2.3903913497924805, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7099069952964783, + "num_tokens": 156435666.0, + "step": 6146 + }, + { + "epoch": 0.6750494179661761, + "grad_norm": 1.9961105585098267, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.7012376189231873, + "num_tokens": 156466423.0, + "step": 6147 + }, + { + "epoch": 0.6751592356687898, + "grad_norm": 2.349738359451294, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.6985093951225281, + "num_tokens": 156488256.0, + "step": 6148 + }, + { + "epoch": 0.6752690533714034, + "grad_norm": 1.987370252609253, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7138999104499817, + "num_tokens": 156518249.0, + "step": 6149 + }, + { + "epoch": 0.6753788710740172, + "grad_norm": 2.350358009338379, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7296736240386963, + "num_tokens": 156540588.0, + "step": 6150 + }, + { + "epoch": 0.6754886887766308, + "grad_norm": 2.3490917682647705, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.6884280443191528, + "num_tokens": 156564028.0, + "step": 6151 + }, + { + "epoch": 0.6755985064792445, + "grad_norm": 1.881812572479248, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7141956090927124, + "num_tokens": 156595552.0, + "step": 6152 + }, + { + "epoch": 0.6757083241818581, + "grad_norm": 2.278151512145996, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.705305814743042, + "num_tokens": 156619074.0, + "step": 6153 + }, + { + "epoch": 0.6758181418844718, + "grad_norm": 1.9232196807861328, + "learning_rate": 1e-06, + "loss": 1.0668, + "mean_token_accuracy": 0.681428849697113, + "num_tokens": 156651146.0, + "step": 6154 + }, + { + "epoch": 0.6759279595870854, + "grad_norm": 2.5473577976226807, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7154257893562317, + "num_tokens": 156670452.0, + "step": 6155 + }, + { + "epoch": 0.676037777289699, + "grad_norm": 2.104796886444092, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7053748369216919, + "num_tokens": 156697171.0, + "step": 6156 + }, + { + "epoch": 0.6761475949923128, + "grad_norm": 2.367227077484131, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7009563446044922, + "num_tokens": 156720116.0, + "step": 6157 + }, + { + "epoch": 0.6762574126949265, + "grad_norm": 1.96258544921875, + "learning_rate": 1e-06, + "loss": 1.0379, + "mean_token_accuracy": 0.6881696581840515, + "num_tokens": 156752739.0, + "step": 6158 + }, + { + "epoch": 0.6763672303975401, + "grad_norm": 2.419694185256958, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7179490327835083, + "num_tokens": 156772735.0, + "step": 6159 + }, + { + "epoch": 0.6764770481001537, + "grad_norm": 2.428677558898926, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7152041792869568, + "num_tokens": 156793954.0, + "step": 6160 + }, + { + "epoch": 0.6765868658027674, + "grad_norm": 2.1934287548065186, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7023723125457764, + "num_tokens": 156818819.0, + "step": 6161 + }, + { + "epoch": 0.676696683505381, + "grad_norm": 1.9579942226409912, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7119541168212891, + "num_tokens": 156848907.0, + "step": 6162 + }, + { + "epoch": 0.6768065012079947, + "grad_norm": 2.3274059295654297, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7070493102073669, + "num_tokens": 156871508.0, + "step": 6163 + }, + { + "epoch": 0.6769163189106084, + "grad_norm": 2.3508639335632324, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7180155515670776, + "num_tokens": 156894731.0, + "step": 6164 + }, + { + "epoch": 0.6770261366132221, + "grad_norm": 2.2357771396636963, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.6984133720397949, + "num_tokens": 156919378.0, + "step": 6165 + }, + { + "epoch": 0.6771359543158357, + "grad_norm": 2.308454751968384, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6959478855133057, + "num_tokens": 156943060.0, + "step": 6166 + }, + { + "epoch": 0.6772457720184494, + "grad_norm": 1.8819262981414795, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7171944975852966, + "num_tokens": 156975965.0, + "step": 6167 + }, + { + "epoch": 0.677355589721063, + "grad_norm": 2.0577235221862793, + "learning_rate": 1e-06, + "loss": 1.0735, + "mean_token_accuracy": 0.680073082447052, + "num_tokens": 157006282.0, + "step": 6168 + }, + { + "epoch": 0.6774654074236767, + "grad_norm": 1.9483376741409302, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6902746558189392, + "num_tokens": 157038640.0, + "step": 6169 + }, + { + "epoch": 0.6775752251262903, + "grad_norm": 2.3106372356414795, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7231247425079346, + "num_tokens": 157060934.0, + "step": 6170 + }, + { + "epoch": 0.6776850428289041, + "grad_norm": 2.135838270187378, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7173647880554199, + "num_tokens": 157085517.0, + "step": 6171 + }, + { + "epoch": 0.6777948605315177, + "grad_norm": 1.8833699226379395, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7029805183410645, + "num_tokens": 157118666.0, + "step": 6172 + }, + { + "epoch": 0.6779046782341314, + "grad_norm": 2.264364719390869, + "learning_rate": 1e-06, + "loss": 1.0484, + "mean_token_accuracy": 0.6862589716911316, + "num_tokens": 157143505.0, + "step": 6173 + }, + { + "epoch": 0.678014495936745, + "grad_norm": 2.014488935470581, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.7023100852966309, + "num_tokens": 157173396.0, + "step": 6174 + }, + { + "epoch": 0.6781243136393587, + "grad_norm": 1.9946932792663574, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.7100695967674255, + "num_tokens": 157204493.0, + "step": 6175 + }, + { + "epoch": 0.6782341313419723, + "grad_norm": 2.1989636421203613, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7252498865127563, + "num_tokens": 157228228.0, + "step": 6176 + }, + { + "epoch": 0.678343949044586, + "grad_norm": 1.9133477210998535, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.7045310139656067, + "num_tokens": 157259554.0, + "step": 6177 + }, + { + "epoch": 0.6784537667471996, + "grad_norm": 2.1818525791168213, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7089733481407166, + "num_tokens": 157283271.0, + "step": 6178 + }, + { + "epoch": 0.6785635844498134, + "grad_norm": 2.2030491828918457, + "learning_rate": 1e-06, + "loss": 1.0506, + "mean_token_accuracy": 0.6838006377220154, + "num_tokens": 157310699.0, + "step": 6179 + }, + { + "epoch": 0.678673402152427, + "grad_norm": 2.4866652488708496, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7302458882331848, + "num_tokens": 157330983.0, + "step": 6180 + }, + { + "epoch": 0.6787832198550406, + "grad_norm": 1.9159984588623047, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7115062475204468, + "num_tokens": 157361715.0, + "step": 6181 + }, + { + "epoch": 0.6788930375576543, + "grad_norm": 2.2619965076446533, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7048338651657104, + "num_tokens": 157384919.0, + "step": 6182 + }, + { + "epoch": 0.6790028552602679, + "grad_norm": 2.1866583824157715, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7078567147254944, + "num_tokens": 157410742.0, + "step": 6183 + }, + { + "epoch": 0.6791126729628816, + "grad_norm": 2.3655037879943848, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.726513147354126, + "num_tokens": 157432309.0, + "step": 6184 + }, + { + "epoch": 0.6792224906654952, + "grad_norm": 2.0280983448028564, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.6930242776870728, + "num_tokens": 157462860.0, + "step": 6185 + }, + { + "epoch": 0.679332308368109, + "grad_norm": 2.136153221130371, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6844826340675354, + "num_tokens": 157491651.0, + "step": 6186 + }, + { + "epoch": 0.6794421260707226, + "grad_norm": 2.165350914001465, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7162374258041382, + "num_tokens": 157515110.0, + "step": 6187 + }, + { + "epoch": 0.6795519437733363, + "grad_norm": 2.221503496170044, + "learning_rate": 1e-06, + "loss": 0.7734, + "mean_token_accuracy": 0.7525466680526733, + "num_tokens": 157536976.0, + "step": 6188 + }, + { + "epoch": 0.6796617614759499, + "grad_norm": 2.1187820434570312, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7168383598327637, + "num_tokens": 157563728.0, + "step": 6189 + }, + { + "epoch": 0.6797715791785636, + "grad_norm": 2.365901231765747, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7201736569404602, + "num_tokens": 157586740.0, + "step": 6190 + }, + { + "epoch": 0.6798813968811772, + "grad_norm": 2.426006555557251, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7229608297348022, + "num_tokens": 157608122.0, + "step": 6191 + }, + { + "epoch": 0.6799912145837909, + "grad_norm": 2.355384111404419, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7092971801757812, + "num_tokens": 157630202.0, + "step": 6192 + }, + { + "epoch": 0.6801010322864046, + "grad_norm": 2.261967420578003, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7174103260040283, + "num_tokens": 157653329.0, + "step": 6193 + }, + { + "epoch": 0.6802108499890183, + "grad_norm": 2.164555072784424, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.6953198909759521, + "num_tokens": 157679325.0, + "step": 6194 + }, + { + "epoch": 0.6803206676916319, + "grad_norm": 2.0781118869781494, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.6982992887496948, + "num_tokens": 157707403.0, + "step": 6195 + }, + { + "epoch": 0.6804304853942456, + "grad_norm": 2.0885725021362305, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6985945701599121, + "num_tokens": 157736468.0, + "step": 6196 + }, + { + "epoch": 0.6805403030968592, + "grad_norm": 2.1262567043304443, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.6981431245803833, + "num_tokens": 157761888.0, + "step": 6197 + }, + { + "epoch": 0.6806501207994728, + "grad_norm": 2.1339008808135986, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7102316617965698, + "num_tokens": 157788100.0, + "step": 6198 + }, + { + "epoch": 0.6807599385020865, + "grad_norm": 2.643580913543701, + "learning_rate": 1e-06, + "loss": 1.0196, + "mean_token_accuracy": 0.6910790801048279, + "num_tokens": 157809134.0, + "step": 6199 + }, + { + "epoch": 0.6808697562047002, + "grad_norm": 2.2871222496032715, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6966292858123779, + "num_tokens": 157833884.0, + "step": 6200 + }, + { + "epoch": 0.6809795739073139, + "grad_norm": 1.9680848121643066, + "learning_rate": 1e-06, + "loss": 1.0254, + "mean_token_accuracy": 0.6894788146018982, + "num_tokens": 157864844.0, + "step": 6201 + }, + { + "epoch": 0.6810893916099275, + "grad_norm": 2.3363959789276123, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7302976846694946, + "num_tokens": 157886951.0, + "step": 6202 + }, + { + "epoch": 0.6811992093125412, + "grad_norm": 2.112278938293457, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6904007792472839, + "num_tokens": 157915537.0, + "step": 6203 + }, + { + "epoch": 0.6813090270151548, + "grad_norm": 2.19818377494812, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7107678651809692, + "num_tokens": 157940012.0, + "step": 6204 + }, + { + "epoch": 0.6814188447177685, + "grad_norm": 2.160987377166748, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.7014005184173584, + "num_tokens": 157963743.0, + "step": 6205 + }, + { + "epoch": 0.6815286624203821, + "grad_norm": 1.9942171573638916, + "learning_rate": 1e-06, + "loss": 1.0709, + "mean_token_accuracy": 0.6784721612930298, + "num_tokens": 157992782.0, + "step": 6206 + }, + { + "epoch": 0.6816384801229959, + "grad_norm": 2.185769557952881, + "learning_rate": 1e-06, + "loss": 1.0388, + "mean_token_accuracy": 0.6852465867996216, + "num_tokens": 158018632.0, + "step": 6207 + }, + { + "epoch": 0.6817482978256095, + "grad_norm": 2.16159987449646, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.712181568145752, + "num_tokens": 158045745.0, + "step": 6208 + }, + { + "epoch": 0.6818581155282232, + "grad_norm": 2.1810619831085205, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7367932796478271, + "num_tokens": 158070095.0, + "step": 6209 + }, + { + "epoch": 0.6819679332308368, + "grad_norm": 2.102966785430908, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6931432485580444, + "num_tokens": 158096648.0, + "step": 6210 + }, + { + "epoch": 0.6820777509334505, + "grad_norm": 2.1728832721710205, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.6838582754135132, + "num_tokens": 158122924.0, + "step": 6211 + }, + { + "epoch": 0.6821875686360641, + "grad_norm": 2.267461061477661, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.6901230812072754, + "num_tokens": 158148452.0, + "step": 6212 + }, + { + "epoch": 0.6822973863386778, + "grad_norm": 2.263639450073242, + "learning_rate": 1e-06, + "loss": 1.0243, + "mean_token_accuracy": 0.6893172264099121, + "num_tokens": 158173983.0, + "step": 6213 + }, + { + "epoch": 0.6824072040412914, + "grad_norm": 2.327577590942383, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7065100073814392, + "num_tokens": 158197354.0, + "step": 6214 + }, + { + "epoch": 0.6825170217439052, + "grad_norm": 2.4039907455444336, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7310289740562439, + "num_tokens": 158217859.0, + "step": 6215 + }, + { + "epoch": 0.6826268394465188, + "grad_norm": 2.4488415718078613, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7266659736633301, + "num_tokens": 158237745.0, + "step": 6216 + }, + { + "epoch": 0.6827366571491325, + "grad_norm": 2.6307780742645264, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7130966186523438, + "num_tokens": 158258573.0, + "step": 6217 + }, + { + "epoch": 0.6828464748517461, + "grad_norm": 2.009042501449585, + "learning_rate": 1e-06, + "loss": 1.0688, + "mean_token_accuracy": 0.6809732913970947, + "num_tokens": 158289911.0, + "step": 6218 + }, + { + "epoch": 0.6829562925543597, + "grad_norm": 2.2827889919281006, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.711754560470581, + "num_tokens": 158313256.0, + "step": 6219 + }, + { + "epoch": 0.6830661102569734, + "grad_norm": 2.3499577045440674, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7035636305809021, + "num_tokens": 158335812.0, + "step": 6220 + }, + { + "epoch": 0.683175927959587, + "grad_norm": 1.9225406646728516, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.6962249279022217, + "num_tokens": 158367351.0, + "step": 6221 + }, + { + "epoch": 0.6832857456622008, + "grad_norm": 2.154047966003418, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7180154323577881, + "num_tokens": 158395050.0, + "step": 6222 + }, + { + "epoch": 0.6833955633648144, + "grad_norm": 2.4358577728271484, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7150819301605225, + "num_tokens": 158415968.0, + "step": 6223 + }, + { + "epoch": 0.6835053810674281, + "grad_norm": 2.269925594329834, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.7086951732635498, + "num_tokens": 158441730.0, + "step": 6224 + }, + { + "epoch": 0.6836151987700417, + "grad_norm": 2.3748936653137207, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.712169885635376, + "num_tokens": 158463354.0, + "step": 6225 + }, + { + "epoch": 0.6837250164726554, + "grad_norm": 2.1690096855163574, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7036402821540833, + "num_tokens": 158488721.0, + "step": 6226 + }, + { + "epoch": 0.683834834175269, + "grad_norm": 2.1592929363250732, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.6839287877082825, + "num_tokens": 158517321.0, + "step": 6227 + }, + { + "epoch": 0.6839446518778827, + "grad_norm": 2.2337071895599365, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7202606797218323, + "num_tokens": 158540438.0, + "step": 6228 + }, + { + "epoch": 0.6840544695804964, + "grad_norm": 2.119614362716675, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7108899354934692, + "num_tokens": 158567273.0, + "step": 6229 + }, + { + "epoch": 0.6841642872831101, + "grad_norm": 2.001525640487671, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7188804149627686, + "num_tokens": 158595774.0, + "step": 6230 + }, + { + "epoch": 0.6842741049857237, + "grad_norm": 2.184128761291504, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7205633521080017, + "num_tokens": 158622142.0, + "step": 6231 + }, + { + "epoch": 0.6843839226883374, + "grad_norm": 1.9594134092330933, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.6914981603622437, + "num_tokens": 158654699.0, + "step": 6232 + }, + { + "epoch": 0.684493740390951, + "grad_norm": 2.5499913692474365, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7281786203384399, + "num_tokens": 158674028.0, + "step": 6233 + }, + { + "epoch": 0.6846035580935647, + "grad_norm": 2.2461307048797607, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7180688977241516, + "num_tokens": 158697284.0, + "step": 6234 + }, + { + "epoch": 0.6847133757961783, + "grad_norm": 2.3886923789978027, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7028400897979736, + "num_tokens": 158719759.0, + "step": 6235 + }, + { + "epoch": 0.6848231934987921, + "grad_norm": 2.065404176712036, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7159984707832336, + "num_tokens": 158745847.0, + "step": 6236 + }, + { + "epoch": 0.6849330112014057, + "grad_norm": 2.3332679271698, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7242967486381531, + "num_tokens": 158767760.0, + "step": 6237 + }, + { + "epoch": 0.6850428289040194, + "grad_norm": 2.212671995162964, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7269200086593628, + "num_tokens": 158791998.0, + "step": 6238 + }, + { + "epoch": 0.685152646606633, + "grad_norm": 2.4416913986206055, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7114020586013794, + "num_tokens": 158814050.0, + "step": 6239 + }, + { + "epoch": 0.6852624643092466, + "grad_norm": 2.261214256286621, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7101662158966064, + "num_tokens": 158838755.0, + "step": 6240 + }, + { + "epoch": 0.6853722820118603, + "grad_norm": 2.039560079574585, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7122845649719238, + "num_tokens": 158866355.0, + "step": 6241 + }, + { + "epoch": 0.6854820997144739, + "grad_norm": 2.121002197265625, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7079104781150818, + "num_tokens": 158892099.0, + "step": 6242 + }, + { + "epoch": 0.6855919174170876, + "grad_norm": 1.9476513862609863, + "learning_rate": 1e-06, + "loss": 1.0054, + "mean_token_accuracy": 0.693428099155426, + "num_tokens": 158923496.0, + "step": 6243 + }, + { + "epoch": 0.6857017351197013, + "grad_norm": 2.196382761001587, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7272992134094238, + "num_tokens": 158947501.0, + "step": 6244 + }, + { + "epoch": 0.685811552822315, + "grad_norm": 2.1481680870056152, + "learning_rate": 1e-06, + "loss": 1.0621, + "mean_token_accuracy": 0.6784864664077759, + "num_tokens": 158974263.0, + "step": 6245 + }, + { + "epoch": 0.6859213705249286, + "grad_norm": 2.052994728088379, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7085851430892944, + "num_tokens": 159004302.0, + "step": 6246 + }, + { + "epoch": 0.6860311882275423, + "grad_norm": 2.3695085048675537, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.720885157585144, + "num_tokens": 159025825.0, + "step": 6247 + }, + { + "epoch": 0.6861410059301559, + "grad_norm": 2.120736598968506, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7321648597717285, + "num_tokens": 159050200.0, + "step": 6248 + }, + { + "epoch": 0.6862508236327696, + "grad_norm": 2.0403287410736084, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.7040919065475464, + "num_tokens": 159077522.0, + "step": 6249 + }, + { + "epoch": 0.6863606413353832, + "grad_norm": 2.2318594455718994, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7121023535728455, + "num_tokens": 159101902.0, + "step": 6250 + }, + { + "epoch": 0.686470459037997, + "grad_norm": 1.9519472122192383, + "learning_rate": 1e-06, + "loss": 1.0585, + "mean_token_accuracy": 0.6788508296012878, + "num_tokens": 159135977.0, + "step": 6251 + }, + { + "epoch": 0.6865802767406106, + "grad_norm": 2.056302309036255, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.6953550577163696, + "num_tokens": 159165282.0, + "step": 6252 + }, + { + "epoch": 0.6866900944432243, + "grad_norm": 2.223318576812744, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7025365233421326, + "num_tokens": 159189650.0, + "step": 6253 + }, + { + "epoch": 0.6867999121458379, + "grad_norm": 2.212156295776367, + "learning_rate": 1e-06, + "loss": 1.0544, + "mean_token_accuracy": 0.6794289350509644, + "num_tokens": 159216609.0, + "step": 6254 + }, + { + "epoch": 0.6869097298484516, + "grad_norm": 2.2184879779815674, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.6974414587020874, + "num_tokens": 159243182.0, + "step": 6255 + }, + { + "epoch": 0.6870195475510652, + "grad_norm": 2.2845616340637207, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7230355143547058, + "num_tokens": 159265640.0, + "step": 6256 + }, + { + "epoch": 0.6871293652536788, + "grad_norm": 2.082573890686035, + "learning_rate": 1e-06, + "loss": 1.0138, + "mean_token_accuracy": 0.6925241947174072, + "num_tokens": 159293988.0, + "step": 6257 + }, + { + "epoch": 0.6872391829562926, + "grad_norm": 2.0097129344940186, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6836215853691101, + "num_tokens": 159323454.0, + "step": 6258 + }, + { + "epoch": 0.6873490006589063, + "grad_norm": 1.963931918144226, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.6960998177528381, + "num_tokens": 159354179.0, + "step": 6259 + }, + { + "epoch": 0.6874588183615199, + "grad_norm": 2.1006367206573486, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.731300950050354, + "num_tokens": 159380146.0, + "step": 6260 + }, + { + "epoch": 0.6875686360641335, + "grad_norm": 2.086862802505493, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6927638053894043, + "num_tokens": 159408631.0, + "step": 6261 + }, + { + "epoch": 0.6876784537667472, + "grad_norm": 2.3575899600982666, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7136320471763611, + "num_tokens": 159431328.0, + "step": 6262 + }, + { + "epoch": 0.6877882714693608, + "grad_norm": 2.123359203338623, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.714857816696167, + "num_tokens": 159457454.0, + "step": 6263 + }, + { + "epoch": 0.6878980891719745, + "grad_norm": 2.8343350887298584, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7372375726699829, + "num_tokens": 159474376.0, + "step": 6264 + }, + { + "epoch": 0.6880079068745882, + "grad_norm": 2.4208920001983643, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7094548344612122, + "num_tokens": 159495186.0, + "step": 6265 + }, + { + "epoch": 0.6881177245772019, + "grad_norm": 1.9358747005462646, + "learning_rate": 1e-06, + "loss": 1.0498, + "mean_token_accuracy": 0.6924189329147339, + "num_tokens": 159526962.0, + "step": 6266 + }, + { + "epoch": 0.6882275422798155, + "grad_norm": 2.1603448390960693, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7237236499786377, + "num_tokens": 159551697.0, + "step": 6267 + }, + { + "epoch": 0.6883373599824292, + "grad_norm": 2.2150862216949463, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7082298994064331, + "num_tokens": 159577420.0, + "step": 6268 + }, + { + "epoch": 0.6884471776850428, + "grad_norm": 2.227494478225708, + "learning_rate": 1e-06, + "loss": 0.9946, + "mean_token_accuracy": 0.7009016275405884, + "num_tokens": 159602129.0, + "step": 6269 + }, + { + "epoch": 0.6885569953876565, + "grad_norm": 2.2839736938476562, + "learning_rate": 1e-06, + "loss": 1.0395, + "mean_token_accuracy": 0.6863727569580078, + "num_tokens": 159626114.0, + "step": 6270 + }, + { + "epoch": 0.6886668130902701, + "grad_norm": 2.35009765625, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7157199382781982, + "num_tokens": 159648434.0, + "step": 6271 + }, + { + "epoch": 0.6887766307928838, + "grad_norm": 2.2859089374542236, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7239720225334167, + "num_tokens": 159671173.0, + "step": 6272 + }, + { + "epoch": 0.6888864484954975, + "grad_norm": 1.921225666999817, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6930360198020935, + "num_tokens": 159702129.0, + "step": 6273 + }, + { + "epoch": 0.6889962661981112, + "grad_norm": 2.2363297939300537, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.7014937400817871, + "num_tokens": 159726615.0, + "step": 6274 + }, + { + "epoch": 0.6891060839007248, + "grad_norm": 2.307699203491211, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7215331792831421, + "num_tokens": 159750512.0, + "step": 6275 + }, + { + "epoch": 0.6892159016033385, + "grad_norm": 1.924512505531311, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7192262411117554, + "num_tokens": 159779843.0, + "step": 6276 + }, + { + "epoch": 0.6893257193059521, + "grad_norm": 2.0697009563446045, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6939526200294495, + "num_tokens": 159808247.0, + "step": 6277 + }, + { + "epoch": 0.6894355370085657, + "grad_norm": 1.9948391914367676, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7053882479667664, + "num_tokens": 159837581.0, + "step": 6278 + }, + { + "epoch": 0.6895453547111794, + "grad_norm": 2.0306622982025146, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7001131176948547, + "num_tokens": 159868310.0, + "step": 6279 + }, + { + "epoch": 0.6896551724137931, + "grad_norm": 2.294332265853882, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7168089151382446, + "num_tokens": 159890362.0, + "step": 6280 + }, + { + "epoch": 0.6897649901164068, + "grad_norm": 2.4448740482330322, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7032275795936584, + "num_tokens": 159912579.0, + "step": 6281 + }, + { + "epoch": 0.6898748078190204, + "grad_norm": 2.12170672416687, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.704905092716217, + "num_tokens": 159937752.0, + "step": 6282 + }, + { + "epoch": 0.6899846255216341, + "grad_norm": 1.7794603109359741, + "learning_rate": 1e-06, + "loss": 1.0559, + "mean_token_accuracy": 0.6829664707183838, + "num_tokens": 159976992.0, + "step": 6283 + }, + { + "epoch": 0.6900944432242477, + "grad_norm": 2.422135829925537, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7287559509277344, + "num_tokens": 159997937.0, + "step": 6284 + }, + { + "epoch": 0.6902042609268614, + "grad_norm": 2.4622092247009277, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7240056991577148, + "num_tokens": 160018225.0, + "step": 6285 + }, + { + "epoch": 0.690314078629475, + "grad_norm": 2.181407928466797, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7087566256523132, + "num_tokens": 160042143.0, + "step": 6286 + }, + { + "epoch": 0.6904238963320888, + "grad_norm": 2.100358724594116, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.724528968334198, + "num_tokens": 160069826.0, + "step": 6287 + }, + { + "epoch": 0.6905337140347024, + "grad_norm": 2.263150930404663, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7090246677398682, + "num_tokens": 160091702.0, + "step": 6288 + }, + { + "epoch": 0.6906435317373161, + "grad_norm": 2.307875871658325, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7124925851821899, + "num_tokens": 160116302.0, + "step": 6289 + }, + { + "epoch": 0.6907533494399297, + "grad_norm": 1.8239245414733887, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7127037048339844, + "num_tokens": 160148628.0, + "step": 6290 + }, + { + "epoch": 0.6908631671425434, + "grad_norm": 2.125544309616089, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7058079242706299, + "num_tokens": 160177229.0, + "step": 6291 + }, + { + "epoch": 0.690972984845157, + "grad_norm": 1.9850479364395142, + "learning_rate": 1e-06, + "loss": 1.0285, + "mean_token_accuracy": 0.6862401962280273, + "num_tokens": 160208721.0, + "step": 6292 + }, + { + "epoch": 0.6910828025477707, + "grad_norm": 2.252035617828369, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7284438610076904, + "num_tokens": 160231991.0, + "step": 6293 + }, + { + "epoch": 0.6911926202503844, + "grad_norm": 2.2392261028289795, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7068508863449097, + "num_tokens": 160257765.0, + "step": 6294 + }, + { + "epoch": 0.6913024379529981, + "grad_norm": 2.369757652282715, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7197952270507812, + "num_tokens": 160278664.0, + "step": 6295 + }, + { + "epoch": 0.6914122556556117, + "grad_norm": 2.234769821166992, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7103670835494995, + "num_tokens": 160304519.0, + "step": 6296 + }, + { + "epoch": 0.6915220733582254, + "grad_norm": 2.3725733757019043, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7057446837425232, + "num_tokens": 160327180.0, + "step": 6297 + }, + { + "epoch": 0.691631891060839, + "grad_norm": 2.234375238418579, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.7071632146835327, + "num_tokens": 160351675.0, + "step": 6298 + }, + { + "epoch": 0.6917417087634526, + "grad_norm": 2.093141794204712, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7384101748466492, + "num_tokens": 160378198.0, + "step": 6299 + }, + { + "epoch": 0.6918515264660663, + "grad_norm": 2.2050065994262695, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.698325514793396, + "num_tokens": 160404028.0, + "step": 6300 + }, + { + "epoch": 0.6919613441686799, + "grad_norm": 1.787622094154358, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7143218517303467, + "num_tokens": 160442804.0, + "step": 6301 + }, + { + "epoch": 0.6920711618712937, + "grad_norm": 2.514036178588867, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.731803297996521, + "num_tokens": 160463034.0, + "step": 6302 + }, + { + "epoch": 0.6921809795739073, + "grad_norm": 2.201202154159546, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7183652520179749, + "num_tokens": 160487645.0, + "step": 6303 + }, + { + "epoch": 0.692290797276521, + "grad_norm": 1.8211853504180908, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.6926117539405823, + "num_tokens": 160521690.0, + "step": 6304 + }, + { + "epoch": 0.6924006149791346, + "grad_norm": 1.9797546863555908, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7219934463500977, + "num_tokens": 160550178.0, + "step": 6305 + }, + { + "epoch": 0.6925104326817483, + "grad_norm": 2.17942476272583, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7211498022079468, + "num_tokens": 160575964.0, + "step": 6306 + }, + { + "epoch": 0.6926202503843619, + "grad_norm": 2.1337199211120605, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7130599021911621, + "num_tokens": 160605379.0, + "step": 6307 + }, + { + "epoch": 0.6927300680869756, + "grad_norm": 2.4649195671081543, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.6993215084075928, + "num_tokens": 160626629.0, + "step": 6308 + }, + { + "epoch": 0.6928398857895893, + "grad_norm": 2.2160351276397705, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.7019721269607544, + "num_tokens": 160655145.0, + "step": 6309 + }, + { + "epoch": 0.692949703492203, + "grad_norm": 2.45605206489563, + "learning_rate": 1e-06, + "loss": 1.0437, + "mean_token_accuracy": 0.6832027435302734, + "num_tokens": 160679014.0, + "step": 6310 + }, + { + "epoch": 0.6930595211948166, + "grad_norm": 2.534733533859253, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7241696119308472, + "num_tokens": 160697872.0, + "step": 6311 + }, + { + "epoch": 0.6931693388974303, + "grad_norm": 2.180872678756714, + "learning_rate": 1e-06, + "loss": 1.1015, + "mean_token_accuracy": 0.6809300184249878, + "num_tokens": 160725543.0, + "step": 6312 + }, + { + "epoch": 0.6932791566000439, + "grad_norm": 2.244377851486206, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7068881392478943, + "num_tokens": 160749253.0, + "step": 6313 + }, + { + "epoch": 0.6933889743026576, + "grad_norm": 2.3177499771118164, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.7062803506851196, + "num_tokens": 160773571.0, + "step": 6314 + }, + { + "epoch": 0.6934987920052712, + "grad_norm": 2.175307273864746, + "learning_rate": 1e-06, + "loss": 0.9846, + "mean_token_accuracy": 0.697152316570282, + "num_tokens": 160798666.0, + "step": 6315 + }, + { + "epoch": 0.693608609707885, + "grad_norm": 2.182025194168091, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.7002059817314148, + "num_tokens": 160825019.0, + "step": 6316 + }, + { + "epoch": 0.6937184274104986, + "grad_norm": 2.2175254821777344, + "learning_rate": 1e-06, + "loss": 1.0323, + "mean_token_accuracy": 0.6868668794631958, + "num_tokens": 160849323.0, + "step": 6317 + }, + { + "epoch": 0.6938282451131123, + "grad_norm": 2.1233274936676025, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.7008892893791199, + "num_tokens": 160875722.0, + "step": 6318 + }, + { + "epoch": 0.6939380628157259, + "grad_norm": 2.0651662349700928, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6880778670310974, + "num_tokens": 160905770.0, + "step": 6319 + }, + { + "epoch": 0.6940478805183395, + "grad_norm": 2.3460397720336914, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.6971244215965271, + "num_tokens": 160928876.0, + "step": 6320 + }, + { + "epoch": 0.6941576982209532, + "grad_norm": 2.2689197063446045, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7068005800247192, + "num_tokens": 160953444.0, + "step": 6321 + }, + { + "epoch": 0.6942675159235668, + "grad_norm": 2.234116792678833, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.7065834999084473, + "num_tokens": 160979041.0, + "step": 6322 + }, + { + "epoch": 0.6943773336261806, + "grad_norm": 2.1610429286956787, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7322707176208496, + "num_tokens": 161002231.0, + "step": 6323 + }, + { + "epoch": 0.6944871513287942, + "grad_norm": 2.4124433994293213, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7163228988647461, + "num_tokens": 161024408.0, + "step": 6324 + }, + { + "epoch": 0.6945969690314079, + "grad_norm": 2.17812442779541, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7404438257217407, + "num_tokens": 161050602.0, + "step": 6325 + }, + { + "epoch": 0.6947067867340215, + "grad_norm": 2.330662250518799, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.710885226726532, + "num_tokens": 161074406.0, + "step": 6326 + }, + { + "epoch": 0.6948166044366352, + "grad_norm": 2.184999942779541, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.701080322265625, + "num_tokens": 161100598.0, + "step": 6327 + }, + { + "epoch": 0.6949264221392488, + "grad_norm": 2.306581497192383, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.711874783039093, + "num_tokens": 161124910.0, + "step": 6328 + }, + { + "epoch": 0.6950362398418625, + "grad_norm": 2.1250252723693848, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6948857307434082, + "num_tokens": 161151933.0, + "step": 6329 + }, + { + "epoch": 0.6951460575444761, + "grad_norm": 2.035127639770508, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6942752599716187, + "num_tokens": 161179837.0, + "step": 6330 + }, + { + "epoch": 0.6952558752470899, + "grad_norm": 2.037318706512451, + "learning_rate": 1e-06, + "loss": 1.132, + "mean_token_accuracy": 0.6662062406539917, + "num_tokens": 161211207.0, + "step": 6331 + }, + { + "epoch": 0.6953656929497035, + "grad_norm": 2.211866617202759, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7063082456588745, + "num_tokens": 161236447.0, + "step": 6332 + }, + { + "epoch": 0.6954755106523172, + "grad_norm": 2.0789968967437744, + "learning_rate": 1e-06, + "loss": 1.0372, + "mean_token_accuracy": 0.6884146332740784, + "num_tokens": 161265282.0, + "step": 6333 + }, + { + "epoch": 0.6955853283549308, + "grad_norm": 2.127329111099243, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7147992253303528, + "num_tokens": 161292907.0, + "step": 6334 + }, + { + "epoch": 0.6956951460575445, + "grad_norm": 2.2191321849823, + "learning_rate": 1e-06, + "loss": 1.0381, + "mean_token_accuracy": 0.6896162629127502, + "num_tokens": 161319853.0, + "step": 6335 + }, + { + "epoch": 0.6958049637601581, + "grad_norm": 2.251519203186035, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7151960134506226, + "num_tokens": 161342273.0, + "step": 6336 + }, + { + "epoch": 0.6959147814627717, + "grad_norm": 2.1596710681915283, + "learning_rate": 1e-06, + "loss": 1.0759, + "mean_token_accuracy": 0.6820095777511597, + "num_tokens": 161369851.0, + "step": 6337 + }, + { + "epoch": 0.6960245991653855, + "grad_norm": 2.363611936569214, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7233536243438721, + "num_tokens": 161391725.0, + "step": 6338 + }, + { + "epoch": 0.6961344168679992, + "grad_norm": 2.271244525909424, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7299278378486633, + "num_tokens": 161412972.0, + "step": 6339 + }, + { + "epoch": 0.6962442345706128, + "grad_norm": 2.1114814281463623, + "learning_rate": 1e-06, + "loss": 1.0711, + "mean_token_accuracy": 0.676865816116333, + "num_tokens": 161442156.0, + "step": 6340 + }, + { + "epoch": 0.6963540522732264, + "grad_norm": 1.979444146156311, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7206439971923828, + "num_tokens": 161471451.0, + "step": 6341 + }, + { + "epoch": 0.6964638699758401, + "grad_norm": 2.214935779571533, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.6933878660202026, + "num_tokens": 161495717.0, + "step": 6342 + }, + { + "epoch": 0.6965736876784537, + "grad_norm": 2.1498351097106934, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7105901837348938, + "num_tokens": 161521870.0, + "step": 6343 + }, + { + "epoch": 0.6966835053810674, + "grad_norm": 2.556140184402466, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7187273502349854, + "num_tokens": 161541243.0, + "step": 6344 + }, + { + "epoch": 0.6967933230836811, + "grad_norm": 2.117359161376953, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.6981490254402161, + "num_tokens": 161567966.0, + "step": 6345 + }, + { + "epoch": 0.6969031407862948, + "grad_norm": 2.6067769527435303, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7157410383224487, + "num_tokens": 161587651.0, + "step": 6346 + }, + { + "epoch": 0.6970129584889084, + "grad_norm": 2.2357985973358154, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6865499019622803, + "num_tokens": 161613359.0, + "step": 6347 + }, + { + "epoch": 0.6971227761915221, + "grad_norm": 2.240811824798584, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7065560817718506, + "num_tokens": 161639034.0, + "step": 6348 + }, + { + "epoch": 0.6972325938941357, + "grad_norm": 2.145678758621216, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7269332408905029, + "num_tokens": 161664324.0, + "step": 6349 + }, + { + "epoch": 0.6973424115967494, + "grad_norm": 2.419384002685547, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.6982718706130981, + "num_tokens": 161686166.0, + "step": 6350 + }, + { + "epoch": 0.697452229299363, + "grad_norm": 2.2590017318725586, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.7019685506820679, + "num_tokens": 161710040.0, + "step": 6351 + }, + { + "epoch": 0.6975620470019768, + "grad_norm": 2.1071696281433105, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6920609474182129, + "num_tokens": 161738295.0, + "step": 6352 + }, + { + "epoch": 0.6976718647045904, + "grad_norm": 2.1693828105926514, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7273552417755127, + "num_tokens": 161763717.0, + "step": 6353 + }, + { + "epoch": 0.6977816824072041, + "grad_norm": 2.462200403213501, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7085055708885193, + "num_tokens": 161783604.0, + "step": 6354 + }, + { + "epoch": 0.6978915001098177, + "grad_norm": 2.2149155139923096, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.7025213837623596, + "num_tokens": 161810651.0, + "step": 6355 + }, + { + "epoch": 0.6980013178124314, + "grad_norm": 2.3216187953948975, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7179657220840454, + "num_tokens": 161833426.0, + "step": 6356 + }, + { + "epoch": 0.698111135515045, + "grad_norm": 2.2193946838378906, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7188597321510315, + "num_tokens": 161858376.0, + "step": 6357 + }, + { + "epoch": 0.6982209532176586, + "grad_norm": 2.157364845275879, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.7042155861854553, + "num_tokens": 161886377.0, + "step": 6358 + }, + { + "epoch": 0.6983307709202724, + "grad_norm": 2.2617616653442383, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6971462965011597, + "num_tokens": 161912127.0, + "step": 6359 + }, + { + "epoch": 0.698440588622886, + "grad_norm": 2.237553834915161, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7050352096557617, + "num_tokens": 161937018.0, + "step": 6360 + }, + { + "epoch": 0.6985504063254997, + "grad_norm": 2.411984443664551, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7165324687957764, + "num_tokens": 161959626.0, + "step": 6361 + }, + { + "epoch": 0.6986602240281133, + "grad_norm": 2.108085870742798, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7085102796554565, + "num_tokens": 161988775.0, + "step": 6362 + }, + { + "epoch": 0.698770041730727, + "grad_norm": 2.2894246578216553, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7200454473495483, + "num_tokens": 162013088.0, + "step": 6363 + }, + { + "epoch": 0.6988798594333406, + "grad_norm": 2.1390433311462402, + "learning_rate": 1e-06, + "loss": 1.0019, + "mean_token_accuracy": 0.6952768564224243, + "num_tokens": 162040479.0, + "step": 6364 + }, + { + "epoch": 0.6989896771359543, + "grad_norm": 2.1697537899017334, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7101900577545166, + "num_tokens": 162064967.0, + "step": 6365 + }, + { + "epoch": 0.6990994948385679, + "grad_norm": 2.481656074523926, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7168276309967041, + "num_tokens": 162086166.0, + "step": 6366 + }, + { + "epoch": 0.6992093125411817, + "grad_norm": 2.367537498474121, + "learning_rate": 1e-06, + "loss": 1.0399, + "mean_token_accuracy": 0.6990793943405151, + "num_tokens": 162109603.0, + "step": 6367 + }, + { + "epoch": 0.6993191302437953, + "grad_norm": 2.0956506729125977, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7164084911346436, + "num_tokens": 162135984.0, + "step": 6368 + }, + { + "epoch": 0.699428947946409, + "grad_norm": 2.2475297451019287, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7143391966819763, + "num_tokens": 162163086.0, + "step": 6369 + }, + { + "epoch": 0.6995387656490226, + "grad_norm": 2.1294655799865723, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7096654772758484, + "num_tokens": 162191057.0, + "step": 6370 + }, + { + "epoch": 0.6996485833516363, + "grad_norm": 2.0567626953125, + "learning_rate": 1e-06, + "loss": 1.0403, + "mean_token_accuracy": 0.6817101836204529, + "num_tokens": 162223074.0, + "step": 6371 + }, + { + "epoch": 0.6997584010542499, + "grad_norm": 2.2569830417633057, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7217506766319275, + "num_tokens": 162246912.0, + "step": 6372 + }, + { + "epoch": 0.6998682187568636, + "grad_norm": 2.66168475151062, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7268968820571899, + "num_tokens": 162265088.0, + "step": 6373 + }, + { + "epoch": 0.6999780364594773, + "grad_norm": 2.1494460105895996, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7123696804046631, + "num_tokens": 162289776.0, + "step": 6374 + }, + { + "epoch": 0.700087854162091, + "grad_norm": 2.0416717529296875, + "learning_rate": 1e-06, + "loss": 1.0375, + "mean_token_accuracy": 0.7005818486213684, + "num_tokens": 162316500.0, + "step": 6375 + }, + { + "epoch": 0.7001976718647046, + "grad_norm": 2.337040662765503, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7061702013015747, + "num_tokens": 162338123.0, + "step": 6376 + }, + { + "epoch": 0.7003074895673183, + "grad_norm": 2.7722654342651367, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7121505737304688, + "num_tokens": 162356123.0, + "step": 6377 + }, + { + "epoch": 0.7004173072699319, + "grad_norm": 2.248490810394287, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7136058211326599, + "num_tokens": 162379558.0, + "step": 6378 + }, + { + "epoch": 0.7005271249725455, + "grad_norm": 2.0115702152252197, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7071109414100647, + "num_tokens": 162412521.0, + "step": 6379 + }, + { + "epoch": 0.7006369426751592, + "grad_norm": 2.1909096240997314, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6889166831970215, + "num_tokens": 162439478.0, + "step": 6380 + }, + { + "epoch": 0.700746760377773, + "grad_norm": 1.9914264678955078, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7409199476242065, + "num_tokens": 162470488.0, + "step": 6381 + }, + { + "epoch": 0.7008565780803866, + "grad_norm": 2.4031424522399902, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7063034772872925, + "num_tokens": 162492692.0, + "step": 6382 + }, + { + "epoch": 0.7009663957830002, + "grad_norm": 1.9944580793380737, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.6933740973472595, + "num_tokens": 162522267.0, + "step": 6383 + }, + { + "epoch": 0.7010762134856139, + "grad_norm": 2.125251531600952, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.7012947797775269, + "num_tokens": 162549930.0, + "step": 6384 + }, + { + "epoch": 0.7011860311882275, + "grad_norm": 2.2995333671569824, + "learning_rate": 1e-06, + "loss": 1.0221, + "mean_token_accuracy": 0.6920923590660095, + "num_tokens": 162577832.0, + "step": 6385 + }, + { + "epoch": 0.7012958488908412, + "grad_norm": 1.9988255500793457, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.6995201110839844, + "num_tokens": 162606351.0, + "step": 6386 + }, + { + "epoch": 0.7014056665934548, + "grad_norm": 2.169553279876709, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.713687539100647, + "num_tokens": 162630522.0, + "step": 6387 + }, + { + "epoch": 0.7015154842960686, + "grad_norm": 2.134643793106079, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7053695917129517, + "num_tokens": 162656084.0, + "step": 6388 + }, + { + "epoch": 0.7016253019986822, + "grad_norm": 2.434917688369751, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7150964140892029, + "num_tokens": 162677685.0, + "step": 6389 + }, + { + "epoch": 0.7017351197012959, + "grad_norm": 2.3213095664978027, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7390432357788086, + "num_tokens": 162700340.0, + "step": 6390 + }, + { + "epoch": 0.7018449374039095, + "grad_norm": 2.0023725032806396, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7237833738327026, + "num_tokens": 162729286.0, + "step": 6391 + }, + { + "epoch": 0.7019547551065232, + "grad_norm": 1.9734764099121094, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.6983680725097656, + "num_tokens": 162760478.0, + "step": 6392 + }, + { + "epoch": 0.7020645728091368, + "grad_norm": 2.164724826812744, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7094494700431824, + "num_tokens": 162786808.0, + "step": 6393 + }, + { + "epoch": 0.7021743905117505, + "grad_norm": 2.124929666519165, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.6999199390411377, + "num_tokens": 162814774.0, + "step": 6394 + }, + { + "epoch": 0.7022842082143641, + "grad_norm": 2.1042940616607666, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.6998729705810547, + "num_tokens": 162841380.0, + "step": 6395 + }, + { + "epoch": 0.7023940259169779, + "grad_norm": 2.190114974975586, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.6945275068283081, + "num_tokens": 162868278.0, + "step": 6396 + }, + { + "epoch": 0.7025038436195915, + "grad_norm": 2.0555450916290283, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7076746225357056, + "num_tokens": 162895107.0, + "step": 6397 + }, + { + "epoch": 0.7026136613222052, + "grad_norm": 1.895875096321106, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.7000973224639893, + "num_tokens": 162929123.0, + "step": 6398 + }, + { + "epoch": 0.7027234790248188, + "grad_norm": 2.281022071838379, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.707388699054718, + "num_tokens": 162952652.0, + "step": 6399 + }, + { + "epoch": 0.7028332967274324, + "grad_norm": 1.9829152822494507, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7083418965339661, + "num_tokens": 162980974.0, + "step": 6400 + }, + { + "epoch": 0.7029431144300461, + "grad_norm": 2.3539862632751465, + "learning_rate": 1e-06, + "loss": 1.027, + "mean_token_accuracy": 0.6887529492378235, + "num_tokens": 163004046.0, + "step": 6401 + }, + { + "epoch": 0.7030529321326597, + "grad_norm": 2.0790586471557617, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.7076027393341064, + "num_tokens": 163031546.0, + "step": 6402 + }, + { + "epoch": 0.7031627498352735, + "grad_norm": 2.7274255752563477, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7253589630126953, + "num_tokens": 163048965.0, + "step": 6403 + }, + { + "epoch": 0.7032725675378871, + "grad_norm": 2.2456700801849365, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.695789098739624, + "num_tokens": 163074682.0, + "step": 6404 + }, + { + "epoch": 0.7033823852405008, + "grad_norm": 2.202343702316284, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7150435447692871, + "num_tokens": 163099457.0, + "step": 6405 + }, + { + "epoch": 0.7034922029431144, + "grad_norm": 2.498229503631592, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7105070352554321, + "num_tokens": 163118260.0, + "step": 6406 + }, + { + "epoch": 0.7036020206457281, + "grad_norm": 2.251166820526123, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7052901983261108, + "num_tokens": 163142072.0, + "step": 6407 + }, + { + "epoch": 0.7037118383483417, + "grad_norm": 2.367033004760742, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7111316919326782, + "num_tokens": 163163601.0, + "step": 6408 + }, + { + "epoch": 0.7038216560509554, + "grad_norm": 2.1839802265167236, + "learning_rate": 1e-06, + "loss": 1.0347, + "mean_token_accuracy": 0.6907438635826111, + "num_tokens": 163190447.0, + "step": 6409 + }, + { + "epoch": 0.7039314737535691, + "grad_norm": 2.14109206199646, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7049497961997986, + "num_tokens": 163216759.0, + "step": 6410 + }, + { + "epoch": 0.7040412914561828, + "grad_norm": 2.4356963634490967, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.710564374923706, + "num_tokens": 163238532.0, + "step": 6411 + }, + { + "epoch": 0.7041511091587964, + "grad_norm": 2.199561357498169, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.721023440361023, + "num_tokens": 163263361.0, + "step": 6412 + }, + { + "epoch": 0.7042609268614101, + "grad_norm": 2.2712459564208984, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7350125908851624, + "num_tokens": 163285823.0, + "step": 6413 + }, + { + "epoch": 0.7043707445640237, + "grad_norm": 2.1131906509399414, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7182515859603882, + "num_tokens": 163311052.0, + "step": 6414 + }, + { + "epoch": 0.7044805622666374, + "grad_norm": 2.1401469707489014, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7195401787757874, + "num_tokens": 163337426.0, + "step": 6415 + }, + { + "epoch": 0.704590379969251, + "grad_norm": 2.387608289718628, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6970933079719543, + "num_tokens": 163361814.0, + "step": 6416 + }, + { + "epoch": 0.7047001976718648, + "grad_norm": 1.8260256052017212, + "learning_rate": 1e-06, + "loss": 1.1132, + "mean_token_accuracy": 0.6704713702201843, + "num_tokens": 163399395.0, + "step": 6417 + }, + { + "epoch": 0.7048100153744784, + "grad_norm": 2.3040146827697754, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.6990999579429626, + "num_tokens": 163422430.0, + "step": 6418 + }, + { + "epoch": 0.704919833077092, + "grad_norm": 3.1010360717773438, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7149873971939087, + "num_tokens": 163443418.0, + "step": 6419 + }, + { + "epoch": 0.7050296507797057, + "grad_norm": 2.4763894081115723, + "learning_rate": 1e-06, + "loss": 1.0296, + "mean_token_accuracy": 0.6865800619125366, + "num_tokens": 163466709.0, + "step": 6420 + }, + { + "epoch": 0.7051394684823193, + "grad_norm": 2.276576280593872, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7008652687072754, + "num_tokens": 163492104.0, + "step": 6421 + }, + { + "epoch": 0.705249286184933, + "grad_norm": 2.0296339988708496, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7185829281806946, + "num_tokens": 163521839.0, + "step": 6422 + }, + { + "epoch": 0.7053591038875466, + "grad_norm": 2.184279203414917, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7123250365257263, + "num_tokens": 163548886.0, + "step": 6423 + }, + { + "epoch": 0.7054689215901603, + "grad_norm": 2.4492106437683105, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.7103342413902283, + "num_tokens": 163570420.0, + "step": 6424 + }, + { + "epoch": 0.705578739292774, + "grad_norm": 2.494922637939453, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6975603103637695, + "num_tokens": 163591161.0, + "step": 6425 + }, + { + "epoch": 0.7056885569953877, + "grad_norm": 2.3473622798919678, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7202482223510742, + "num_tokens": 163612330.0, + "step": 6426 + }, + { + "epoch": 0.7057983746980013, + "grad_norm": 2.0885567665100098, + "learning_rate": 1e-06, + "loss": 1.067, + "mean_token_accuracy": 0.6772365570068359, + "num_tokens": 163641908.0, + "step": 6427 + }, + { + "epoch": 0.705908192400615, + "grad_norm": 2.0724363327026367, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.7043265104293823, + "num_tokens": 163671244.0, + "step": 6428 + }, + { + "epoch": 0.7060180101032286, + "grad_norm": 2.220275640487671, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7025046348571777, + "num_tokens": 163694900.0, + "step": 6429 + }, + { + "epoch": 0.7061278278058423, + "grad_norm": 2.2294652462005615, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7301926612854004, + "num_tokens": 163718615.0, + "step": 6430 + }, + { + "epoch": 0.7062376455084559, + "grad_norm": 2.0282745361328125, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7272387742996216, + "num_tokens": 163745319.0, + "step": 6431 + }, + { + "epoch": 0.7063474632110697, + "grad_norm": 2.1374897956848145, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7066474556922913, + "num_tokens": 163773883.0, + "step": 6432 + }, + { + "epoch": 0.7064572809136833, + "grad_norm": 2.280510902404785, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.6990319490432739, + "num_tokens": 163798405.0, + "step": 6433 + }, + { + "epoch": 0.706567098616297, + "grad_norm": 2.0731139183044434, + "learning_rate": 1e-06, + "loss": 1.0485, + "mean_token_accuracy": 0.7041326761245728, + "num_tokens": 163825865.0, + "step": 6434 + }, + { + "epoch": 0.7066769163189106, + "grad_norm": 2.2903964519500732, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7043052911758423, + "num_tokens": 163848442.0, + "step": 6435 + }, + { + "epoch": 0.7067867340215243, + "grad_norm": 2.0272505283355713, + "learning_rate": 1e-06, + "loss": 1.0252, + "mean_token_accuracy": 0.6853978633880615, + "num_tokens": 163881158.0, + "step": 6436 + }, + { + "epoch": 0.7068965517241379, + "grad_norm": 2.0724453926086426, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.6975486874580383, + "num_tokens": 163911674.0, + "step": 6437 + }, + { + "epoch": 0.7070063694267515, + "grad_norm": 2.392143726348877, + "learning_rate": 1e-06, + "loss": 1.0539, + "mean_token_accuracy": 0.6804927587509155, + "num_tokens": 163936748.0, + "step": 6438 + }, + { + "epoch": 0.7071161871293653, + "grad_norm": 2.2365329265594482, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6901740431785583, + "num_tokens": 163962802.0, + "step": 6439 + }, + { + "epoch": 0.707226004831979, + "grad_norm": 1.9526689052581787, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.707750678062439, + "num_tokens": 163993511.0, + "step": 6440 + }, + { + "epoch": 0.7073358225345926, + "grad_norm": 2.2665507793426514, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7359150648117065, + "num_tokens": 164015145.0, + "step": 6441 + }, + { + "epoch": 0.7074456402372062, + "grad_norm": 2.0922763347625732, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7122769951820374, + "num_tokens": 164041152.0, + "step": 6442 + }, + { + "epoch": 0.7075554579398199, + "grad_norm": 2.191479444503784, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.7013830542564392, + "num_tokens": 164067585.0, + "step": 6443 + }, + { + "epoch": 0.7076652756424335, + "grad_norm": 2.1506567001342773, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6969906091690063, + "num_tokens": 164094012.0, + "step": 6444 + }, + { + "epoch": 0.7077750933450472, + "grad_norm": 2.3720099925994873, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6894056797027588, + "num_tokens": 164116063.0, + "step": 6445 + }, + { + "epoch": 0.7078849110476609, + "grad_norm": 2.2602922916412354, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.698462724685669, + "num_tokens": 164140138.0, + "step": 6446 + }, + { + "epoch": 0.7079947287502746, + "grad_norm": 1.9285389184951782, + "learning_rate": 1e-06, + "loss": 1.0641, + "mean_token_accuracy": 0.6819809675216675, + "num_tokens": 164173976.0, + "step": 6447 + }, + { + "epoch": 0.7081045464528882, + "grad_norm": 1.9847121238708496, + "learning_rate": 1e-06, + "loss": 1.0522, + "mean_token_accuracy": 0.683903157711029, + "num_tokens": 164207080.0, + "step": 6448 + }, + { + "epoch": 0.7082143641555019, + "grad_norm": 2.3953452110290527, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7338742017745972, + "num_tokens": 164228161.0, + "step": 6449 + }, + { + "epoch": 0.7083241818581155, + "grad_norm": 2.138780117034912, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.7075440883636475, + "num_tokens": 164252483.0, + "step": 6450 + }, + { + "epoch": 0.7084339995607292, + "grad_norm": 2.6695570945739746, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7146825194358826, + "num_tokens": 164272946.0, + "step": 6451 + }, + { + "epoch": 0.7085438172633428, + "grad_norm": 2.082308769226074, + "learning_rate": 1e-06, + "loss": 1.0446, + "mean_token_accuracy": 0.697396993637085, + "num_tokens": 164302152.0, + "step": 6452 + }, + { + "epoch": 0.7086536349659565, + "grad_norm": 2.3725016117095947, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7040935754776001, + "num_tokens": 164326380.0, + "step": 6453 + }, + { + "epoch": 0.7087634526685702, + "grad_norm": 2.6451592445373535, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7091768980026245, + "num_tokens": 164346626.0, + "step": 6454 + }, + { + "epoch": 0.7088732703711839, + "grad_norm": 3.623690366744995, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.726234495639801, + "num_tokens": 164363898.0, + "step": 6455 + }, + { + "epoch": 0.7089830880737975, + "grad_norm": 2.2489829063415527, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7025477886199951, + "num_tokens": 164388438.0, + "step": 6456 + }, + { + "epoch": 0.7090929057764112, + "grad_norm": 2.218554973602295, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7221337556838989, + "num_tokens": 164413418.0, + "step": 6457 + }, + { + "epoch": 0.7092027234790248, + "grad_norm": 2.112703323364258, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7126669883728027, + "num_tokens": 164441275.0, + "step": 6458 + }, + { + "epoch": 0.7093125411816384, + "grad_norm": 2.1496317386627197, + "learning_rate": 1e-06, + "loss": 1.0031, + "mean_token_accuracy": 0.6984745860099792, + "num_tokens": 164468010.0, + "step": 6459 + }, + { + "epoch": 0.7094223588842521, + "grad_norm": 2.2630324363708496, + "learning_rate": 1e-06, + "loss": 1.0763, + "mean_token_accuracy": 0.6764398813247681, + "num_tokens": 164493155.0, + "step": 6460 + }, + { + "epoch": 0.7095321765868658, + "grad_norm": 2.126850128173828, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7086591720581055, + "num_tokens": 164517419.0, + "step": 6461 + }, + { + "epoch": 0.7096419942894795, + "grad_norm": 2.1355042457580566, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.6929610967636108, + "num_tokens": 164544384.0, + "step": 6462 + }, + { + "epoch": 0.7097518119920931, + "grad_norm": 2.441511392593384, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7007279396057129, + "num_tokens": 164566943.0, + "step": 6463 + }, + { + "epoch": 0.7098616296947068, + "grad_norm": 2.042937755584717, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6955317258834839, + "num_tokens": 164596914.0, + "step": 6464 + }, + { + "epoch": 0.7099714473973204, + "grad_norm": 2.1881213188171387, + "learning_rate": 1e-06, + "loss": 1.0721, + "mean_token_accuracy": 0.6674386262893677, + "num_tokens": 164621643.0, + "step": 6465 + }, + { + "epoch": 0.7100812650999341, + "grad_norm": 2.080246686935425, + "learning_rate": 1e-06, + "loss": 1.0442, + "mean_token_accuracy": 0.6903679370880127, + "num_tokens": 164649858.0, + "step": 6466 + }, + { + "epoch": 0.7101910828025477, + "grad_norm": 2.2167623043060303, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6948462128639221, + "num_tokens": 164676838.0, + "step": 6467 + }, + { + "epoch": 0.7103009005051615, + "grad_norm": 2.0577049255371094, + "learning_rate": 1e-06, + "loss": 1.0893, + "mean_token_accuracy": 0.6781131029129028, + "num_tokens": 164707680.0, + "step": 6468 + }, + { + "epoch": 0.7104107182077751, + "grad_norm": 2.0155563354492188, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6875468492507935, + "num_tokens": 164739497.0, + "step": 6469 + }, + { + "epoch": 0.7105205359103888, + "grad_norm": 1.8928066492080688, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7008624076843262, + "num_tokens": 164773236.0, + "step": 6470 + }, + { + "epoch": 0.7106303536130024, + "grad_norm": 2.2119333744049072, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7027952671051025, + "num_tokens": 164798492.0, + "step": 6471 + }, + { + "epoch": 0.7107401713156161, + "grad_norm": 2.3980517387390137, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7042344808578491, + "num_tokens": 164820201.0, + "step": 6472 + }, + { + "epoch": 0.7108499890182297, + "grad_norm": 1.9972984790802002, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7209800481796265, + "num_tokens": 164847216.0, + "step": 6473 + }, + { + "epoch": 0.7109598067208434, + "grad_norm": 2.0089447498321533, + "learning_rate": 1e-06, + "loss": 1.0555, + "mean_token_accuracy": 0.6871844530105591, + "num_tokens": 164878217.0, + "step": 6474 + }, + { + "epoch": 0.7110696244234571, + "grad_norm": 2.2289228439331055, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.709019660949707, + "num_tokens": 164901452.0, + "step": 6475 + }, + { + "epoch": 0.7111794421260708, + "grad_norm": 2.0406887531280518, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7130414247512817, + "num_tokens": 164928807.0, + "step": 6476 + }, + { + "epoch": 0.7112892598286844, + "grad_norm": 2.033674955368042, + "learning_rate": 1e-06, + "loss": 1.0872, + "mean_token_accuracy": 0.6823593378067017, + "num_tokens": 164958985.0, + "step": 6477 + }, + { + "epoch": 0.711399077531298, + "grad_norm": 2.32780385017395, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7000630497932434, + "num_tokens": 164983952.0, + "step": 6478 + }, + { + "epoch": 0.7115088952339117, + "grad_norm": 1.9596458673477173, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.6840816140174866, + "num_tokens": 165016090.0, + "step": 6479 + }, + { + "epoch": 0.7116187129365253, + "grad_norm": 2.2277727127075195, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7101502418518066, + "num_tokens": 165042362.0, + "step": 6480 + }, + { + "epoch": 0.711728530639139, + "grad_norm": 1.949535608291626, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.6953188180923462, + "num_tokens": 165073148.0, + "step": 6481 + }, + { + "epoch": 0.7118383483417526, + "grad_norm": 2.0758490562438965, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7003610134124756, + "num_tokens": 165100477.0, + "step": 6482 + }, + { + "epoch": 0.7119481660443664, + "grad_norm": 2.2708070278167725, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7036371827125549, + "num_tokens": 165123937.0, + "step": 6483 + }, + { + "epoch": 0.71205798374698, + "grad_norm": 2.2611770629882812, + "learning_rate": 1e-06, + "loss": 0.9723, + "mean_token_accuracy": 0.6954684853553772, + "num_tokens": 165146493.0, + "step": 6484 + }, + { + "epoch": 0.7121678014495937, + "grad_norm": 2.1377973556518555, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.6987258195877075, + "num_tokens": 165175001.0, + "step": 6485 + }, + { + "epoch": 0.7122776191522073, + "grad_norm": 2.086038827896118, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.701842188835144, + "num_tokens": 165200944.0, + "step": 6486 + }, + { + "epoch": 0.712387436854821, + "grad_norm": 2.373840093612671, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7157601118087769, + "num_tokens": 165223294.0, + "step": 6487 + }, + { + "epoch": 0.7124972545574346, + "grad_norm": 2.3826916217803955, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7162312269210815, + "num_tokens": 165245622.0, + "step": 6488 + }, + { + "epoch": 0.7126070722600483, + "grad_norm": 2.5755374431610107, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7262595891952515, + "num_tokens": 165264930.0, + "step": 6489 + }, + { + "epoch": 0.712716889962662, + "grad_norm": 2.203382968902588, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7077767848968506, + "num_tokens": 165288619.0, + "step": 6490 + }, + { + "epoch": 0.7128267076652757, + "grad_norm": 2.180032968521118, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7189682126045227, + "num_tokens": 165315502.0, + "step": 6491 + }, + { + "epoch": 0.7129365253678893, + "grad_norm": 2.445411205291748, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7192143201828003, + "num_tokens": 165335205.0, + "step": 6492 + }, + { + "epoch": 0.713046343070503, + "grad_norm": 2.0246810913085938, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.7030425071716309, + "num_tokens": 165367109.0, + "step": 6493 + }, + { + "epoch": 0.7131561607731166, + "grad_norm": 2.1491317749023438, + "learning_rate": 1e-06, + "loss": 1.0885, + "mean_token_accuracy": 0.6748522520065308, + "num_tokens": 165394384.0, + "step": 6494 + }, + { + "epoch": 0.7132659784757303, + "grad_norm": 2.0757412910461426, + "learning_rate": 1e-06, + "loss": 1.0402, + "mean_token_accuracy": 0.685171902179718, + "num_tokens": 165423275.0, + "step": 6495 + }, + { + "epoch": 0.7133757961783439, + "grad_norm": 2.0361437797546387, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6924301385879517, + "num_tokens": 165453621.0, + "step": 6496 + }, + { + "epoch": 0.7134856138809577, + "grad_norm": 2.1808810234069824, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.726393461227417, + "num_tokens": 165479218.0, + "step": 6497 + }, + { + "epoch": 0.7135954315835713, + "grad_norm": 2.3268380165100098, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.703947126865387, + "num_tokens": 165504126.0, + "step": 6498 + }, + { + "epoch": 0.713705249286185, + "grad_norm": 2.5396358966827393, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6993693113327026, + "num_tokens": 165528696.0, + "step": 6499 + }, + { + "epoch": 0.7138150669887986, + "grad_norm": 1.9054991006851196, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.707443118095398, + "num_tokens": 165559013.0, + "step": 6500 + }, + { + "epoch": 0.7139248846914122, + "grad_norm": 2.134751081466675, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6984245777130127, + "num_tokens": 165586231.0, + "step": 6501 + }, + { + "epoch": 0.7140347023940259, + "grad_norm": 2.149182081222534, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.696526288986206, + "num_tokens": 165612988.0, + "step": 6502 + }, + { + "epoch": 0.7141445200966395, + "grad_norm": 2.5113120079040527, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7088817358016968, + "num_tokens": 165637339.0, + "step": 6503 + }, + { + "epoch": 0.7142543377992533, + "grad_norm": 2.305894136428833, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7109217643737793, + "num_tokens": 165661135.0, + "step": 6504 + }, + { + "epoch": 0.7143641555018669, + "grad_norm": 1.8863660097122192, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.7091890573501587, + "num_tokens": 165692536.0, + "step": 6505 + }, + { + "epoch": 0.7144739732044806, + "grad_norm": 2.4449238777160645, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7369496822357178, + "num_tokens": 165712832.0, + "step": 6506 + }, + { + "epoch": 0.7145837909070942, + "grad_norm": 2.1357274055480957, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7057017087936401, + "num_tokens": 165738983.0, + "step": 6507 + }, + { + "epoch": 0.7146936086097079, + "grad_norm": 2.04038143157959, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.695067822933197, + "num_tokens": 165767519.0, + "step": 6508 + }, + { + "epoch": 0.7148034263123215, + "grad_norm": 2.096497058868408, + "learning_rate": 1e-06, + "loss": 0.9804, + "mean_token_accuracy": 0.7019402980804443, + "num_tokens": 165794862.0, + "step": 6509 + }, + { + "epoch": 0.7149132440149352, + "grad_norm": 1.8748935461044312, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7054851651191711, + "num_tokens": 165827882.0, + "step": 6510 + }, + { + "epoch": 0.7150230617175488, + "grad_norm": 2.3566622734069824, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7033461332321167, + "num_tokens": 165851969.0, + "step": 6511 + }, + { + "epoch": 0.7151328794201626, + "grad_norm": 2.214850425720215, + "learning_rate": 1e-06, + "loss": 1.0716, + "mean_token_accuracy": 0.6802762746810913, + "num_tokens": 165881248.0, + "step": 6512 + }, + { + "epoch": 0.7152426971227762, + "grad_norm": 2.2690982818603516, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.6903477907180786, + "num_tokens": 165907712.0, + "step": 6513 + }, + { + "epoch": 0.7153525148253899, + "grad_norm": 2.332564353942871, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6806867718696594, + "num_tokens": 165932973.0, + "step": 6514 + }, + { + "epoch": 0.7154623325280035, + "grad_norm": 2.1748435497283936, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.691733181476593, + "num_tokens": 165959332.0, + "step": 6515 + }, + { + "epoch": 0.7155721502306172, + "grad_norm": 2.263040065765381, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7182158827781677, + "num_tokens": 165981962.0, + "step": 6516 + }, + { + "epoch": 0.7156819679332308, + "grad_norm": 2.0992143154144287, + "learning_rate": 1e-06, + "loss": 1.1211, + "mean_token_accuracy": 0.6602144837379456, + "num_tokens": 166011165.0, + "step": 6517 + }, + { + "epoch": 0.7157917856358444, + "grad_norm": 2.2437150478363037, + "learning_rate": 1e-06, + "loss": 0.974, + "mean_token_accuracy": 0.7083346843719482, + "num_tokens": 166034315.0, + "step": 6518 + }, + { + "epoch": 0.7159016033384582, + "grad_norm": 2.3585455417633057, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.7164456248283386, + "num_tokens": 166056843.0, + "step": 6519 + }, + { + "epoch": 0.7160114210410718, + "grad_norm": 2.3762223720550537, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7198298573493958, + "num_tokens": 166079527.0, + "step": 6520 + }, + { + "epoch": 0.7161212387436855, + "grad_norm": 1.9445987939834595, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7234136462211609, + "num_tokens": 166110544.0, + "step": 6521 + }, + { + "epoch": 0.7162310564462991, + "grad_norm": 2.3230628967285156, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7151775360107422, + "num_tokens": 166133313.0, + "step": 6522 + }, + { + "epoch": 0.7163408741489128, + "grad_norm": 2.0121781826019287, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7174195051193237, + "num_tokens": 166161037.0, + "step": 6523 + }, + { + "epoch": 0.7164506918515264, + "grad_norm": 2.2280805110931396, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7101331353187561, + "num_tokens": 166186475.0, + "step": 6524 + }, + { + "epoch": 0.7165605095541401, + "grad_norm": 2.426666498184204, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7142795324325562, + "num_tokens": 166206903.0, + "step": 6525 + }, + { + "epoch": 0.7166703272567538, + "grad_norm": 2.551462173461914, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7223806977272034, + "num_tokens": 166226026.0, + "step": 6526 + }, + { + "epoch": 0.7167801449593675, + "grad_norm": 2.199429512023926, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.6999251842498779, + "num_tokens": 166250304.0, + "step": 6527 + }, + { + "epoch": 0.7168899626619811, + "grad_norm": 2.2953062057495117, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7079278826713562, + "num_tokens": 166272940.0, + "step": 6528 + }, + { + "epoch": 0.7169997803645948, + "grad_norm": 1.9317927360534668, + "learning_rate": 1e-06, + "loss": 1.0497, + "mean_token_accuracy": 0.682257890701294, + "num_tokens": 166305617.0, + "step": 6529 + }, + { + "epoch": 0.7171095980672084, + "grad_norm": 2.0481226444244385, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6917456388473511, + "num_tokens": 166333631.0, + "step": 6530 + }, + { + "epoch": 0.7172194157698221, + "grad_norm": 2.2119803428649902, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7121938467025757, + "num_tokens": 166357296.0, + "step": 6531 + }, + { + "epoch": 0.7173292334724357, + "grad_norm": 2.2505061626434326, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.6920294165611267, + "num_tokens": 166382530.0, + "step": 6532 + }, + { + "epoch": 0.7174390511750495, + "grad_norm": 2.4065279960632324, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7142058610916138, + "num_tokens": 166405697.0, + "step": 6533 + }, + { + "epoch": 0.7175488688776631, + "grad_norm": 2.269078493118286, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7053734064102173, + "num_tokens": 166430087.0, + "step": 6534 + }, + { + "epoch": 0.7176586865802768, + "grad_norm": 2.2316784858703613, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7394768595695496, + "num_tokens": 166452723.0, + "step": 6535 + }, + { + "epoch": 0.7177685042828904, + "grad_norm": 2.3002235889434814, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7117323279380798, + "num_tokens": 166476862.0, + "step": 6536 + }, + { + "epoch": 0.717878321985504, + "grad_norm": 2.3047378063201904, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7074736952781677, + "num_tokens": 166500790.0, + "step": 6537 + }, + { + "epoch": 0.7179881396881177, + "grad_norm": 2.5620527267456055, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7169437408447266, + "num_tokens": 166518924.0, + "step": 6538 + }, + { + "epoch": 0.7180979573907313, + "grad_norm": 2.2955708503723145, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.7011944055557251, + "num_tokens": 166542406.0, + "step": 6539 + }, + { + "epoch": 0.7182077750933451, + "grad_norm": 2.4883103370666504, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7264605164527893, + "num_tokens": 166560935.0, + "step": 6540 + }, + { + "epoch": 0.7183175927959587, + "grad_norm": 2.241851806640625, + "learning_rate": 1e-06, + "loss": 1.0286, + "mean_token_accuracy": 0.6889516115188599, + "num_tokens": 166586434.0, + "step": 6541 + }, + { + "epoch": 0.7184274104985724, + "grad_norm": 2.228288412094116, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.7131696939468384, + "num_tokens": 166610547.0, + "step": 6542 + }, + { + "epoch": 0.718537228201186, + "grad_norm": 2.125399589538574, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7347149848937988, + "num_tokens": 166634647.0, + "step": 6543 + }, + { + "epoch": 0.7186470459037997, + "grad_norm": 2.492502212524414, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7129529714584351, + "num_tokens": 166656283.0, + "step": 6544 + }, + { + "epoch": 0.7187568636064133, + "grad_norm": 2.387463331222534, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7311957478523254, + "num_tokens": 166676844.0, + "step": 6545 + }, + { + "epoch": 0.718866681309027, + "grad_norm": 2.1346518993377686, + "learning_rate": 1e-06, + "loss": 1.0435, + "mean_token_accuracy": 0.6822429299354553, + "num_tokens": 166702796.0, + "step": 6546 + }, + { + "epoch": 0.7189764990116406, + "grad_norm": 2.578504800796509, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7221193313598633, + "num_tokens": 166723436.0, + "step": 6547 + }, + { + "epoch": 0.7190863167142544, + "grad_norm": 2.1687979698181152, + "learning_rate": 1e-06, + "loss": 1.0182, + "mean_token_accuracy": 0.689232349395752, + "num_tokens": 166750992.0, + "step": 6548 + }, + { + "epoch": 0.719196134416868, + "grad_norm": 2.3733389377593994, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.718679666519165, + "num_tokens": 166773521.0, + "step": 6549 + }, + { + "epoch": 0.7193059521194817, + "grad_norm": 2.1765894889831543, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.6993809938430786, + "num_tokens": 166802011.0, + "step": 6550 + }, + { + "epoch": 0.7194157698220953, + "grad_norm": 2.1509480476379395, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7147831320762634, + "num_tokens": 166828089.0, + "step": 6551 + }, + { + "epoch": 0.719525587524709, + "grad_norm": 2.246403932571411, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7043501138687134, + "num_tokens": 166852916.0, + "step": 6552 + }, + { + "epoch": 0.7196354052273226, + "grad_norm": 2.0147857666015625, + "learning_rate": 1e-06, + "loss": 0.9721, + "mean_token_accuracy": 0.6970102787017822, + "num_tokens": 166881202.0, + "step": 6553 + }, + { + "epoch": 0.7197452229299363, + "grad_norm": 2.213799238204956, + "learning_rate": 1e-06, + "loss": 1.0449, + "mean_token_accuracy": 0.6956304311752319, + "num_tokens": 166906847.0, + "step": 6554 + }, + { + "epoch": 0.71985504063255, + "grad_norm": 2.427055835723877, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7007423639297485, + "num_tokens": 166929343.0, + "step": 6555 + }, + { + "epoch": 0.7199648583351637, + "grad_norm": 2.1809377670288086, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7152374982833862, + "num_tokens": 166953820.0, + "step": 6556 + }, + { + "epoch": 0.7200746760377773, + "grad_norm": 2.079608917236328, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7168089747428894, + "num_tokens": 166980888.0, + "step": 6557 + }, + { + "epoch": 0.720184493740391, + "grad_norm": 2.408168077468872, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7102454900741577, + "num_tokens": 167003108.0, + "step": 6558 + }, + { + "epoch": 0.7202943114430046, + "grad_norm": 2.5335440635681152, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.719704806804657, + "num_tokens": 167021946.0, + "step": 6559 + }, + { + "epoch": 0.7204041291456182, + "grad_norm": 2.3203299045562744, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7164214253425598, + "num_tokens": 167045556.0, + "step": 6560 + }, + { + "epoch": 0.7205139468482319, + "grad_norm": 2.1733336448669434, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7165402173995972, + "num_tokens": 167069426.0, + "step": 6561 + }, + { + "epoch": 0.7206237645508456, + "grad_norm": 2.0820980072021484, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.719215452671051, + "num_tokens": 167094282.0, + "step": 6562 + }, + { + "epoch": 0.7207335822534593, + "grad_norm": 2.4092013835906982, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7019454836845398, + "num_tokens": 167115735.0, + "step": 6563 + }, + { + "epoch": 0.7208433999560729, + "grad_norm": 2.031663179397583, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7204858064651489, + "num_tokens": 167144616.0, + "step": 6564 + }, + { + "epoch": 0.7209532176586866, + "grad_norm": 2.103447914123535, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7267878651618958, + "num_tokens": 167172146.0, + "step": 6565 + }, + { + "epoch": 0.7210630353613002, + "grad_norm": 2.2733917236328125, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.6975066661834717, + "num_tokens": 167197079.0, + "step": 6566 + }, + { + "epoch": 0.7211728530639139, + "grad_norm": 2.0203726291656494, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.708010196685791, + "num_tokens": 167226187.0, + "step": 6567 + }, + { + "epoch": 0.7212826707665275, + "grad_norm": 2.1561524868011475, + "learning_rate": 1e-06, + "loss": 1.0302, + "mean_token_accuracy": 0.6914840936660767, + "num_tokens": 167255292.0, + "step": 6568 + }, + { + "epoch": 0.7213924884691413, + "grad_norm": 2.4351344108581543, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7227131128311157, + "num_tokens": 167275945.0, + "step": 6569 + }, + { + "epoch": 0.7215023061717549, + "grad_norm": 2.1187732219696045, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6938982009887695, + "num_tokens": 167304888.0, + "step": 6570 + }, + { + "epoch": 0.7216121238743686, + "grad_norm": 2.105318069458008, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7157564759254456, + "num_tokens": 167331939.0, + "step": 6571 + }, + { + "epoch": 0.7217219415769822, + "grad_norm": 2.1509783267974854, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7080740332603455, + "num_tokens": 167356942.0, + "step": 6572 + }, + { + "epoch": 0.7218317592795959, + "grad_norm": 2.316406726837158, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7118819952011108, + "num_tokens": 167379671.0, + "step": 6573 + }, + { + "epoch": 0.7219415769822095, + "grad_norm": 1.9654076099395752, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7094712853431702, + "num_tokens": 167410415.0, + "step": 6574 + }, + { + "epoch": 0.7220513946848232, + "grad_norm": 2.2039828300476074, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6973475217819214, + "num_tokens": 167435269.0, + "step": 6575 + }, + { + "epoch": 0.7221612123874368, + "grad_norm": 2.318833112716675, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7090567350387573, + "num_tokens": 167459960.0, + "step": 6576 + }, + { + "epoch": 0.7222710300900506, + "grad_norm": 2.1031692028045654, + "learning_rate": 1e-06, + "loss": 1.0171, + "mean_token_accuracy": 0.700136661529541, + "num_tokens": 167488059.0, + "step": 6577 + }, + { + "epoch": 0.7223808477926642, + "grad_norm": 2.324254274368286, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7122596502304077, + "num_tokens": 167511741.0, + "step": 6578 + }, + { + "epoch": 0.7224906654952779, + "grad_norm": 2.2861886024475098, + "learning_rate": 1e-06, + "loss": 1.0042, + "mean_token_accuracy": 0.6969981789588928, + "num_tokens": 167537081.0, + "step": 6579 + }, + { + "epoch": 0.7226004831978915, + "grad_norm": 2.1018307209014893, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7152663469314575, + "num_tokens": 167564908.0, + "step": 6580 + }, + { + "epoch": 0.7227103009005051, + "grad_norm": 2.2032129764556885, + "learning_rate": 1e-06, + "loss": 1.0529, + "mean_token_accuracy": 0.6783268451690674, + "num_tokens": 167591870.0, + "step": 6581 + }, + { + "epoch": 0.7228201186031188, + "grad_norm": 2.347447395324707, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7055041790008545, + "num_tokens": 167614520.0, + "step": 6582 + }, + { + "epoch": 0.7229299363057324, + "grad_norm": 2.4480974674224854, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7046806812286377, + "num_tokens": 167634119.0, + "step": 6583 + }, + { + "epoch": 0.7230397540083462, + "grad_norm": 2.178286075592041, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.7020145654678345, + "num_tokens": 167660510.0, + "step": 6584 + }, + { + "epoch": 0.7231495717109598, + "grad_norm": 2.0057778358459473, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6883675456047058, + "num_tokens": 167690174.0, + "step": 6585 + }, + { + "epoch": 0.7232593894135735, + "grad_norm": 2.397965669631958, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6952225565910339, + "num_tokens": 167712186.0, + "step": 6586 + }, + { + "epoch": 0.7233692071161871, + "grad_norm": 2.005476713180542, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.6956130862236023, + "num_tokens": 167741299.0, + "step": 6587 + }, + { + "epoch": 0.7234790248188008, + "grad_norm": 2.261573076248169, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7203711271286011, + "num_tokens": 167765180.0, + "step": 6588 + }, + { + "epoch": 0.7235888425214144, + "grad_norm": 2.1870598793029785, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7094824314117432, + "num_tokens": 167790944.0, + "step": 6589 + }, + { + "epoch": 0.7236986602240281, + "grad_norm": 2.2458503246307373, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7119952440261841, + "num_tokens": 167813487.0, + "step": 6590 + }, + { + "epoch": 0.7238084779266418, + "grad_norm": 2.3512449264526367, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7024354934692383, + "num_tokens": 167836917.0, + "step": 6591 + }, + { + "epoch": 0.7239182956292555, + "grad_norm": 2.226954460144043, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7017302513122559, + "num_tokens": 167861561.0, + "step": 6592 + }, + { + "epoch": 0.7240281133318691, + "grad_norm": 2.3406951427459717, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7246624231338501, + "num_tokens": 167885306.0, + "step": 6593 + }, + { + "epoch": 0.7241379310344828, + "grad_norm": 2.047893762588501, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7060191631317139, + "num_tokens": 167913253.0, + "step": 6594 + }, + { + "epoch": 0.7242477487370964, + "grad_norm": 2.647644519805908, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7177010774612427, + "num_tokens": 167931391.0, + "step": 6595 + }, + { + "epoch": 0.72435756643971, + "grad_norm": 2.279412031173706, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7018527984619141, + "num_tokens": 167954055.0, + "step": 6596 + }, + { + "epoch": 0.7244673841423237, + "grad_norm": 2.235764980316162, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7135825753211975, + "num_tokens": 167979432.0, + "step": 6597 + }, + { + "epoch": 0.7245772018449375, + "grad_norm": 2.0573525428771973, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7090215682983398, + "num_tokens": 168007348.0, + "step": 6598 + }, + { + "epoch": 0.7246870195475511, + "grad_norm": 2.0332210063934326, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6929154396057129, + "num_tokens": 168036869.0, + "step": 6599 + }, + { + "epoch": 0.7247968372501647, + "grad_norm": 2.2839486598968506, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7203299403190613, + "num_tokens": 168060981.0, + "step": 6600 + }, + { + "epoch": 0.7249066549527784, + "grad_norm": 2.3231098651885986, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.7016324996948242, + "num_tokens": 168086408.0, + "step": 6601 + }, + { + "epoch": 0.725016472655392, + "grad_norm": 2.228466749191284, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7078808546066284, + "num_tokens": 168110189.0, + "step": 6602 + }, + { + "epoch": 0.7251262903580057, + "grad_norm": 2.1518607139587402, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6873732805252075, + "num_tokens": 168136644.0, + "step": 6603 + }, + { + "epoch": 0.7252361080606193, + "grad_norm": 2.1694986820220947, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7353823184967041, + "num_tokens": 168159274.0, + "step": 6604 + }, + { + "epoch": 0.725345925763233, + "grad_norm": 2.3859667778015137, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7256300449371338, + "num_tokens": 168179959.0, + "step": 6605 + }, + { + "epoch": 0.7254557434658467, + "grad_norm": 2.0635414123535156, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7090662121772766, + "num_tokens": 168209314.0, + "step": 6606 + }, + { + "epoch": 0.7255655611684604, + "grad_norm": 2.3911750316619873, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7231463193893433, + "num_tokens": 168231516.0, + "step": 6607 + }, + { + "epoch": 0.725675378871074, + "grad_norm": 2.1535158157348633, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7190672159194946, + "num_tokens": 168257668.0, + "step": 6608 + }, + { + "epoch": 0.7257851965736877, + "grad_norm": 2.032702684402466, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6945444345474243, + "num_tokens": 168287990.0, + "step": 6609 + }, + { + "epoch": 0.7258950142763013, + "grad_norm": 2.475131034851074, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7271441221237183, + "num_tokens": 168305827.0, + "step": 6610 + }, + { + "epoch": 0.726004831978915, + "grad_norm": 2.152745485305786, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7259341478347778, + "num_tokens": 168329732.0, + "step": 6611 + }, + { + "epoch": 0.7261146496815286, + "grad_norm": 2.2900800704956055, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7054147124290466, + "num_tokens": 168353301.0, + "step": 6612 + }, + { + "epoch": 0.7262244673841424, + "grad_norm": 2.2157022953033447, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6945319175720215, + "num_tokens": 168379566.0, + "step": 6613 + }, + { + "epoch": 0.726334285086756, + "grad_norm": 2.1622426509857178, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6790868043899536, + "num_tokens": 168405636.0, + "step": 6614 + }, + { + "epoch": 0.7264441027893697, + "grad_norm": 2.6260523796081543, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7059942483901978, + "num_tokens": 168426013.0, + "step": 6615 + }, + { + "epoch": 0.7265539204919833, + "grad_norm": 2.0007541179656982, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7218430042266846, + "num_tokens": 168456347.0, + "step": 6616 + }, + { + "epoch": 0.726663738194597, + "grad_norm": 2.44096302986145, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7199196815490723, + "num_tokens": 168477543.0, + "step": 6617 + }, + { + "epoch": 0.7267735558972106, + "grad_norm": 2.3015952110290527, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6888619661331177, + "num_tokens": 168500770.0, + "step": 6618 + }, + { + "epoch": 0.7268833735998242, + "grad_norm": 2.317901611328125, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6955632567405701, + "num_tokens": 168524585.0, + "step": 6619 + }, + { + "epoch": 0.726993191302438, + "grad_norm": 2.0248732566833496, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7188878059387207, + "num_tokens": 168552956.0, + "step": 6620 + }, + { + "epoch": 0.7271030090050516, + "grad_norm": 1.9954310655593872, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7179214954376221, + "num_tokens": 168584313.0, + "step": 6621 + }, + { + "epoch": 0.7272128267076653, + "grad_norm": 1.968253254890442, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.7006515264511108, + "num_tokens": 168614756.0, + "step": 6622 + }, + { + "epoch": 0.7273226444102789, + "grad_norm": 2.2782645225524902, + "learning_rate": 1e-06, + "loss": 1.0213, + "mean_token_accuracy": 0.6940352916717529, + "num_tokens": 168641194.0, + "step": 6623 + }, + { + "epoch": 0.7274324621128926, + "grad_norm": 2.2028613090515137, + "learning_rate": 1e-06, + "loss": 1.076, + "mean_token_accuracy": 0.6751030683517456, + "num_tokens": 168666995.0, + "step": 6624 + }, + { + "epoch": 0.7275422798155062, + "grad_norm": 2.3813915252685547, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6949214935302734, + "num_tokens": 168688911.0, + "step": 6625 + }, + { + "epoch": 0.7276520975181199, + "grad_norm": 2.3607089519500732, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7290288209915161, + "num_tokens": 168709226.0, + "step": 6626 + }, + { + "epoch": 0.7277619152207336, + "grad_norm": 1.8565646409988403, + "learning_rate": 1e-06, + "loss": 1.0218, + "mean_token_accuracy": 0.6966311931610107, + "num_tokens": 168742523.0, + "step": 6627 + }, + { + "epoch": 0.7278717329233473, + "grad_norm": 2.4096193313598633, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7229745388031006, + "num_tokens": 168764692.0, + "step": 6628 + }, + { + "epoch": 0.7279815506259609, + "grad_norm": 2.080517530441284, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.6984487771987915, + "num_tokens": 168793300.0, + "step": 6629 + }, + { + "epoch": 0.7280913683285746, + "grad_norm": 2.4024176597595215, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.707122802734375, + "num_tokens": 168814462.0, + "step": 6630 + }, + { + "epoch": 0.7282011860311882, + "grad_norm": 1.8932335376739502, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7108179330825806, + "num_tokens": 168846127.0, + "step": 6631 + }, + { + "epoch": 0.7283110037338019, + "grad_norm": 2.1543314456939697, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.6938043236732483, + "num_tokens": 168869450.0, + "step": 6632 + }, + { + "epoch": 0.7284208214364155, + "grad_norm": 2.3154993057250977, + "learning_rate": 1e-06, + "loss": 0.8132, + "mean_token_accuracy": 0.744056224822998, + "num_tokens": 168890194.0, + "step": 6633 + }, + { + "epoch": 0.7285306391390292, + "grad_norm": 2.4099130630493164, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7163505554199219, + "num_tokens": 168912725.0, + "step": 6634 + }, + { + "epoch": 0.7286404568416429, + "grad_norm": 2.2881853580474854, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7160522937774658, + "num_tokens": 168937021.0, + "step": 6635 + }, + { + "epoch": 0.7287502745442566, + "grad_norm": 1.943945050239563, + "learning_rate": 1e-06, + "loss": 1.0297, + "mean_token_accuracy": 0.6945950388908386, + "num_tokens": 168968600.0, + "step": 6636 + }, + { + "epoch": 0.7288600922468702, + "grad_norm": 2.279299259185791, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7070446610450745, + "num_tokens": 168991784.0, + "step": 6637 + }, + { + "epoch": 0.7289699099494839, + "grad_norm": 1.9699513912200928, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7151444554328918, + "num_tokens": 169020619.0, + "step": 6638 + }, + { + "epoch": 0.7290797276520975, + "grad_norm": 2.189159393310547, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.72292160987854, + "num_tokens": 169044315.0, + "step": 6639 + }, + { + "epoch": 0.7291895453547111, + "grad_norm": 2.1824262142181396, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.7013274431228638, + "num_tokens": 169069487.0, + "step": 6640 + }, + { + "epoch": 0.7292993630573248, + "grad_norm": 2.470365047454834, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7113876938819885, + "num_tokens": 169092924.0, + "step": 6641 + }, + { + "epoch": 0.7294091807599385, + "grad_norm": 2.1521899700164795, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6909571886062622, + "num_tokens": 169118941.0, + "step": 6642 + }, + { + "epoch": 0.7295189984625522, + "grad_norm": 2.004441738128662, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6976464986801147, + "num_tokens": 169148557.0, + "step": 6643 + }, + { + "epoch": 0.7296288161651658, + "grad_norm": 2.0770351886749268, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7016029357910156, + "num_tokens": 169175688.0, + "step": 6644 + }, + { + "epoch": 0.7297386338677795, + "grad_norm": 2.130082368850708, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7036776542663574, + "num_tokens": 169204274.0, + "step": 6645 + }, + { + "epoch": 0.7298484515703931, + "grad_norm": 1.8752806186676025, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.7064819931983948, + "num_tokens": 169236400.0, + "step": 6646 + }, + { + "epoch": 0.7299582692730068, + "grad_norm": 2.172775983810425, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7027653455734253, + "num_tokens": 169259990.0, + "step": 6647 + }, + { + "epoch": 0.7300680869756204, + "grad_norm": 2.4469223022460938, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7151585817337036, + "num_tokens": 169281068.0, + "step": 6648 + }, + { + "epoch": 0.7301779046782342, + "grad_norm": 2.4825825691223145, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7415236234664917, + "num_tokens": 169299576.0, + "step": 6649 + }, + { + "epoch": 0.7302877223808478, + "grad_norm": 2.151000738143921, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.727379560470581, + "num_tokens": 169323176.0, + "step": 6650 + }, + { + "epoch": 0.7303975400834615, + "grad_norm": 2.014763832092285, + "learning_rate": 1e-06, + "loss": 1.074, + "mean_token_accuracy": 0.6794357299804688, + "num_tokens": 169355773.0, + "step": 6651 + }, + { + "epoch": 0.7305073577860751, + "grad_norm": 2.1851425170898438, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7198807597160339, + "num_tokens": 169378864.0, + "step": 6652 + }, + { + "epoch": 0.7306171754886888, + "grad_norm": 2.32395339012146, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7120671272277832, + "num_tokens": 169400311.0, + "step": 6653 + }, + { + "epoch": 0.7307269931913024, + "grad_norm": 2.1532633304595947, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7267135381698608, + "num_tokens": 169424704.0, + "step": 6654 + }, + { + "epoch": 0.7308368108939161, + "grad_norm": 2.3203673362731934, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7015472054481506, + "num_tokens": 169448212.0, + "step": 6655 + }, + { + "epoch": 0.7309466285965298, + "grad_norm": 2.2578327655792236, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7094222903251648, + "num_tokens": 169471305.0, + "step": 6656 + }, + { + "epoch": 0.7310564462991435, + "grad_norm": 2.3076629638671875, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6925566792488098, + "num_tokens": 169496376.0, + "step": 6657 + }, + { + "epoch": 0.7311662640017571, + "grad_norm": 2.2706117630004883, + "learning_rate": 1e-06, + "loss": 1.0584, + "mean_token_accuracy": 0.6884025931358337, + "num_tokens": 169521974.0, + "step": 6658 + }, + { + "epoch": 0.7312760817043708, + "grad_norm": 2.055689811706543, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7033373713493347, + "num_tokens": 169551672.0, + "step": 6659 + }, + { + "epoch": 0.7313858994069844, + "grad_norm": 2.117823839187622, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6978681683540344, + "num_tokens": 169579474.0, + "step": 6660 + }, + { + "epoch": 0.731495717109598, + "grad_norm": 2.0693917274475098, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6984105110168457, + "num_tokens": 169608953.0, + "step": 6661 + }, + { + "epoch": 0.7316055348122117, + "grad_norm": 2.43196702003479, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7210040092468262, + "num_tokens": 169630302.0, + "step": 6662 + }, + { + "epoch": 0.7317153525148253, + "grad_norm": 2.0569632053375244, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6944900751113892, + "num_tokens": 169658182.0, + "step": 6663 + }, + { + "epoch": 0.7318251702174391, + "grad_norm": 2.206246852874756, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7277944087982178, + "num_tokens": 169682284.0, + "step": 6664 + }, + { + "epoch": 0.7319349879200527, + "grad_norm": 2.184490442276001, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.699307382106781, + "num_tokens": 169710535.0, + "step": 6665 + }, + { + "epoch": 0.7320448056226664, + "grad_norm": 2.5880837440490723, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7381918430328369, + "num_tokens": 169729019.0, + "step": 6666 + }, + { + "epoch": 0.73215462332528, + "grad_norm": 2.7217044830322266, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7294387817382812, + "num_tokens": 169745320.0, + "step": 6667 + }, + { + "epoch": 0.7322644410278937, + "grad_norm": 1.9349266290664673, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.6992950439453125, + "num_tokens": 169774913.0, + "step": 6668 + }, + { + "epoch": 0.7323742587305073, + "grad_norm": 2.1846330165863037, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6884951591491699, + "num_tokens": 169802419.0, + "step": 6669 + }, + { + "epoch": 0.732484076433121, + "grad_norm": 2.2407608032226562, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7169056534767151, + "num_tokens": 169825637.0, + "step": 6670 + }, + { + "epoch": 0.7325938941357347, + "grad_norm": 2.11020565032959, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7043518424034119, + "num_tokens": 169852105.0, + "step": 6671 + }, + { + "epoch": 0.7327037118383484, + "grad_norm": 2.2666993141174316, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7147705554962158, + "num_tokens": 169875747.0, + "step": 6672 + }, + { + "epoch": 0.732813529540962, + "grad_norm": 2.0055887699127197, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.727980375289917, + "num_tokens": 169904563.0, + "step": 6673 + }, + { + "epoch": 0.7329233472435757, + "grad_norm": 1.9251512289047241, + "learning_rate": 1e-06, + "loss": 1.0468, + "mean_token_accuracy": 0.6862087249755859, + "num_tokens": 169938298.0, + "step": 6674 + }, + { + "epoch": 0.7330331649461893, + "grad_norm": 2.493555784225464, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.7127811908721924, + "num_tokens": 169959959.0, + "step": 6675 + }, + { + "epoch": 0.733142982648803, + "grad_norm": 2.365861654281616, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7163972854614258, + "num_tokens": 169982158.0, + "step": 6676 + }, + { + "epoch": 0.7332528003514166, + "grad_norm": 2.100153923034668, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7139217257499695, + "num_tokens": 170008410.0, + "step": 6677 + }, + { + "epoch": 0.7333626180540304, + "grad_norm": 2.099250316619873, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7252190113067627, + "num_tokens": 170034324.0, + "step": 6678 + }, + { + "epoch": 0.733472435756644, + "grad_norm": 2.4490692615509033, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7120171785354614, + "num_tokens": 170054062.0, + "step": 6679 + }, + { + "epoch": 0.7335822534592576, + "grad_norm": 2.124598503112793, + "learning_rate": 1e-06, + "loss": 1.0583, + "mean_token_accuracy": 0.6817472577095032, + "num_tokens": 170081608.0, + "step": 6680 + }, + { + "epoch": 0.7336920711618713, + "grad_norm": 2.083742380142212, + "learning_rate": 1e-06, + "loss": 1.0537, + "mean_token_accuracy": 0.6820695400238037, + "num_tokens": 170110626.0, + "step": 6681 + }, + { + "epoch": 0.7338018888644849, + "grad_norm": 1.9028884172439575, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7123943567276001, + "num_tokens": 170141912.0, + "step": 6682 + }, + { + "epoch": 0.7339117065670986, + "grad_norm": 1.8079816102981567, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6925089359283447, + "num_tokens": 170180178.0, + "step": 6683 + }, + { + "epoch": 0.7340215242697122, + "grad_norm": 2.1729562282562256, + "learning_rate": 1e-06, + "loss": 1.0505, + "mean_token_accuracy": 0.6929594874382019, + "num_tokens": 170206970.0, + "step": 6684 + }, + { + "epoch": 0.734131341972326, + "grad_norm": 2.2271509170532227, + "learning_rate": 1e-06, + "loss": 1.0482, + "mean_token_accuracy": 0.6966581344604492, + "num_tokens": 170232323.0, + "step": 6685 + }, + { + "epoch": 0.7342411596749396, + "grad_norm": 2.0209779739379883, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7233612537384033, + "num_tokens": 170261549.0, + "step": 6686 + }, + { + "epoch": 0.7343509773775533, + "grad_norm": 2.0988476276397705, + "learning_rate": 1e-06, + "loss": 1.0746, + "mean_token_accuracy": 0.6702660322189331, + "num_tokens": 170290935.0, + "step": 6687 + }, + { + "epoch": 0.7344607950801669, + "grad_norm": 2.2980523109436035, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7148487567901611, + "num_tokens": 170316858.0, + "step": 6688 + }, + { + "epoch": 0.7345706127827806, + "grad_norm": 2.0063929557800293, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6956342458724976, + "num_tokens": 170345776.0, + "step": 6689 + }, + { + "epoch": 0.7346804304853942, + "grad_norm": 2.166949987411499, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.708254337310791, + "num_tokens": 170372475.0, + "step": 6690 + }, + { + "epoch": 0.7347902481880079, + "grad_norm": 2.3200461864471436, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.72077876329422, + "num_tokens": 170394587.0, + "step": 6691 + }, + { + "epoch": 0.7349000658906216, + "grad_norm": 2.1006298065185547, + "learning_rate": 1e-06, + "loss": 1.055, + "mean_token_accuracy": 0.6883944272994995, + "num_tokens": 170422649.0, + "step": 6692 + }, + { + "epoch": 0.7350098835932353, + "grad_norm": 2.036153554916382, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.7043836116790771, + "num_tokens": 170451177.0, + "step": 6693 + }, + { + "epoch": 0.7351197012958489, + "grad_norm": 2.055105686187744, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7203462719917297, + "num_tokens": 170479793.0, + "step": 6694 + }, + { + "epoch": 0.7352295189984626, + "grad_norm": 2.443748712539673, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7105496525764465, + "num_tokens": 170499582.0, + "step": 6695 + }, + { + "epoch": 0.7353393367010762, + "grad_norm": 2.304030418395996, + "learning_rate": 1e-06, + "loss": 1.0312, + "mean_token_accuracy": 0.6868830919265747, + "num_tokens": 170525366.0, + "step": 6696 + }, + { + "epoch": 0.7354491544036899, + "grad_norm": 2.252851963043213, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.7010911703109741, + "num_tokens": 170549472.0, + "step": 6697 + }, + { + "epoch": 0.7355589721063035, + "grad_norm": 2.2112319469451904, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.6962459087371826, + "num_tokens": 170576968.0, + "step": 6698 + }, + { + "epoch": 0.7356687898089171, + "grad_norm": 2.2583324909210205, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7144253849983215, + "num_tokens": 170601988.0, + "step": 6699 + }, + { + "epoch": 0.7357786075115309, + "grad_norm": 2.368572473526001, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7225779294967651, + "num_tokens": 170622482.0, + "step": 6700 + }, + { + "epoch": 0.7358884252141445, + "grad_norm": 2.29935359954834, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7087355852127075, + "num_tokens": 170644943.0, + "step": 6701 + }, + { + "epoch": 0.7359982429167582, + "grad_norm": 1.9085173606872559, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7171072959899902, + "num_tokens": 170676608.0, + "step": 6702 + }, + { + "epoch": 0.7361080606193718, + "grad_norm": 1.8602741956710815, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7125866413116455, + "num_tokens": 170708155.0, + "step": 6703 + }, + { + "epoch": 0.7362178783219855, + "grad_norm": 2.47757625579834, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.731813371181488, + "num_tokens": 170728713.0, + "step": 6704 + }, + { + "epoch": 0.7363276960245991, + "grad_norm": 2.361950159072876, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7121679782867432, + "num_tokens": 170751203.0, + "step": 6705 + }, + { + "epoch": 0.7364375137272128, + "grad_norm": 2.6116127967834473, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7072744965553284, + "num_tokens": 170770719.0, + "step": 6706 + }, + { + "epoch": 0.7365473314298265, + "grad_norm": 2.2271065711975098, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7243900299072266, + "num_tokens": 170793543.0, + "step": 6707 + }, + { + "epoch": 0.7366571491324402, + "grad_norm": 2.346735715866089, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7180769443511963, + "num_tokens": 170814686.0, + "step": 6708 + }, + { + "epoch": 0.7367669668350538, + "grad_norm": 2.112741231918335, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6916029453277588, + "num_tokens": 170843865.0, + "step": 6709 + }, + { + "epoch": 0.7368767845376675, + "grad_norm": 2.4838502407073975, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6929739117622375, + "num_tokens": 170865294.0, + "step": 6710 + }, + { + "epoch": 0.7369866022402811, + "grad_norm": 2.449186325073242, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.719865083694458, + "num_tokens": 170885433.0, + "step": 6711 + }, + { + "epoch": 0.7370964199428948, + "grad_norm": 2.039198875427246, + "learning_rate": 1e-06, + "loss": 1.0392, + "mean_token_accuracy": 0.6900978088378906, + "num_tokens": 170915134.0, + "step": 6712 + }, + { + "epoch": 0.7372062376455084, + "grad_norm": 2.131120204925537, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.7023111581802368, + "num_tokens": 170940634.0, + "step": 6713 + }, + { + "epoch": 0.7373160553481222, + "grad_norm": 2.1011080741882324, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7176073789596558, + "num_tokens": 170966574.0, + "step": 6714 + }, + { + "epoch": 0.7374258730507358, + "grad_norm": 1.9697518348693848, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.710098147392273, + "num_tokens": 170997334.0, + "step": 6715 + }, + { + "epoch": 0.7375356907533495, + "grad_norm": 2.082866907119751, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6981297135353088, + "num_tokens": 171024453.0, + "step": 6716 + }, + { + "epoch": 0.7376455084559631, + "grad_norm": 2.5278759002685547, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.6885061860084534, + "num_tokens": 171046283.0, + "step": 6717 + }, + { + "epoch": 0.7377553261585768, + "grad_norm": 2.296278476715088, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7112421989440918, + "num_tokens": 171069495.0, + "step": 6718 + }, + { + "epoch": 0.7378651438611904, + "grad_norm": 2.349080801010132, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7092108130455017, + "num_tokens": 171092749.0, + "step": 6719 + }, + { + "epoch": 0.737974961563804, + "grad_norm": 2.334556818008423, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7245376706123352, + "num_tokens": 171115387.0, + "step": 6720 + }, + { + "epoch": 0.7380847792664178, + "grad_norm": 2.2747809886932373, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7107149958610535, + "num_tokens": 171138595.0, + "step": 6721 + }, + { + "epoch": 0.7381945969690314, + "grad_norm": 2.2392561435699463, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7007743716239929, + "num_tokens": 171161917.0, + "step": 6722 + }, + { + "epoch": 0.7383044146716451, + "grad_norm": 2.2902908325195312, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7109180688858032, + "num_tokens": 171186182.0, + "step": 6723 + }, + { + "epoch": 0.7384142323742587, + "grad_norm": 2.3193986415863037, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7129527926445007, + "num_tokens": 171209522.0, + "step": 6724 + }, + { + "epoch": 0.7385240500768724, + "grad_norm": 2.383037805557251, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.7373015880584717, + "num_tokens": 171227713.0, + "step": 6725 + }, + { + "epoch": 0.738633867779486, + "grad_norm": 2.2070207595825195, + "learning_rate": 1e-06, + "loss": 0.9816, + "mean_token_accuracy": 0.6981903314590454, + "num_tokens": 171253456.0, + "step": 6726 + }, + { + "epoch": 0.7387436854820997, + "grad_norm": 2.273012399673462, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7228893041610718, + "num_tokens": 171276136.0, + "step": 6727 + }, + { + "epoch": 0.7388535031847133, + "grad_norm": 1.877320647239685, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7079111337661743, + "num_tokens": 171308242.0, + "step": 6728 + }, + { + "epoch": 0.7389633208873271, + "grad_norm": 2.5773448944091797, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7213143110275269, + "num_tokens": 171327612.0, + "step": 6729 + }, + { + "epoch": 0.7390731385899407, + "grad_norm": 2.1291348934173584, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7033799886703491, + "num_tokens": 171354359.0, + "step": 6730 + }, + { + "epoch": 0.7391829562925544, + "grad_norm": 2.125704526901245, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7128887176513672, + "num_tokens": 171379943.0, + "step": 6731 + }, + { + "epoch": 0.739292773995168, + "grad_norm": 2.490173101425171, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.7042449712753296, + "num_tokens": 171399760.0, + "step": 6732 + }, + { + "epoch": 0.7394025916977817, + "grad_norm": 2.334479570388794, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7225366234779358, + "num_tokens": 171422007.0, + "step": 6733 + }, + { + "epoch": 0.7395124094003953, + "grad_norm": 2.3629581928253174, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6930670738220215, + "num_tokens": 171446071.0, + "step": 6734 + }, + { + "epoch": 0.739622227103009, + "grad_norm": 2.183772325515747, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7280631065368652, + "num_tokens": 171471726.0, + "step": 6735 + }, + { + "epoch": 0.7397320448056227, + "grad_norm": 2.272258758544922, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6930146217346191, + "num_tokens": 171500156.0, + "step": 6736 + }, + { + "epoch": 0.7398418625082364, + "grad_norm": 2.7259209156036377, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7059233784675598, + "num_tokens": 171517704.0, + "step": 6737 + }, + { + "epoch": 0.73995168021085, + "grad_norm": 2.381844997406006, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7045904397964478, + "num_tokens": 171539551.0, + "step": 6738 + }, + { + "epoch": 0.7400614979134637, + "grad_norm": 2.2173311710357666, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7124075889587402, + "num_tokens": 171565565.0, + "step": 6739 + }, + { + "epoch": 0.7401713156160773, + "grad_norm": 2.1984260082244873, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7208588719367981, + "num_tokens": 171591350.0, + "step": 6740 + }, + { + "epoch": 0.7402811333186909, + "grad_norm": 2.3376853466033936, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7366925477981567, + "num_tokens": 171613147.0, + "step": 6741 + }, + { + "epoch": 0.7403909510213046, + "grad_norm": 2.483583688735962, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7501628398895264, + "num_tokens": 171632816.0, + "step": 6742 + }, + { + "epoch": 0.7405007687239183, + "grad_norm": 2.004836320877075, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.711472749710083, + "num_tokens": 171662418.0, + "step": 6743 + }, + { + "epoch": 0.740610586426532, + "grad_norm": 1.998295545578003, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6973512172698975, + "num_tokens": 171693391.0, + "step": 6744 + }, + { + "epoch": 0.7407204041291456, + "grad_norm": 2.459446907043457, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7226503491401672, + "num_tokens": 171711746.0, + "step": 6745 + }, + { + "epoch": 0.7408302218317593, + "grad_norm": 2.280093193054199, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7130622863769531, + "num_tokens": 171736779.0, + "step": 6746 + }, + { + "epoch": 0.7409400395343729, + "grad_norm": 2.2557952404022217, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7088063955307007, + "num_tokens": 171762275.0, + "step": 6747 + }, + { + "epoch": 0.7410498572369866, + "grad_norm": 2.01275634765625, + "learning_rate": 1e-06, + "loss": 1.0464, + "mean_token_accuracy": 0.6835472583770752, + "num_tokens": 171793482.0, + "step": 6748 + }, + { + "epoch": 0.7411596749396002, + "grad_norm": 2.0460145473480225, + "learning_rate": 1e-06, + "loss": 1.0171, + "mean_token_accuracy": 0.6908292770385742, + "num_tokens": 171822795.0, + "step": 6749 + }, + { + "epoch": 0.741269492642214, + "grad_norm": 2.3665499687194824, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.7058902382850647, + "num_tokens": 171845288.0, + "step": 6750 + }, + { + "epoch": 0.7413793103448276, + "grad_norm": 2.3218235969543457, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6931775212287903, + "num_tokens": 171868715.0, + "step": 6751 + }, + { + "epoch": 0.7414891280474413, + "grad_norm": 2.3667423725128174, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7359597086906433, + "num_tokens": 171889819.0, + "step": 6752 + }, + { + "epoch": 0.7415989457500549, + "grad_norm": 2.399820327758789, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.707382321357727, + "num_tokens": 171911391.0, + "step": 6753 + }, + { + "epoch": 0.7417087634526686, + "grad_norm": 2.5082437992095947, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6869603395462036, + "num_tokens": 171934545.0, + "step": 6754 + }, + { + "epoch": 0.7418185811552822, + "grad_norm": 2.3274662494659424, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.7026743292808533, + "num_tokens": 171960262.0, + "step": 6755 + }, + { + "epoch": 0.7419283988578959, + "grad_norm": 2.098219156265259, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.71403568983078, + "num_tokens": 171986514.0, + "step": 6756 + }, + { + "epoch": 0.7420382165605095, + "grad_norm": 2.1269021034240723, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6971238851547241, + "num_tokens": 172014288.0, + "step": 6757 + }, + { + "epoch": 0.7421480342631233, + "grad_norm": 2.2151243686676025, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.6887861490249634, + "num_tokens": 172040416.0, + "step": 6758 + }, + { + "epoch": 0.7422578519657369, + "grad_norm": 2.157278060913086, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.6887722015380859, + "num_tokens": 172066288.0, + "step": 6759 + }, + { + "epoch": 0.7423676696683505, + "grad_norm": 2.0910871028900146, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7054437398910522, + "num_tokens": 172093435.0, + "step": 6760 + }, + { + "epoch": 0.7424774873709642, + "grad_norm": 2.478778600692749, + "learning_rate": 1e-06, + "loss": 1.0061, + "mean_token_accuracy": 0.7025396227836609, + "num_tokens": 172114709.0, + "step": 6761 + }, + { + "epoch": 0.7425873050735778, + "grad_norm": 1.988857626914978, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6984677314758301, + "num_tokens": 172143408.0, + "step": 6762 + }, + { + "epoch": 0.7426971227761915, + "grad_norm": 2.0666255950927734, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.7026472687721252, + "num_tokens": 172169954.0, + "step": 6763 + }, + { + "epoch": 0.7428069404788051, + "grad_norm": 2.2775843143463135, + "learning_rate": 1e-06, + "loss": 1.0353, + "mean_token_accuracy": 0.6818434000015259, + "num_tokens": 172195110.0, + "step": 6764 + }, + { + "epoch": 0.7429167581814189, + "grad_norm": 2.136650323867798, + "learning_rate": 1e-06, + "loss": 1.0731, + "mean_token_accuracy": 0.6786496639251709, + "num_tokens": 172223410.0, + "step": 6765 + }, + { + "epoch": 0.7430265758840325, + "grad_norm": 2.1501851081848145, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7052552103996277, + "num_tokens": 172249550.0, + "step": 6766 + }, + { + "epoch": 0.7431363935866462, + "grad_norm": 2.0622754096984863, + "learning_rate": 1e-06, + "loss": 1.0127, + "mean_token_accuracy": 0.692085862159729, + "num_tokens": 172277449.0, + "step": 6767 + }, + { + "epoch": 0.7432462112892598, + "grad_norm": 2.086949110031128, + "learning_rate": 1e-06, + "loss": 0.9931, + "mean_token_accuracy": 0.6958984136581421, + "num_tokens": 172305087.0, + "step": 6768 + }, + { + "epoch": 0.7433560289918735, + "grad_norm": 2.326660394668579, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7210420370101929, + "num_tokens": 172327353.0, + "step": 6769 + }, + { + "epoch": 0.7434658466944871, + "grad_norm": 2.2721667289733887, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6940072178840637, + "num_tokens": 172351861.0, + "step": 6770 + }, + { + "epoch": 0.7435756643971008, + "grad_norm": 2.2433061599731445, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7249446511268616, + "num_tokens": 172375218.0, + "step": 6771 + }, + { + "epoch": 0.7436854820997145, + "grad_norm": 2.1240851879119873, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7286840081214905, + "num_tokens": 172401262.0, + "step": 6772 + }, + { + "epoch": 0.7437952998023282, + "grad_norm": 2.2062580585479736, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7254989743232727, + "num_tokens": 172424705.0, + "step": 6773 + }, + { + "epoch": 0.7439051175049418, + "grad_norm": 1.979431390762329, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.7107342481613159, + "num_tokens": 172453974.0, + "step": 6774 + }, + { + "epoch": 0.7440149352075555, + "grad_norm": 2.1725568771362305, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7103729248046875, + "num_tokens": 172478704.0, + "step": 6775 + }, + { + "epoch": 0.7441247529101691, + "grad_norm": 2.1902780532836914, + "learning_rate": 1e-06, + "loss": 1.0205, + "mean_token_accuracy": 0.7018221020698547, + "num_tokens": 172504030.0, + "step": 6776 + }, + { + "epoch": 0.7442345706127828, + "grad_norm": 2.2903964519500732, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.707504153251648, + "num_tokens": 172526583.0, + "step": 6777 + }, + { + "epoch": 0.7443443883153964, + "grad_norm": 2.5263662338256836, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7083921432495117, + "num_tokens": 172546773.0, + "step": 6778 + }, + { + "epoch": 0.7444542060180102, + "grad_norm": 2.3874728679656982, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7122342586517334, + "num_tokens": 172567332.0, + "step": 6779 + }, + { + "epoch": 0.7445640237206238, + "grad_norm": 2.2499427795410156, + "learning_rate": 1e-06, + "loss": 0.8544, + "mean_token_accuracy": 0.7337605953216553, + "num_tokens": 172591765.0, + "step": 6780 + }, + { + "epoch": 0.7446738414232374, + "grad_norm": 2.189337968826294, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.6958598494529724, + "num_tokens": 172617945.0, + "step": 6781 + }, + { + "epoch": 0.7447836591258511, + "grad_norm": 2.3168869018554688, + "learning_rate": 1e-06, + "loss": 1.032, + "mean_token_accuracy": 0.6894805431365967, + "num_tokens": 172640239.0, + "step": 6782 + }, + { + "epoch": 0.7448934768284647, + "grad_norm": 2.587350368499756, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7305078506469727, + "num_tokens": 172658749.0, + "step": 6783 + }, + { + "epoch": 0.7450032945310784, + "grad_norm": 2.2174861431121826, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6935542821884155, + "num_tokens": 172684118.0, + "step": 6784 + }, + { + "epoch": 0.745113112233692, + "grad_norm": 2.0803298950195312, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6969514489173889, + "num_tokens": 172711042.0, + "step": 6785 + }, + { + "epoch": 0.7452229299363057, + "grad_norm": 2.147434949874878, + "learning_rate": 1e-06, + "loss": 1.0348, + "mean_token_accuracy": 0.6902065873146057, + "num_tokens": 172738165.0, + "step": 6786 + }, + { + "epoch": 0.7453327476389194, + "grad_norm": 2.5769083499908447, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.6991072297096252, + "num_tokens": 172757572.0, + "step": 6787 + }, + { + "epoch": 0.7454425653415331, + "grad_norm": 1.9301738739013672, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6939661502838135, + "num_tokens": 172789086.0, + "step": 6788 + }, + { + "epoch": 0.7455523830441467, + "grad_norm": 2.2584471702575684, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7030973434448242, + "num_tokens": 172813166.0, + "step": 6789 + }, + { + "epoch": 0.7456622007467604, + "grad_norm": 2.537778377532959, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7215004563331604, + "num_tokens": 172832196.0, + "step": 6790 + }, + { + "epoch": 0.745772018449374, + "grad_norm": 2.5498533248901367, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7269279360771179, + "num_tokens": 172851268.0, + "step": 6791 + }, + { + "epoch": 0.7458818361519877, + "grad_norm": 2.0669240951538086, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7080449461936951, + "num_tokens": 172878597.0, + "step": 6792 + }, + { + "epoch": 0.7459916538546013, + "grad_norm": 1.882282018661499, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6960362792015076, + "num_tokens": 172913175.0, + "step": 6793 + }, + { + "epoch": 0.7461014715572151, + "grad_norm": 2.1773343086242676, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6913312673568726, + "num_tokens": 172940602.0, + "step": 6794 + }, + { + "epoch": 0.7462112892598287, + "grad_norm": 2.4227395057678223, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6935601234436035, + "num_tokens": 172964606.0, + "step": 6795 + }, + { + "epoch": 0.7463211069624424, + "grad_norm": 2.4356069564819336, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.717926561832428, + "num_tokens": 172987279.0, + "step": 6796 + }, + { + "epoch": 0.746430924665056, + "grad_norm": 2.0466983318328857, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.749126672744751, + "num_tokens": 173011951.0, + "step": 6797 + }, + { + "epoch": 0.7465407423676697, + "grad_norm": 2.243021249771118, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7324420213699341, + "num_tokens": 173035296.0, + "step": 6798 + }, + { + "epoch": 0.7466505600702833, + "grad_norm": 2.088806390762329, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.718890905380249, + "num_tokens": 173062752.0, + "step": 6799 + }, + { + "epoch": 0.7467603777728969, + "grad_norm": 2.2373125553131104, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7329433560371399, + "num_tokens": 173086381.0, + "step": 6800 + }, + { + "epoch": 0.7468701954755107, + "grad_norm": 2.351914882659912, + "learning_rate": 1e-06, + "loss": 1.0115, + "mean_token_accuracy": 0.6984071731567383, + "num_tokens": 173108585.0, + "step": 6801 + }, + { + "epoch": 0.7469800131781243, + "grad_norm": 2.417511463165283, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.704300045967102, + "num_tokens": 173132093.0, + "step": 6802 + }, + { + "epoch": 0.747089830880738, + "grad_norm": 2.156076192855835, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7273358702659607, + "num_tokens": 173156235.0, + "step": 6803 + }, + { + "epoch": 0.7471996485833516, + "grad_norm": 2.547424554824829, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7178727388381958, + "num_tokens": 173176720.0, + "step": 6804 + }, + { + "epoch": 0.7473094662859653, + "grad_norm": 2.27947735786438, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6938468217849731, + "num_tokens": 173200355.0, + "step": 6805 + }, + { + "epoch": 0.7474192839885789, + "grad_norm": 2.4006993770599365, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.710951566696167, + "num_tokens": 173222733.0, + "step": 6806 + }, + { + "epoch": 0.7475291016911926, + "grad_norm": 2.1059770584106445, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.699425458908081, + "num_tokens": 173251055.0, + "step": 6807 + }, + { + "epoch": 0.7476389193938063, + "grad_norm": 2.225921869277954, + "learning_rate": 1e-06, + "loss": 1.0464, + "mean_token_accuracy": 0.6795931458473206, + "num_tokens": 173277201.0, + "step": 6808 + }, + { + "epoch": 0.74774873709642, + "grad_norm": 2.221176862716675, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.728122353553772, + "num_tokens": 173301996.0, + "step": 6809 + }, + { + "epoch": 0.7478585547990336, + "grad_norm": 1.9984699487686157, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7110767364501953, + "num_tokens": 173331518.0, + "step": 6810 + }, + { + "epoch": 0.7479683725016473, + "grad_norm": 2.519198417663574, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7261261940002441, + "num_tokens": 173350239.0, + "step": 6811 + }, + { + "epoch": 0.7480781902042609, + "grad_norm": 2.280090093612671, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7235639691352844, + "num_tokens": 173372332.0, + "step": 6812 + }, + { + "epoch": 0.7481880079068746, + "grad_norm": 2.2262096405029297, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7068343758583069, + "num_tokens": 173399503.0, + "step": 6813 + }, + { + "epoch": 0.7482978256094882, + "grad_norm": 2.097031354904175, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7170583009719849, + "num_tokens": 173425764.0, + "step": 6814 + }, + { + "epoch": 0.7484076433121019, + "grad_norm": 2.0330610275268555, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7100149989128113, + "num_tokens": 173453685.0, + "step": 6815 + }, + { + "epoch": 0.7485174610147156, + "grad_norm": 2.422607421875, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7023311853408813, + "num_tokens": 173475588.0, + "step": 6816 + }, + { + "epoch": 0.7486272787173293, + "grad_norm": 2.409728765487671, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.6935659050941467, + "num_tokens": 173497353.0, + "step": 6817 + }, + { + "epoch": 0.7487370964199429, + "grad_norm": 1.9813306331634521, + "learning_rate": 1e-06, + "loss": 1.0387, + "mean_token_accuracy": 0.6937153339385986, + "num_tokens": 173525649.0, + "step": 6818 + }, + { + "epoch": 0.7488469141225566, + "grad_norm": 2.2825911045074463, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.720833957195282, + "num_tokens": 173547942.0, + "step": 6819 + }, + { + "epoch": 0.7489567318251702, + "grad_norm": 2.3788490295410156, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7074074745178223, + "num_tokens": 173569975.0, + "step": 6820 + }, + { + "epoch": 0.7490665495277838, + "grad_norm": 2.5422186851501465, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7012275457382202, + "num_tokens": 173590763.0, + "step": 6821 + }, + { + "epoch": 0.7491763672303975, + "grad_norm": 2.162708282470703, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6908137798309326, + "num_tokens": 173620714.0, + "step": 6822 + }, + { + "epoch": 0.7492861849330112, + "grad_norm": 2.36733078956604, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.694183886051178, + "num_tokens": 173643924.0, + "step": 6823 + }, + { + "epoch": 0.7493960026356249, + "grad_norm": 2.0711259841918945, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7155874967575073, + "num_tokens": 173671150.0, + "step": 6824 + }, + { + "epoch": 0.7495058203382385, + "grad_norm": 2.354156017303467, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7175473570823669, + "num_tokens": 173694758.0, + "step": 6825 + }, + { + "epoch": 0.7496156380408522, + "grad_norm": 2.1890273094177246, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7059459090232849, + "num_tokens": 173721774.0, + "step": 6826 + }, + { + "epoch": 0.7497254557434658, + "grad_norm": 2.164213180541992, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7093100547790527, + "num_tokens": 173748172.0, + "step": 6827 + }, + { + "epoch": 0.7498352734460795, + "grad_norm": 2.131819248199463, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.703089714050293, + "num_tokens": 173774611.0, + "step": 6828 + }, + { + "epoch": 0.7499450911486931, + "grad_norm": 2.0438435077667236, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7165429592132568, + "num_tokens": 173802206.0, + "step": 6829 + }, + { + "epoch": 0.7500549088513069, + "grad_norm": 2.039560317993164, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7266286611557007, + "num_tokens": 173829141.0, + "step": 6830 + }, + { + "epoch": 0.7501647265539205, + "grad_norm": 2.21858811378479, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.685776948928833, + "num_tokens": 173855194.0, + "step": 6831 + }, + { + "epoch": 0.7502745442565342, + "grad_norm": 1.9787383079528809, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7088506817817688, + "num_tokens": 173883098.0, + "step": 6832 + }, + { + "epoch": 0.7503843619591478, + "grad_norm": 2.3193435668945312, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7086489200592041, + "num_tokens": 173905990.0, + "step": 6833 + }, + { + "epoch": 0.7504941796617615, + "grad_norm": 2.2274703979492188, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7354113459587097, + "num_tokens": 173929595.0, + "step": 6834 + }, + { + "epoch": 0.7506039973643751, + "grad_norm": 2.3235604763031006, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7065100073814392, + "num_tokens": 173951672.0, + "step": 6835 + }, + { + "epoch": 0.7507138150669888, + "grad_norm": 2.1760618686676025, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6918975114822388, + "num_tokens": 173980217.0, + "step": 6836 + }, + { + "epoch": 0.7508236327696025, + "grad_norm": 2.332961320877075, + "learning_rate": 1e-06, + "loss": 1.0392, + "mean_token_accuracy": 0.684715747833252, + "num_tokens": 174004822.0, + "step": 6837 + }, + { + "epoch": 0.7509334504722162, + "grad_norm": 2.0996530055999756, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7189731597900391, + "num_tokens": 174032813.0, + "step": 6838 + }, + { + "epoch": 0.7510432681748298, + "grad_norm": 1.9554921388626099, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7102804183959961, + "num_tokens": 174065264.0, + "step": 6839 + }, + { + "epoch": 0.7511530858774434, + "grad_norm": 2.1341147422790527, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7116305828094482, + "num_tokens": 174089659.0, + "step": 6840 + }, + { + "epoch": 0.7512629035800571, + "grad_norm": 2.287468671798706, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7134727835655212, + "num_tokens": 174112045.0, + "step": 6841 + }, + { + "epoch": 0.7513727212826707, + "grad_norm": 2.2839581966400146, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.729069709777832, + "num_tokens": 174134793.0, + "step": 6842 + }, + { + "epoch": 0.7514825389852844, + "grad_norm": 1.849968671798706, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7060908675193787, + "num_tokens": 174168775.0, + "step": 6843 + }, + { + "epoch": 0.7515923566878981, + "grad_norm": 2.2829864025115967, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.7004898190498352, + "num_tokens": 174193253.0, + "step": 6844 + }, + { + "epoch": 0.7517021743905118, + "grad_norm": 2.1437888145446777, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.708579957485199, + "num_tokens": 174219728.0, + "step": 6845 + }, + { + "epoch": 0.7518119920931254, + "grad_norm": 1.7668315172195435, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7087849974632263, + "num_tokens": 174255640.0, + "step": 6846 + }, + { + "epoch": 0.7519218097957391, + "grad_norm": 2.052537679672241, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6957947015762329, + "num_tokens": 174284664.0, + "step": 6847 + }, + { + "epoch": 0.7520316274983527, + "grad_norm": 2.048896312713623, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7021615505218506, + "num_tokens": 174312432.0, + "step": 6848 + }, + { + "epoch": 0.7521414452009664, + "grad_norm": 1.906554102897644, + "learning_rate": 1e-06, + "loss": 1.0627, + "mean_token_accuracy": 0.6837210655212402, + "num_tokens": 174344440.0, + "step": 6849 + }, + { + "epoch": 0.75225126290358, + "grad_norm": 2.0181291103363037, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7025644183158875, + "num_tokens": 174373078.0, + "step": 6850 + }, + { + "epoch": 0.7523610806061937, + "grad_norm": 2.0288710594177246, + "learning_rate": 1e-06, + "loss": 1.035, + "mean_token_accuracy": 0.6859441995620728, + "num_tokens": 174402451.0, + "step": 6851 + }, + { + "epoch": 0.7524708983088074, + "grad_norm": 2.2047886848449707, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.6910276412963867, + "num_tokens": 174428609.0, + "step": 6852 + }, + { + "epoch": 0.7525807160114211, + "grad_norm": 2.3388726711273193, + "learning_rate": 1e-06, + "loss": 1.044, + "mean_token_accuracy": 0.7106925249099731, + "num_tokens": 174453478.0, + "step": 6853 + }, + { + "epoch": 0.7526905337140347, + "grad_norm": 2.125049591064453, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.7027832865715027, + "num_tokens": 174478977.0, + "step": 6854 + }, + { + "epoch": 0.7528003514166484, + "grad_norm": 2.069808006286621, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7231755256652832, + "num_tokens": 174505677.0, + "step": 6855 + }, + { + "epoch": 0.752910169119262, + "grad_norm": 2.134608268737793, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7055860161781311, + "num_tokens": 174534178.0, + "step": 6856 + }, + { + "epoch": 0.7530199868218757, + "grad_norm": 2.143280029296875, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6920535564422607, + "num_tokens": 174559667.0, + "step": 6857 + }, + { + "epoch": 0.7531298045244893, + "grad_norm": 2.1504268646240234, + "learning_rate": 1e-06, + "loss": 1.0521, + "mean_token_accuracy": 0.6779235601425171, + "num_tokens": 174586810.0, + "step": 6858 + }, + { + "epoch": 0.7532396222271031, + "grad_norm": 1.9317489862442017, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7062572836875916, + "num_tokens": 174616944.0, + "step": 6859 + }, + { + "epoch": 0.7533494399297167, + "grad_norm": 2.269108772277832, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7106155157089233, + "num_tokens": 174639818.0, + "step": 6860 + }, + { + "epoch": 0.7534592576323303, + "grad_norm": 2.3203558921813965, + "learning_rate": 1e-06, + "loss": 1.033, + "mean_token_accuracy": 0.693196177482605, + "num_tokens": 174665058.0, + "step": 6861 + }, + { + "epoch": 0.753569075334944, + "grad_norm": 2.0921790599823, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7104209661483765, + "num_tokens": 174692981.0, + "step": 6862 + }, + { + "epoch": 0.7536788930375576, + "grad_norm": 1.9503380060195923, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7228329181671143, + "num_tokens": 174724017.0, + "step": 6863 + }, + { + "epoch": 0.7537887107401713, + "grad_norm": 2.166921377182007, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.6965476274490356, + "num_tokens": 174749860.0, + "step": 6864 + }, + { + "epoch": 0.7538985284427849, + "grad_norm": 2.068751335144043, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.7093802094459534, + "num_tokens": 174776734.0, + "step": 6865 + }, + { + "epoch": 0.7540083461453987, + "grad_norm": 2.1249821186065674, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7198370099067688, + "num_tokens": 174804019.0, + "step": 6866 + }, + { + "epoch": 0.7541181638480123, + "grad_norm": 2.5586280822753906, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7002201080322266, + "num_tokens": 174823958.0, + "step": 6867 + }, + { + "epoch": 0.754227981550626, + "grad_norm": 2.151458501815796, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7266637682914734, + "num_tokens": 174848545.0, + "step": 6868 + }, + { + "epoch": 0.7543377992532396, + "grad_norm": 2.0384762287139893, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7037867307662964, + "num_tokens": 174878346.0, + "step": 6869 + }, + { + "epoch": 0.7544476169558533, + "grad_norm": 1.9626113176345825, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7060005068778992, + "num_tokens": 174907414.0, + "step": 6870 + }, + { + "epoch": 0.7545574346584669, + "grad_norm": 2.4281835556030273, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7129473090171814, + "num_tokens": 174927705.0, + "step": 6871 + }, + { + "epoch": 0.7546672523610806, + "grad_norm": 2.238426685333252, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7112789154052734, + "num_tokens": 174953677.0, + "step": 6872 + }, + { + "epoch": 0.7547770700636943, + "grad_norm": 2.1568636894226074, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7030504941940308, + "num_tokens": 174978665.0, + "step": 6873 + }, + { + "epoch": 0.754886887766308, + "grad_norm": 1.9878510236740112, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7145237326622009, + "num_tokens": 175008728.0, + "step": 6874 + }, + { + "epoch": 0.7549967054689216, + "grad_norm": 2.2658703327178955, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.7039461731910706, + "num_tokens": 175032284.0, + "step": 6875 + }, + { + "epoch": 0.7551065231715353, + "grad_norm": 2.3026621341705322, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.6975030899047852, + "num_tokens": 175057966.0, + "step": 6876 + }, + { + "epoch": 0.7552163408741489, + "grad_norm": 2.4279098510742188, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7024514675140381, + "num_tokens": 175078628.0, + "step": 6877 + }, + { + "epoch": 0.7553261585767626, + "grad_norm": 2.2798986434936523, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.6830657124519348, + "num_tokens": 175103137.0, + "step": 6878 + }, + { + "epoch": 0.7554359762793762, + "grad_norm": 1.9833065271377563, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7211476564407349, + "num_tokens": 175134135.0, + "step": 6879 + }, + { + "epoch": 0.7555457939819898, + "grad_norm": 2.7514212131500244, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7164678573608398, + "num_tokens": 175150642.0, + "step": 6880 + }, + { + "epoch": 0.7556556116846036, + "grad_norm": 2.1246137619018555, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7025623917579651, + "num_tokens": 175178547.0, + "step": 6881 + }, + { + "epoch": 0.7557654293872172, + "grad_norm": 2.017350673675537, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7159233093261719, + "num_tokens": 175205856.0, + "step": 6882 + }, + { + "epoch": 0.7558752470898309, + "grad_norm": 2.2548015117645264, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.7035028338432312, + "num_tokens": 175232147.0, + "step": 6883 + }, + { + "epoch": 0.7559850647924445, + "grad_norm": 2.087675094604492, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.70232093334198, + "num_tokens": 175258873.0, + "step": 6884 + }, + { + "epoch": 0.7560948824950582, + "grad_norm": 2.469390869140625, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7192391157150269, + "num_tokens": 175279177.0, + "step": 6885 + }, + { + "epoch": 0.7562047001976718, + "grad_norm": 1.9158865213394165, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.6964055299758911, + "num_tokens": 175311665.0, + "step": 6886 + }, + { + "epoch": 0.7563145179002855, + "grad_norm": 2.2117836475372314, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.719691276550293, + "num_tokens": 175334979.0, + "step": 6887 + }, + { + "epoch": 0.7564243356028992, + "grad_norm": 2.0728187561035156, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.6978338956832886, + "num_tokens": 175364776.0, + "step": 6888 + }, + { + "epoch": 0.7565341533055129, + "grad_norm": 2.044989824295044, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6922013759613037, + "num_tokens": 175392956.0, + "step": 6889 + }, + { + "epoch": 0.7566439710081265, + "grad_norm": 2.340014934539795, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7081351280212402, + "num_tokens": 175417388.0, + "step": 6890 + }, + { + "epoch": 0.7567537887107402, + "grad_norm": 1.9594467878341675, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7291797399520874, + "num_tokens": 175445935.0, + "step": 6891 + }, + { + "epoch": 0.7568636064133538, + "grad_norm": 2.447117567062378, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7068110108375549, + "num_tokens": 175465758.0, + "step": 6892 + }, + { + "epoch": 0.7569734241159675, + "grad_norm": 2.301426410675049, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6973887085914612, + "num_tokens": 175489964.0, + "step": 6893 + }, + { + "epoch": 0.7570832418185811, + "grad_norm": 2.483262300491333, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7203561663627625, + "num_tokens": 175508740.0, + "step": 6894 + }, + { + "epoch": 0.7571930595211949, + "grad_norm": 2.135258197784424, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7203351259231567, + "num_tokens": 175532207.0, + "step": 6895 + }, + { + "epoch": 0.7573028772238085, + "grad_norm": 2.3948726654052734, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7204474806785583, + "num_tokens": 175553406.0, + "step": 6896 + }, + { + "epoch": 0.7574126949264222, + "grad_norm": 2.1986753940582275, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.6993744373321533, + "num_tokens": 175577574.0, + "step": 6897 + }, + { + "epoch": 0.7575225126290358, + "grad_norm": 2.344367742538452, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.713941216468811, + "num_tokens": 175598488.0, + "step": 6898 + }, + { + "epoch": 0.7576323303316495, + "grad_norm": 1.9336154460906982, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.6863268613815308, + "num_tokens": 175630596.0, + "step": 6899 + }, + { + "epoch": 0.7577421480342631, + "grad_norm": 2.100790023803711, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6973969340324402, + "num_tokens": 175657982.0, + "step": 6900 + }, + { + "epoch": 0.7578519657368767, + "grad_norm": 1.934380292892456, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7089366316795349, + "num_tokens": 175690986.0, + "step": 6901 + }, + { + "epoch": 0.7579617834394905, + "grad_norm": 2.4761805534362793, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.6979644894599915, + "num_tokens": 175709922.0, + "step": 6902 + }, + { + "epoch": 0.7580716011421041, + "grad_norm": 2.072944164276123, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7104430198669434, + "num_tokens": 175735935.0, + "step": 6903 + }, + { + "epoch": 0.7581814188447178, + "grad_norm": 2.084688663482666, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7090131640434265, + "num_tokens": 175763053.0, + "step": 6904 + }, + { + "epoch": 0.7582912365473314, + "grad_norm": 2.0018832683563232, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7168147563934326, + "num_tokens": 175791630.0, + "step": 6905 + }, + { + "epoch": 0.7584010542499451, + "grad_norm": 2.097904920578003, + "learning_rate": 1e-06, + "loss": 1.0948, + "mean_token_accuracy": 0.6730712056159973, + "num_tokens": 175818400.0, + "step": 6906 + }, + { + "epoch": 0.7585108719525587, + "grad_norm": 2.3905186653137207, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7353830337524414, + "num_tokens": 175839636.0, + "step": 6907 + }, + { + "epoch": 0.7586206896551724, + "grad_norm": 2.122415542602539, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6927652359008789, + "num_tokens": 175867352.0, + "step": 6908 + }, + { + "epoch": 0.758730507357786, + "grad_norm": 2.184715509414673, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.71161949634552, + "num_tokens": 175891964.0, + "step": 6909 + }, + { + "epoch": 0.7588403250603998, + "grad_norm": 2.3761789798736572, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7249521017074585, + "num_tokens": 175912603.0, + "step": 6910 + }, + { + "epoch": 0.7589501427630134, + "grad_norm": 1.9691495895385742, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7394604086875916, + "num_tokens": 175939180.0, + "step": 6911 + }, + { + "epoch": 0.7590599604656271, + "grad_norm": 1.9752731323242188, + "learning_rate": 1e-06, + "loss": 1.1041, + "mean_token_accuracy": 0.6680595278739929, + "num_tokens": 175972744.0, + "step": 6912 + }, + { + "epoch": 0.7591697781682407, + "grad_norm": 2.5646915435791016, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7102431654930115, + "num_tokens": 175992819.0, + "step": 6913 + }, + { + "epoch": 0.7592795958708544, + "grad_norm": 2.085799217224121, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7083234786987305, + "num_tokens": 176019716.0, + "step": 6914 + }, + { + "epoch": 0.759389413573468, + "grad_norm": 2.180941581726074, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7070055603981018, + "num_tokens": 176045301.0, + "step": 6915 + }, + { + "epoch": 0.7594992312760817, + "grad_norm": 2.3301103115081787, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6966530680656433, + "num_tokens": 176068541.0, + "step": 6916 + }, + { + "epoch": 0.7596090489786954, + "grad_norm": 1.991587519645691, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7011513710021973, + "num_tokens": 176099538.0, + "step": 6917 + }, + { + "epoch": 0.7597188666813091, + "grad_norm": 2.185250997543335, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.6884670853614807, + "num_tokens": 176125673.0, + "step": 6918 + }, + { + "epoch": 0.7598286843839227, + "grad_norm": 2.208829164505005, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7301297783851624, + "num_tokens": 176150879.0, + "step": 6919 + }, + { + "epoch": 0.7599385020865363, + "grad_norm": 2.0561845302581787, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7024392485618591, + "num_tokens": 176178295.0, + "step": 6920 + }, + { + "epoch": 0.76004831978915, + "grad_norm": 1.9479169845581055, + "learning_rate": 1e-06, + "loss": 1.0986, + "mean_token_accuracy": 0.6716702580451965, + "num_tokens": 176211584.0, + "step": 6921 + }, + { + "epoch": 0.7601581374917636, + "grad_norm": 2.240978240966797, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.731706440448761, + "num_tokens": 176236337.0, + "step": 6922 + }, + { + "epoch": 0.7602679551943773, + "grad_norm": 2.250180244445801, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7363008260726929, + "num_tokens": 176259073.0, + "step": 6923 + }, + { + "epoch": 0.760377772896991, + "grad_norm": 2.189016103744507, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6939094662666321, + "num_tokens": 176286551.0, + "step": 6924 + }, + { + "epoch": 0.7604875905996047, + "grad_norm": 2.6624040603637695, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7179663181304932, + "num_tokens": 176305690.0, + "step": 6925 + }, + { + "epoch": 0.7605974083022183, + "grad_norm": 2.4209837913513184, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7277891635894775, + "num_tokens": 176326379.0, + "step": 6926 + }, + { + "epoch": 0.760707226004832, + "grad_norm": 2.1145191192626953, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7133843898773193, + "num_tokens": 176352295.0, + "step": 6927 + }, + { + "epoch": 0.7608170437074456, + "grad_norm": 1.895180583000183, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.6964213252067566, + "num_tokens": 176386248.0, + "step": 6928 + }, + { + "epoch": 0.7609268614100593, + "grad_norm": 2.626450300216675, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7200796604156494, + "num_tokens": 176403696.0, + "step": 6929 + }, + { + "epoch": 0.7610366791126729, + "grad_norm": 2.5629122257232666, + "learning_rate": 1e-06, + "loss": 1.0398, + "mean_token_accuracy": 0.6890593767166138, + "num_tokens": 176424831.0, + "step": 6930 + }, + { + "epoch": 0.7611464968152867, + "grad_norm": 2.305741548538208, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.709141731262207, + "num_tokens": 176449000.0, + "step": 6931 + }, + { + "epoch": 0.7612563145179003, + "grad_norm": 2.204444408416748, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6936041116714478, + "num_tokens": 176475306.0, + "step": 6932 + }, + { + "epoch": 0.761366132220514, + "grad_norm": 2.1187820434570312, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7100272178649902, + "num_tokens": 176500312.0, + "step": 6933 + }, + { + "epoch": 0.7614759499231276, + "grad_norm": 2.1665961742401123, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.722387969493866, + "num_tokens": 176524419.0, + "step": 6934 + }, + { + "epoch": 0.7615857676257413, + "grad_norm": 2.1805825233459473, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7075445652008057, + "num_tokens": 176549612.0, + "step": 6935 + }, + { + "epoch": 0.7616955853283549, + "grad_norm": 1.8014380931854248, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.6963028311729431, + "num_tokens": 176585425.0, + "step": 6936 + }, + { + "epoch": 0.7618054030309686, + "grad_norm": 2.473747968673706, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7141467332839966, + "num_tokens": 176605930.0, + "step": 6937 + }, + { + "epoch": 0.7619152207335822, + "grad_norm": 2.3045902252197266, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7326740622520447, + "num_tokens": 176627539.0, + "step": 6938 + }, + { + "epoch": 0.762025038436196, + "grad_norm": 2.0534114837646484, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7266867756843567, + "num_tokens": 176655126.0, + "step": 6939 + }, + { + "epoch": 0.7621348561388096, + "grad_norm": 2.220743179321289, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7063066363334656, + "num_tokens": 176681105.0, + "step": 6940 + }, + { + "epoch": 0.7622446738414232, + "grad_norm": 2.323803663253784, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7259342074394226, + "num_tokens": 176702875.0, + "step": 6941 + }, + { + "epoch": 0.7623544915440369, + "grad_norm": 2.0863733291625977, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7352450489997864, + "num_tokens": 176728870.0, + "step": 6942 + }, + { + "epoch": 0.7624643092466505, + "grad_norm": 2.1016340255737305, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7432072162628174, + "num_tokens": 176754431.0, + "step": 6943 + }, + { + "epoch": 0.7625741269492642, + "grad_norm": 2.2530808448791504, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.6937709450721741, + "num_tokens": 176778705.0, + "step": 6944 + }, + { + "epoch": 0.7626839446518778, + "grad_norm": 2.1027257442474365, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7125996351242065, + "num_tokens": 176805004.0, + "step": 6945 + }, + { + "epoch": 0.7627937623544916, + "grad_norm": 2.1026837825775146, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7114313840866089, + "num_tokens": 176833278.0, + "step": 6946 + }, + { + "epoch": 0.7629035800571052, + "grad_norm": 2.5823214054107666, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7122217416763306, + "num_tokens": 176851848.0, + "step": 6947 + }, + { + "epoch": 0.7630133977597189, + "grad_norm": 2.14224910736084, + "learning_rate": 1e-06, + "loss": 1.0939, + "mean_token_accuracy": 0.6762076616287231, + "num_tokens": 176880435.0, + "step": 6948 + }, + { + "epoch": 0.7631232154623325, + "grad_norm": 2.2171685695648193, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7246205806732178, + "num_tokens": 176903675.0, + "step": 6949 + }, + { + "epoch": 0.7632330331649462, + "grad_norm": 2.310133934020996, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7124327421188354, + "num_tokens": 176926319.0, + "step": 6950 + }, + { + "epoch": 0.7633428508675598, + "grad_norm": 2.0269813537597656, + "learning_rate": 1e-06, + "loss": 1.0356, + "mean_token_accuracy": 0.6906144022941589, + "num_tokens": 176957674.0, + "step": 6951 + }, + { + "epoch": 0.7634526685701735, + "grad_norm": 2.101950168609619, + "learning_rate": 1e-06, + "loss": 1.0132, + "mean_token_accuracy": 0.6934974193572998, + "num_tokens": 176985898.0, + "step": 6952 + }, + { + "epoch": 0.7635624862727872, + "grad_norm": 2.4071173667907715, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7270538210868835, + "num_tokens": 177007384.0, + "step": 6953 + }, + { + "epoch": 0.7636723039754009, + "grad_norm": 2.2238707542419434, + "learning_rate": 1e-06, + "loss": 1.0228, + "mean_token_accuracy": 0.6883914470672607, + "num_tokens": 177033098.0, + "step": 6954 + }, + { + "epoch": 0.7637821216780145, + "grad_norm": 2.0410566329956055, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7370807528495789, + "num_tokens": 177058415.0, + "step": 6955 + }, + { + "epoch": 0.7638919393806282, + "grad_norm": 2.0174076557159424, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.6980761289596558, + "num_tokens": 177086494.0, + "step": 6956 + }, + { + "epoch": 0.7640017570832418, + "grad_norm": 2.3842387199401855, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.6997778415679932, + "num_tokens": 177107394.0, + "step": 6957 + }, + { + "epoch": 0.7641115747858555, + "grad_norm": 2.069085121154785, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.6925948858261108, + "num_tokens": 177135159.0, + "step": 6958 + }, + { + "epoch": 0.7642213924884691, + "grad_norm": 2.1641948223114014, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7250951528549194, + "num_tokens": 177159153.0, + "step": 6959 + }, + { + "epoch": 0.7643312101910829, + "grad_norm": 2.391289234161377, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7165408134460449, + "num_tokens": 177179382.0, + "step": 6960 + }, + { + "epoch": 0.7644410278936965, + "grad_norm": 2.745086193084717, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7123069167137146, + "num_tokens": 177197267.0, + "step": 6961 + }, + { + "epoch": 0.7645508455963101, + "grad_norm": 2.160780668258667, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6917572021484375, + "num_tokens": 177223058.0, + "step": 6962 + }, + { + "epoch": 0.7646606632989238, + "grad_norm": 1.8616948127746582, + "learning_rate": 1e-06, + "loss": 1.0122, + "mean_token_accuracy": 0.700168788433075, + "num_tokens": 177256833.0, + "step": 6963 + }, + { + "epoch": 0.7647704810015374, + "grad_norm": 2.1732497215270996, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.698688268661499, + "num_tokens": 177282468.0, + "step": 6964 + }, + { + "epoch": 0.7648802987041511, + "grad_norm": 2.0987939834594727, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7224318981170654, + "num_tokens": 177309000.0, + "step": 6965 + }, + { + "epoch": 0.7649901164067647, + "grad_norm": 2.1301181316375732, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7061231136322021, + "num_tokens": 177338640.0, + "step": 6966 + }, + { + "epoch": 0.7650999341093784, + "grad_norm": 2.428703546524048, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7293753623962402, + "num_tokens": 177359943.0, + "step": 6967 + }, + { + "epoch": 0.7652097518119921, + "grad_norm": 2.2012476921081543, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6976816058158875, + "num_tokens": 177385911.0, + "step": 6968 + }, + { + "epoch": 0.7653195695146058, + "grad_norm": 2.084618330001831, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7179888486862183, + "num_tokens": 177412454.0, + "step": 6969 + }, + { + "epoch": 0.7654293872172194, + "grad_norm": 1.9366557598114014, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6884194016456604, + "num_tokens": 177442479.0, + "step": 6970 + }, + { + "epoch": 0.7655392049198331, + "grad_norm": 2.256540298461914, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.6982269287109375, + "num_tokens": 177466138.0, + "step": 6971 + }, + { + "epoch": 0.7656490226224467, + "grad_norm": 2.371302366256714, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.713578462600708, + "num_tokens": 177487529.0, + "step": 6972 + }, + { + "epoch": 0.7657588403250604, + "grad_norm": 2.3057167530059814, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7267556190490723, + "num_tokens": 177509696.0, + "step": 6973 + }, + { + "epoch": 0.765868658027674, + "grad_norm": 2.296236753463745, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.703143298625946, + "num_tokens": 177532566.0, + "step": 6974 + }, + { + "epoch": 0.7659784757302878, + "grad_norm": 2.0239360332489014, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7101258039474487, + "num_tokens": 177563640.0, + "step": 6975 + }, + { + "epoch": 0.7660882934329014, + "grad_norm": 2.232668876647949, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6949634552001953, + "num_tokens": 177588084.0, + "step": 6976 + }, + { + "epoch": 0.7661981111355151, + "grad_norm": 2.249997854232788, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.704421877861023, + "num_tokens": 177611587.0, + "step": 6977 + }, + { + "epoch": 0.7663079288381287, + "grad_norm": 2.453786611557007, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.724636435508728, + "num_tokens": 177630249.0, + "step": 6978 + }, + { + "epoch": 0.7664177465407424, + "grad_norm": 2.4437520503997803, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.6997683644294739, + "num_tokens": 177651136.0, + "step": 6979 + }, + { + "epoch": 0.766527564243356, + "grad_norm": 2.0459039211273193, + "learning_rate": 1e-06, + "loss": 1.038, + "mean_token_accuracy": 0.6863547563552856, + "num_tokens": 177680022.0, + "step": 6980 + }, + { + "epoch": 0.7666373819459696, + "grad_norm": 2.1579463481903076, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.7119370698928833, + "num_tokens": 177705523.0, + "step": 6981 + }, + { + "epoch": 0.7667471996485834, + "grad_norm": 2.1293752193450928, + "learning_rate": 1e-06, + "loss": 1.0697, + "mean_token_accuracy": 0.6743506193161011, + "num_tokens": 177733495.0, + "step": 6982 + }, + { + "epoch": 0.766857017351197, + "grad_norm": 2.692333221435547, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7075966596603394, + "num_tokens": 177751304.0, + "step": 6983 + }, + { + "epoch": 0.7669668350538107, + "grad_norm": 2.01603364944458, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7058541774749756, + "num_tokens": 177781395.0, + "step": 6984 + }, + { + "epoch": 0.7670766527564243, + "grad_norm": 2.1034317016601562, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7306771278381348, + "num_tokens": 177807528.0, + "step": 6985 + }, + { + "epoch": 0.767186470459038, + "grad_norm": 2.3861451148986816, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7145402431488037, + "num_tokens": 177828954.0, + "step": 6986 + }, + { + "epoch": 0.7672962881616516, + "grad_norm": 2.328134298324585, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7086302042007446, + "num_tokens": 177851367.0, + "step": 6987 + }, + { + "epoch": 0.7674061058642653, + "grad_norm": 1.9391461610794067, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6948720812797546, + "num_tokens": 177883789.0, + "step": 6988 + }, + { + "epoch": 0.767515923566879, + "grad_norm": 2.208881378173828, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7051692605018616, + "num_tokens": 177908034.0, + "step": 6989 + }, + { + "epoch": 0.7676257412694927, + "grad_norm": 2.0602481365203857, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.7021827101707458, + "num_tokens": 177939243.0, + "step": 6990 + }, + { + "epoch": 0.7677355589721063, + "grad_norm": 2.2290871143341064, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7014026641845703, + "num_tokens": 177963629.0, + "step": 6991 + }, + { + "epoch": 0.76784537667472, + "grad_norm": 1.9628101587295532, + "learning_rate": 1e-06, + "loss": 1.0551, + "mean_token_accuracy": 0.6828227043151855, + "num_tokens": 177997484.0, + "step": 6992 + }, + { + "epoch": 0.7679551943773336, + "grad_norm": 2.067840337753296, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.713930070400238, + "num_tokens": 178025473.0, + "step": 6993 + }, + { + "epoch": 0.7680650120799473, + "grad_norm": 2.2520036697387695, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7325365543365479, + "num_tokens": 178047955.0, + "step": 6994 + }, + { + "epoch": 0.7681748297825609, + "grad_norm": 2.11612868309021, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7035598158836365, + "num_tokens": 178074621.0, + "step": 6995 + }, + { + "epoch": 0.7682846474851747, + "grad_norm": 2.0729894638061523, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6910197734832764, + "num_tokens": 178104386.0, + "step": 6996 + }, + { + "epoch": 0.7683944651877883, + "grad_norm": 2.279677391052246, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.697110652923584, + "num_tokens": 178128448.0, + "step": 6997 + }, + { + "epoch": 0.768504282890402, + "grad_norm": 2.228257656097412, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.68821781873703, + "num_tokens": 178153596.0, + "step": 6998 + }, + { + "epoch": 0.7686141005930156, + "grad_norm": 2.4103546142578125, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.713401198387146, + "num_tokens": 178181702.0, + "step": 6999 + }, + { + "epoch": 0.7687239182956292, + "grad_norm": 2.547393321990967, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7279466986656189, + "num_tokens": 178203050.0, + "step": 7000 + }, + { + "epoch": 0.7688337359982429, + "grad_norm": 2.327025890350342, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7105162143707275, + "num_tokens": 178223890.0, + "step": 7001 + }, + { + "epoch": 0.7689435537008565, + "grad_norm": 2.1891422271728516, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.7024722099304199, + "num_tokens": 178249863.0, + "step": 7002 + }, + { + "epoch": 0.7690533714034702, + "grad_norm": 2.0546371936798096, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7065478563308716, + "num_tokens": 178276544.0, + "step": 7003 + }, + { + "epoch": 0.7691631891060839, + "grad_norm": 2.0351600646972656, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7266548275947571, + "num_tokens": 178305004.0, + "step": 7004 + }, + { + "epoch": 0.7692730068086976, + "grad_norm": 2.333514928817749, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7122777104377747, + "num_tokens": 178326280.0, + "step": 7005 + }, + { + "epoch": 0.7693828245113112, + "grad_norm": 2.4311065673828125, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7163963913917542, + "num_tokens": 178346540.0, + "step": 7006 + }, + { + "epoch": 0.7694926422139249, + "grad_norm": 2.0496432781219482, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7222365140914917, + "num_tokens": 178373420.0, + "step": 7007 + }, + { + "epoch": 0.7696024599165385, + "grad_norm": 2.254685401916504, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7012313604354858, + "num_tokens": 178398617.0, + "step": 7008 + }, + { + "epoch": 0.7697122776191522, + "grad_norm": 2.2479162216186523, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.72989821434021, + "num_tokens": 178422000.0, + "step": 7009 + }, + { + "epoch": 0.7698220953217658, + "grad_norm": 1.9461196660995483, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6949743032455444, + "num_tokens": 178452739.0, + "step": 7010 + }, + { + "epoch": 0.7699319130243796, + "grad_norm": 2.1855318546295166, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.702295184135437, + "num_tokens": 178477155.0, + "step": 7011 + }, + { + "epoch": 0.7700417307269932, + "grad_norm": 2.219754934310913, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.7025277614593506, + "num_tokens": 178502553.0, + "step": 7012 + }, + { + "epoch": 0.7701515484296069, + "grad_norm": 2.482142210006714, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7275657057762146, + "num_tokens": 178522084.0, + "step": 7013 + }, + { + "epoch": 0.7702613661322205, + "grad_norm": 2.500006675720215, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7177085876464844, + "num_tokens": 178541524.0, + "step": 7014 + }, + { + "epoch": 0.7703711838348342, + "grad_norm": 2.047194480895996, + "learning_rate": 1e-06, + "loss": 1.0121, + "mean_token_accuracy": 0.6940702199935913, + "num_tokens": 178568947.0, + "step": 7015 + }, + { + "epoch": 0.7704810015374478, + "grad_norm": 2.163480281829834, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7113129496574402, + "num_tokens": 178594525.0, + "step": 7016 + }, + { + "epoch": 0.7705908192400615, + "grad_norm": 2.3478803634643555, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.7092151641845703, + "num_tokens": 178616555.0, + "step": 7017 + }, + { + "epoch": 0.7707006369426752, + "grad_norm": 2.3884267807006836, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.712221622467041, + "num_tokens": 178636518.0, + "step": 7018 + }, + { + "epoch": 0.7708104546452889, + "grad_norm": 2.08904767036438, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6915736794471741, + "num_tokens": 178664731.0, + "step": 7019 + }, + { + "epoch": 0.7709202723479025, + "grad_norm": 2.035137414932251, + "learning_rate": 1e-06, + "loss": 1.0411, + "mean_token_accuracy": 0.6829183101654053, + "num_tokens": 178694066.0, + "step": 7020 + }, + { + "epoch": 0.7710300900505161, + "grad_norm": 2.0629773139953613, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7190752029418945, + "num_tokens": 178719246.0, + "step": 7021 + }, + { + "epoch": 0.7711399077531298, + "grad_norm": 2.1872191429138184, + "learning_rate": 1e-06, + "loss": 1.0485, + "mean_token_accuracy": 0.6870625019073486, + "num_tokens": 178746815.0, + "step": 7022 + }, + { + "epoch": 0.7712497254557434, + "grad_norm": 2.3773062229156494, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.7116773128509521, + "num_tokens": 178768383.0, + "step": 7023 + }, + { + "epoch": 0.7713595431583571, + "grad_norm": 2.2693028450012207, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7173696756362915, + "num_tokens": 178792530.0, + "step": 7024 + }, + { + "epoch": 0.7714693608609708, + "grad_norm": 2.1503520011901855, + "learning_rate": 1e-06, + "loss": 1.0181, + "mean_token_accuracy": 0.6962374448776245, + "num_tokens": 178820932.0, + "step": 7025 + }, + { + "epoch": 0.7715791785635845, + "grad_norm": 2.213486433029175, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7015779614448547, + "num_tokens": 178845988.0, + "step": 7026 + }, + { + "epoch": 0.7716889962661981, + "grad_norm": 2.3396050930023193, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7111973762512207, + "num_tokens": 178871880.0, + "step": 7027 + }, + { + "epoch": 0.7717988139688118, + "grad_norm": 2.3159432411193848, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7071323394775391, + "num_tokens": 178894657.0, + "step": 7028 + }, + { + "epoch": 0.7719086316714254, + "grad_norm": 2.361875534057617, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7223526239395142, + "num_tokens": 178919269.0, + "step": 7029 + }, + { + "epoch": 0.7720184493740391, + "grad_norm": 2.5379700660705566, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.7094937562942505, + "num_tokens": 178939752.0, + "step": 7030 + }, + { + "epoch": 0.7721282670766527, + "grad_norm": 2.2473831176757812, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7275493741035461, + "num_tokens": 178963914.0, + "step": 7031 + }, + { + "epoch": 0.7722380847792664, + "grad_norm": 2.318270206451416, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.6898471713066101, + "num_tokens": 178988180.0, + "step": 7032 + }, + { + "epoch": 0.7723479024818801, + "grad_norm": 2.413767099380493, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7072485685348511, + "num_tokens": 179010560.0, + "step": 7033 + }, + { + "epoch": 0.7724577201844938, + "grad_norm": 2.1634130477905273, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.6914404630661011, + "num_tokens": 179036996.0, + "step": 7034 + }, + { + "epoch": 0.7725675378871074, + "grad_norm": 2.4127447605133057, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7286100387573242, + "num_tokens": 179057948.0, + "step": 7035 + }, + { + "epoch": 0.7726773555897211, + "grad_norm": 2.010438919067383, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7229050397872925, + "num_tokens": 179085423.0, + "step": 7036 + }, + { + "epoch": 0.7727871732923347, + "grad_norm": 2.237415313720703, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.69572514295578, + "num_tokens": 179112364.0, + "step": 7037 + }, + { + "epoch": 0.7728969909949484, + "grad_norm": 2.526712656021118, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7119787931442261, + "num_tokens": 179134038.0, + "step": 7038 + }, + { + "epoch": 0.773006808697562, + "grad_norm": 2.2665586471557617, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.704947292804718, + "num_tokens": 179157565.0, + "step": 7039 + }, + { + "epoch": 0.7731166264001758, + "grad_norm": 2.338656425476074, + "learning_rate": 1e-06, + "loss": 1.0144, + "mean_token_accuracy": 0.6966036558151245, + "num_tokens": 179181547.0, + "step": 7040 + }, + { + "epoch": 0.7732264441027894, + "grad_norm": 2.2615857124328613, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.6982327699661255, + "num_tokens": 179206537.0, + "step": 7041 + }, + { + "epoch": 0.773336261805403, + "grad_norm": 2.0217535495758057, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7133514881134033, + "num_tokens": 179234498.0, + "step": 7042 + }, + { + "epoch": 0.7734460795080167, + "grad_norm": 2.2005980014801025, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7196398973464966, + "num_tokens": 179257519.0, + "step": 7043 + }, + { + "epoch": 0.7735558972106303, + "grad_norm": 2.16092848777771, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7186211347579956, + "num_tokens": 179283973.0, + "step": 7044 + }, + { + "epoch": 0.773665714913244, + "grad_norm": 2.051358699798584, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.6872943639755249, + "num_tokens": 179314493.0, + "step": 7045 + }, + { + "epoch": 0.7737755326158576, + "grad_norm": 2.0165324211120605, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7029315233230591, + "num_tokens": 179341092.0, + "step": 7046 + }, + { + "epoch": 0.7738853503184714, + "grad_norm": 2.0918171405792236, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7136253118515015, + "num_tokens": 179368657.0, + "step": 7047 + }, + { + "epoch": 0.773995168021085, + "grad_norm": 2.303001642227173, + "learning_rate": 1e-06, + "loss": 1.0319, + "mean_token_accuracy": 0.6868008971214294, + "num_tokens": 179394378.0, + "step": 7048 + }, + { + "epoch": 0.7741049857236987, + "grad_norm": 2.234266757965088, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7116117477416992, + "num_tokens": 179419118.0, + "step": 7049 + }, + { + "epoch": 0.7742148034263123, + "grad_norm": 2.344121217727661, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7307698726654053, + "num_tokens": 179440779.0, + "step": 7050 + }, + { + "epoch": 0.774324621128926, + "grad_norm": 2.244955062866211, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7102899551391602, + "num_tokens": 179465189.0, + "step": 7051 + }, + { + "epoch": 0.7744344388315396, + "grad_norm": 2.190399646759033, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7182759046554565, + "num_tokens": 179490224.0, + "step": 7052 + }, + { + "epoch": 0.7745442565341533, + "grad_norm": 2.664813280105591, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7159676551818848, + "num_tokens": 179508843.0, + "step": 7053 + }, + { + "epoch": 0.774654074236767, + "grad_norm": 2.263019561767578, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7127641439437866, + "num_tokens": 179533884.0, + "step": 7054 + }, + { + "epoch": 0.7747638919393807, + "grad_norm": 2.2872252464294434, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7209081649780273, + "num_tokens": 179557209.0, + "step": 7055 + }, + { + "epoch": 0.7748737096419943, + "grad_norm": 2.2179408073425293, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7091213464736938, + "num_tokens": 179582750.0, + "step": 7056 + }, + { + "epoch": 0.774983527344608, + "grad_norm": 2.3676483631134033, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7099995017051697, + "num_tokens": 179604551.0, + "step": 7057 + }, + { + "epoch": 0.7750933450472216, + "grad_norm": 2.2376294136047363, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7043637037277222, + "num_tokens": 179629996.0, + "step": 7058 + }, + { + "epoch": 0.7752031627498353, + "grad_norm": 2.139610528945923, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6943565607070923, + "num_tokens": 179660533.0, + "step": 7059 + }, + { + "epoch": 0.7753129804524489, + "grad_norm": 2.570162296295166, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7124526500701904, + "num_tokens": 179680458.0, + "step": 7060 + }, + { + "epoch": 0.7754227981550625, + "grad_norm": 2.809044361114502, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7307072281837463, + "num_tokens": 179698347.0, + "step": 7061 + }, + { + "epoch": 0.7755326158576763, + "grad_norm": 2.33400821685791, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7200609445571899, + "num_tokens": 179720529.0, + "step": 7062 + }, + { + "epoch": 0.77564243356029, + "grad_norm": 2.5035533905029297, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.729834258556366, + "num_tokens": 179739199.0, + "step": 7063 + }, + { + "epoch": 0.7757522512629036, + "grad_norm": 2.015939950942993, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.709296464920044, + "num_tokens": 179767192.0, + "step": 7064 + }, + { + "epoch": 0.7758620689655172, + "grad_norm": 2.2370896339416504, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.7039629220962524, + "num_tokens": 179791297.0, + "step": 7065 + }, + { + "epoch": 0.7759718866681309, + "grad_norm": 2.1393558979034424, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.7005105018615723, + "num_tokens": 179815584.0, + "step": 7066 + }, + { + "epoch": 0.7760817043707445, + "grad_norm": 2.3485267162323, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.733845591545105, + "num_tokens": 179837594.0, + "step": 7067 + }, + { + "epoch": 0.7761915220733582, + "grad_norm": 1.9820647239685059, + "learning_rate": 1e-06, + "loss": 1.0082, + "mean_token_accuracy": 0.6999461650848389, + "num_tokens": 179867410.0, + "step": 7068 + }, + { + "epoch": 0.7763013397759719, + "grad_norm": 2.1995654106140137, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7077186703681946, + "num_tokens": 179891976.0, + "step": 7069 + }, + { + "epoch": 0.7764111574785856, + "grad_norm": 2.4502363204956055, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.720923662185669, + "num_tokens": 179911209.0, + "step": 7070 + }, + { + "epoch": 0.7765209751811992, + "grad_norm": 2.255143404006958, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7116295099258423, + "num_tokens": 179936031.0, + "step": 7071 + }, + { + "epoch": 0.7766307928838129, + "grad_norm": 2.5902698040008545, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7079870104789734, + "num_tokens": 179956644.0, + "step": 7072 + }, + { + "epoch": 0.7767406105864265, + "grad_norm": 2.3826937675476074, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.6899970769882202, + "num_tokens": 179980011.0, + "step": 7073 + }, + { + "epoch": 0.7768504282890402, + "grad_norm": 2.2036972045898438, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7193991541862488, + "num_tokens": 180003439.0, + "step": 7074 + }, + { + "epoch": 0.7769602459916538, + "grad_norm": 2.2933502197265625, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7165653705596924, + "num_tokens": 180025974.0, + "step": 7075 + }, + { + "epoch": 0.7770700636942676, + "grad_norm": 2.0076282024383545, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7227152585983276, + "num_tokens": 180053941.0, + "step": 7076 + }, + { + "epoch": 0.7771798813968812, + "grad_norm": 2.1549062728881836, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6929835081100464, + "num_tokens": 180078429.0, + "step": 7077 + }, + { + "epoch": 0.7772896990994949, + "grad_norm": 1.935451865196228, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6895453333854675, + "num_tokens": 180110981.0, + "step": 7078 + }, + { + "epoch": 0.7773995168021085, + "grad_norm": 2.2806217670440674, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7216495871543884, + "num_tokens": 180132853.0, + "step": 7079 + }, + { + "epoch": 0.7775093345047221, + "grad_norm": 2.0662243366241455, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6906290054321289, + "num_tokens": 180161310.0, + "step": 7080 + }, + { + "epoch": 0.7776191522073358, + "grad_norm": 2.0358171463012695, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7095612287521362, + "num_tokens": 180188679.0, + "step": 7081 + }, + { + "epoch": 0.7777289699099494, + "grad_norm": 2.440117120742798, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.7039654850959778, + "num_tokens": 180209774.0, + "step": 7082 + }, + { + "epoch": 0.7778387876125632, + "grad_norm": 1.8775887489318848, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.708965003490448, + "num_tokens": 180241188.0, + "step": 7083 + }, + { + "epoch": 0.7779486053151768, + "grad_norm": 2.1366870403289795, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6927603483200073, + "num_tokens": 180268438.0, + "step": 7084 + }, + { + "epoch": 0.7780584230177905, + "grad_norm": 2.7081897258758545, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7172584533691406, + "num_tokens": 180285232.0, + "step": 7085 + }, + { + "epoch": 0.7781682407204041, + "grad_norm": 2.2002980709075928, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.717560887336731, + "num_tokens": 180310109.0, + "step": 7086 + }, + { + "epoch": 0.7782780584230178, + "grad_norm": 2.322772741317749, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.703023374080658, + "num_tokens": 180334336.0, + "step": 7087 + }, + { + "epoch": 0.7783878761256314, + "grad_norm": 2.065047264099121, + "learning_rate": 1e-06, + "loss": 1.0941, + "mean_token_accuracy": 0.6799150705337524, + "num_tokens": 180365304.0, + "step": 7088 + }, + { + "epoch": 0.7784976938282451, + "grad_norm": 2.4373116493225098, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7100350856781006, + "num_tokens": 180386313.0, + "step": 7089 + }, + { + "epoch": 0.7786075115308587, + "grad_norm": 2.504131317138672, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7425276041030884, + "num_tokens": 180403150.0, + "step": 7090 + }, + { + "epoch": 0.7787173292334725, + "grad_norm": 2.4553871154785156, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7008295655250549, + "num_tokens": 180425642.0, + "step": 7091 + }, + { + "epoch": 0.7788271469360861, + "grad_norm": 2.2099976539611816, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.6788656711578369, + "num_tokens": 180452958.0, + "step": 7092 + }, + { + "epoch": 0.7789369646386998, + "grad_norm": 1.9196833372116089, + "learning_rate": 1e-06, + "loss": 1.0497, + "mean_token_accuracy": 0.6822556853294373, + "num_tokens": 180486941.0, + "step": 7093 + }, + { + "epoch": 0.7790467823413134, + "grad_norm": 2.4300646781921387, + "learning_rate": 1e-06, + "loss": 1.0279, + "mean_token_accuracy": 0.6873807907104492, + "num_tokens": 180508003.0, + "step": 7094 + }, + { + "epoch": 0.7791566000439271, + "grad_norm": 2.405653476715088, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7171603441238403, + "num_tokens": 180529733.0, + "step": 7095 + }, + { + "epoch": 0.7792664177465407, + "grad_norm": 2.3486053943634033, + "learning_rate": 1e-06, + "loss": 1.0193, + "mean_token_accuracy": 0.6949982643127441, + "num_tokens": 180554920.0, + "step": 7096 + }, + { + "epoch": 0.7793762354491544, + "grad_norm": 2.272878885269165, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7080156803131104, + "num_tokens": 180576829.0, + "step": 7097 + }, + { + "epoch": 0.7794860531517681, + "grad_norm": 2.2584493160247803, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7211204767227173, + "num_tokens": 180599707.0, + "step": 7098 + }, + { + "epoch": 0.7795958708543818, + "grad_norm": 2.0589652061462402, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7045738101005554, + "num_tokens": 180628871.0, + "step": 7099 + }, + { + "epoch": 0.7797056885569954, + "grad_norm": 2.147455930709839, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7127938270568848, + "num_tokens": 180655754.0, + "step": 7100 + }, + { + "epoch": 0.779815506259609, + "grad_norm": 2.1833221912384033, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6921144723892212, + "num_tokens": 180682688.0, + "step": 7101 + }, + { + "epoch": 0.7799253239622227, + "grad_norm": 2.0722649097442627, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7135425806045532, + "num_tokens": 180710652.0, + "step": 7102 + }, + { + "epoch": 0.7800351416648363, + "grad_norm": 2.285128355026245, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7121323347091675, + "num_tokens": 180733516.0, + "step": 7103 + }, + { + "epoch": 0.78014495936745, + "grad_norm": 2.1312835216522217, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.7019756436347961, + "num_tokens": 180759660.0, + "step": 7104 + }, + { + "epoch": 0.7802547770700637, + "grad_norm": 2.0694477558135986, + "learning_rate": 1e-06, + "loss": 1.0774, + "mean_token_accuracy": 0.6761258840560913, + "num_tokens": 180789974.0, + "step": 7105 + }, + { + "epoch": 0.7803645947726774, + "grad_norm": 1.8867919445037842, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7047412395477295, + "num_tokens": 180821025.0, + "step": 7106 + }, + { + "epoch": 0.780474412475291, + "grad_norm": 2.3215126991271973, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6935642957687378, + "num_tokens": 180844085.0, + "step": 7107 + }, + { + "epoch": 0.7805842301779047, + "grad_norm": 1.9866058826446533, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7133152484893799, + "num_tokens": 180869452.0, + "step": 7108 + }, + { + "epoch": 0.7806940478805183, + "grad_norm": 2.0200607776641846, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7156033515930176, + "num_tokens": 180895867.0, + "step": 7109 + }, + { + "epoch": 0.780803865583132, + "grad_norm": 2.2874679565429688, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7133011817932129, + "num_tokens": 180918613.0, + "step": 7110 + }, + { + "epoch": 0.7809136832857456, + "grad_norm": 2.1897032260894775, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6904687881469727, + "num_tokens": 180944219.0, + "step": 7111 + }, + { + "epoch": 0.7810235009883594, + "grad_norm": 1.9530644416809082, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7008522748947144, + "num_tokens": 180974124.0, + "step": 7112 + }, + { + "epoch": 0.781133318690973, + "grad_norm": 2.2525863647460938, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7183753848075867, + "num_tokens": 180999093.0, + "step": 7113 + }, + { + "epoch": 0.7812431363935867, + "grad_norm": 2.2509336471557617, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7193088531494141, + "num_tokens": 181022126.0, + "step": 7114 + }, + { + "epoch": 0.7813529540962003, + "grad_norm": 2.2771360874176025, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7091630697250366, + "num_tokens": 181045494.0, + "step": 7115 + }, + { + "epoch": 0.781462771798814, + "grad_norm": 2.175954818725586, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7087693214416504, + "num_tokens": 181070152.0, + "step": 7116 + }, + { + "epoch": 0.7815725895014276, + "grad_norm": 2.121731996536255, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6990137100219727, + "num_tokens": 181097285.0, + "step": 7117 + }, + { + "epoch": 0.7816824072040413, + "grad_norm": 2.409994125366211, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7171352505683899, + "num_tokens": 181118526.0, + "step": 7118 + }, + { + "epoch": 0.7817922249066549, + "grad_norm": 2.0701496601104736, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7063425779342651, + "num_tokens": 181147073.0, + "step": 7119 + }, + { + "epoch": 0.7819020426092687, + "grad_norm": 2.160616397857666, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7053614854812622, + "num_tokens": 181173251.0, + "step": 7120 + }, + { + "epoch": 0.7820118603118823, + "grad_norm": 2.222991704940796, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7006815671920776, + "num_tokens": 181198456.0, + "step": 7121 + }, + { + "epoch": 0.782121678014496, + "grad_norm": 2.362018346786499, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.6966714859008789, + "num_tokens": 181222012.0, + "step": 7122 + }, + { + "epoch": 0.7822314957171096, + "grad_norm": 2.0584065914154053, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7012622356414795, + "num_tokens": 181252513.0, + "step": 7123 + }, + { + "epoch": 0.7823413134197232, + "grad_norm": 2.523916721343994, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.7022656202316284, + "num_tokens": 181280732.0, + "step": 7124 + }, + { + "epoch": 0.7824511311223369, + "grad_norm": 2.0320136547088623, + "learning_rate": 1e-06, + "loss": 1.0368, + "mean_token_accuracy": 0.6871711611747742, + "num_tokens": 181310753.0, + "step": 7125 + }, + { + "epoch": 0.7825609488249505, + "grad_norm": 1.9773437976837158, + "learning_rate": 1e-06, + "loss": 1.0768, + "mean_token_accuracy": 0.6741818189620972, + "num_tokens": 181339524.0, + "step": 7126 + }, + { + "epoch": 0.7826707665275643, + "grad_norm": 2.3602588176727295, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7313635349273682, + "num_tokens": 181360024.0, + "step": 7127 + }, + { + "epoch": 0.7827805842301779, + "grad_norm": 2.1328125, + "learning_rate": 1e-06, + "loss": 1.0818, + "mean_token_accuracy": 0.6733928918838501, + "num_tokens": 181387210.0, + "step": 7128 + }, + { + "epoch": 0.7828904019327916, + "grad_norm": 2.1983397006988525, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6964049339294434, + "num_tokens": 181412403.0, + "step": 7129 + }, + { + "epoch": 0.7830002196354052, + "grad_norm": 2.22096848487854, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7345068454742432, + "num_tokens": 181434225.0, + "step": 7130 + }, + { + "epoch": 0.7831100373380189, + "grad_norm": 2.147475242614746, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6924663782119751, + "num_tokens": 181461568.0, + "step": 7131 + }, + { + "epoch": 0.7832198550406325, + "grad_norm": 2.1995396614074707, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.705579400062561, + "num_tokens": 181487980.0, + "step": 7132 + }, + { + "epoch": 0.7833296727432462, + "grad_norm": 2.0473313331604004, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6983088850975037, + "num_tokens": 181516578.0, + "step": 7133 + }, + { + "epoch": 0.7834394904458599, + "grad_norm": 2.3076484203338623, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7085336446762085, + "num_tokens": 181540444.0, + "step": 7134 + }, + { + "epoch": 0.7835493081484736, + "grad_norm": 2.391990900039673, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7148414254188538, + "num_tokens": 181561433.0, + "step": 7135 + }, + { + "epoch": 0.7836591258510872, + "grad_norm": 2.3396177291870117, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6882325410842896, + "num_tokens": 181582807.0, + "step": 7136 + }, + { + "epoch": 0.7837689435537009, + "grad_norm": 2.030627965927124, + "learning_rate": 1e-06, + "loss": 1.0528, + "mean_token_accuracy": 0.6893717646598816, + "num_tokens": 181612964.0, + "step": 7137 + }, + { + "epoch": 0.7838787612563145, + "grad_norm": 2.571991205215454, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7084308862686157, + "num_tokens": 181635611.0, + "step": 7138 + }, + { + "epoch": 0.7839885789589282, + "grad_norm": 2.3502113819122314, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7348207235336304, + "num_tokens": 181655960.0, + "step": 7139 + }, + { + "epoch": 0.7840983966615418, + "grad_norm": 2.1286637783050537, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7299633026123047, + "num_tokens": 181681485.0, + "step": 7140 + }, + { + "epoch": 0.7842082143641556, + "grad_norm": 1.9363974332809448, + "learning_rate": 1e-06, + "loss": 1.0421, + "mean_token_accuracy": 0.6904017925262451, + "num_tokens": 181713980.0, + "step": 7141 + }, + { + "epoch": 0.7843180320667692, + "grad_norm": 2.093181848526001, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7074697017669678, + "num_tokens": 181740035.0, + "step": 7142 + }, + { + "epoch": 0.7844278497693828, + "grad_norm": 2.1376922130584717, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.6942099928855896, + "num_tokens": 181767683.0, + "step": 7143 + }, + { + "epoch": 0.7845376674719965, + "grad_norm": 2.207066774368286, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7081483602523804, + "num_tokens": 181791549.0, + "step": 7144 + }, + { + "epoch": 0.7846474851746101, + "grad_norm": 2.0167617797851562, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7050365805625916, + "num_tokens": 181818877.0, + "step": 7145 + }, + { + "epoch": 0.7847573028772238, + "grad_norm": 2.355783462524414, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.7039515972137451, + "num_tokens": 181841573.0, + "step": 7146 + }, + { + "epoch": 0.7848671205798374, + "grad_norm": 2.4016382694244385, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.7123461961746216, + "num_tokens": 181866249.0, + "step": 7147 + }, + { + "epoch": 0.7849769382824512, + "grad_norm": 2.271408796310425, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.7098497152328491, + "num_tokens": 181890311.0, + "step": 7148 + }, + { + "epoch": 0.7850867559850648, + "grad_norm": 1.9657727479934692, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.7021688222885132, + "num_tokens": 181919165.0, + "step": 7149 + }, + { + "epoch": 0.7851965736876785, + "grad_norm": 2.115602493286133, + "learning_rate": 1e-06, + "loss": 0.9966, + "mean_token_accuracy": 0.7019276022911072, + "num_tokens": 181946701.0, + "step": 7150 + }, + { + "epoch": 0.7853063913902921, + "grad_norm": 2.1931707859039307, + "learning_rate": 1e-06, + "loss": 1.0355, + "mean_token_accuracy": 0.6840739250183105, + "num_tokens": 181971937.0, + "step": 7151 + }, + { + "epoch": 0.7854162090929058, + "grad_norm": 2.343602180480957, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7201829552650452, + "num_tokens": 181994051.0, + "step": 7152 + }, + { + "epoch": 0.7855260267955194, + "grad_norm": 2.047348976135254, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7044540047645569, + "num_tokens": 182022119.0, + "step": 7153 + }, + { + "epoch": 0.7856358444981331, + "grad_norm": 2.143852472305298, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.7043976187705994, + "num_tokens": 182049961.0, + "step": 7154 + }, + { + "epoch": 0.7857456622007467, + "grad_norm": 2.4457144737243652, + "learning_rate": 1e-06, + "loss": 1.0391, + "mean_token_accuracy": 0.6876962780952454, + "num_tokens": 182071534.0, + "step": 7155 + }, + { + "epoch": 0.7858554799033605, + "grad_norm": 2.2053492069244385, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.7146626710891724, + "num_tokens": 182096427.0, + "step": 7156 + }, + { + "epoch": 0.7859652976059741, + "grad_norm": 2.1830203533172607, + "learning_rate": 1e-06, + "loss": 0.9903, + "mean_token_accuracy": 0.7013662457466125, + "num_tokens": 182122723.0, + "step": 7157 + }, + { + "epoch": 0.7860751153085878, + "grad_norm": 2.096858501434326, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.742032527923584, + "num_tokens": 182145579.0, + "step": 7158 + }, + { + "epoch": 0.7861849330112014, + "grad_norm": 2.5997183322906494, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7019121646881104, + "num_tokens": 182165515.0, + "step": 7159 + }, + { + "epoch": 0.786294750713815, + "grad_norm": 2.3602888584136963, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7230134606361389, + "num_tokens": 182188246.0, + "step": 7160 + }, + { + "epoch": 0.7864045684164287, + "grad_norm": 2.1553092002868652, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.6966407299041748, + "num_tokens": 182213760.0, + "step": 7161 + }, + { + "epoch": 0.7865143861190423, + "grad_norm": 2.1302733421325684, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7351981401443481, + "num_tokens": 182237726.0, + "step": 7162 + }, + { + "epoch": 0.7866242038216561, + "grad_norm": 2.3128347396850586, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.6825869083404541, + "num_tokens": 182263151.0, + "step": 7163 + }, + { + "epoch": 0.7867340215242697, + "grad_norm": 2.078749656677246, + "learning_rate": 1e-06, + "loss": 1.002, + "mean_token_accuracy": 0.6981328725814819, + "num_tokens": 182293045.0, + "step": 7164 + }, + { + "epoch": 0.7868438392268834, + "grad_norm": 2.0677096843719482, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7097628712654114, + "num_tokens": 182320721.0, + "step": 7165 + }, + { + "epoch": 0.786953656929497, + "grad_norm": 2.091054677963257, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7234663367271423, + "num_tokens": 182345658.0, + "step": 7166 + }, + { + "epoch": 0.7870634746321107, + "grad_norm": 2.4835596084594727, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7039179801940918, + "num_tokens": 182368522.0, + "step": 7167 + }, + { + "epoch": 0.7871732923347243, + "grad_norm": 2.3854596614837646, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7276493906974792, + "num_tokens": 182389752.0, + "step": 7168 + }, + { + "epoch": 0.787283110037338, + "grad_norm": 1.9217654466629028, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.7043388485908508, + "num_tokens": 182420669.0, + "step": 7169 + }, + { + "epoch": 0.7873929277399517, + "grad_norm": 2.126695156097412, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7084437608718872, + "num_tokens": 182447591.0, + "step": 7170 + }, + { + "epoch": 0.7875027454425654, + "grad_norm": 2.2529077529907227, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7120983600616455, + "num_tokens": 182470597.0, + "step": 7171 + }, + { + "epoch": 0.787612563145179, + "grad_norm": 2.3909177780151367, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7120009660720825, + "num_tokens": 182491651.0, + "step": 7172 + }, + { + "epoch": 0.7877223808477927, + "grad_norm": 2.042095184326172, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.7019277811050415, + "num_tokens": 182519540.0, + "step": 7173 + }, + { + "epoch": 0.7878321985504063, + "grad_norm": 2.2380847930908203, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6979446411132812, + "num_tokens": 182544837.0, + "step": 7174 + }, + { + "epoch": 0.78794201625302, + "grad_norm": 2.548527240753174, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.699455201625824, + "num_tokens": 182565924.0, + "step": 7175 + }, + { + "epoch": 0.7880518339556336, + "grad_norm": 2.448127269744873, + "learning_rate": 1e-06, + "loss": 1.0863, + "mean_token_accuracy": 0.6705012321472168, + "num_tokens": 182590013.0, + "step": 7176 + }, + { + "epoch": 0.7881616516582474, + "grad_norm": 2.2515180110931396, + "learning_rate": 1e-06, + "loss": 1.0209, + "mean_token_accuracy": 0.7008556723594666, + "num_tokens": 182613995.0, + "step": 7177 + }, + { + "epoch": 0.788271469360861, + "grad_norm": 2.168828248977661, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7120231986045837, + "num_tokens": 182639699.0, + "step": 7178 + }, + { + "epoch": 0.7883812870634747, + "grad_norm": 2.2341878414154053, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6967717409133911, + "num_tokens": 182666867.0, + "step": 7179 + }, + { + "epoch": 0.7884911047660883, + "grad_norm": 2.4218645095825195, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7104508876800537, + "num_tokens": 182689749.0, + "step": 7180 + }, + { + "epoch": 0.788600922468702, + "grad_norm": 2.0454485416412354, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7286548018455505, + "num_tokens": 182714395.0, + "step": 7181 + }, + { + "epoch": 0.7887107401713156, + "grad_norm": 2.1954123973846436, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7125943303108215, + "num_tokens": 182738744.0, + "step": 7182 + }, + { + "epoch": 0.7888205578739292, + "grad_norm": 2.3213512897491455, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7024093866348267, + "num_tokens": 182761328.0, + "step": 7183 + }, + { + "epoch": 0.7889303755765429, + "grad_norm": 2.47625470161438, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7081823348999023, + "num_tokens": 182783213.0, + "step": 7184 + }, + { + "epoch": 0.7890401932791566, + "grad_norm": 2.3305647373199463, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.6959034204483032, + "num_tokens": 182807712.0, + "step": 7185 + }, + { + "epoch": 0.7891500109817703, + "grad_norm": 1.8942286968231201, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7011933326721191, + "num_tokens": 182838961.0, + "step": 7186 + }, + { + "epoch": 0.7892598286843839, + "grad_norm": 2.054285764694214, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7114388942718506, + "num_tokens": 182866199.0, + "step": 7187 + }, + { + "epoch": 0.7893696463869976, + "grad_norm": 2.46640944480896, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7141256928443909, + "num_tokens": 182886788.0, + "step": 7188 + }, + { + "epoch": 0.7894794640896112, + "grad_norm": 2.3012912273406982, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7145568132400513, + "num_tokens": 182909263.0, + "step": 7189 + }, + { + "epoch": 0.7895892817922249, + "grad_norm": 2.2254092693328857, + "learning_rate": 1e-06, + "loss": 1.0433, + "mean_token_accuracy": 0.6932134032249451, + "num_tokens": 182934485.0, + "step": 7190 + }, + { + "epoch": 0.7896990994948385, + "grad_norm": 2.3676371574401855, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7064663171768188, + "num_tokens": 182956924.0, + "step": 7191 + }, + { + "epoch": 0.7898089171974523, + "grad_norm": 1.9564132690429688, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7085954546928406, + "num_tokens": 182987697.0, + "step": 7192 + }, + { + "epoch": 0.7899187349000659, + "grad_norm": 1.9793205261230469, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7046023607254028, + "num_tokens": 183016407.0, + "step": 7193 + }, + { + "epoch": 0.7900285526026796, + "grad_norm": 2.2605772018432617, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7259318828582764, + "num_tokens": 183039056.0, + "step": 7194 + }, + { + "epoch": 0.7901383703052932, + "grad_norm": 1.9675763845443726, + "learning_rate": 1e-06, + "loss": 1.0018, + "mean_token_accuracy": 0.7012271881103516, + "num_tokens": 183070673.0, + "step": 7195 + }, + { + "epoch": 0.7902481880079069, + "grad_norm": 2.3926682472229004, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7251685857772827, + "num_tokens": 183092147.0, + "step": 7196 + }, + { + "epoch": 0.7903580057105205, + "grad_norm": 1.9363700151443481, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7091653943061829, + "num_tokens": 183122549.0, + "step": 7197 + }, + { + "epoch": 0.7904678234131342, + "grad_norm": 2.2922990322113037, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7186272144317627, + "num_tokens": 183144973.0, + "step": 7198 + }, + { + "epoch": 0.7905776411157479, + "grad_norm": 2.263669013977051, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7127895355224609, + "num_tokens": 183168517.0, + "step": 7199 + }, + { + "epoch": 0.7906874588183616, + "grad_norm": 2.2653629779815674, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7240413427352905, + "num_tokens": 183192016.0, + "step": 7200 + }, + { + "epoch": 0.7907972765209752, + "grad_norm": 2.3297922611236572, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7285680174827576, + "num_tokens": 183215052.0, + "step": 7201 + }, + { + "epoch": 0.7909070942235888, + "grad_norm": 2.0874664783477783, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7133663296699524, + "num_tokens": 183240806.0, + "step": 7202 + }, + { + "epoch": 0.7910169119262025, + "grad_norm": 2.0575811862945557, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7193489670753479, + "num_tokens": 183265706.0, + "step": 7203 + }, + { + "epoch": 0.7911267296288161, + "grad_norm": 2.2893617153167725, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7108150720596313, + "num_tokens": 183289462.0, + "step": 7204 + }, + { + "epoch": 0.7912365473314298, + "grad_norm": 2.2620182037353516, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6925760507583618, + "num_tokens": 183313341.0, + "step": 7205 + }, + { + "epoch": 0.7913463650340435, + "grad_norm": 2.0406723022460938, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7227723598480225, + "num_tokens": 183338503.0, + "step": 7206 + }, + { + "epoch": 0.7914561827366572, + "grad_norm": 2.2154555320739746, + "learning_rate": 1e-06, + "loss": 1.0382, + "mean_token_accuracy": 0.688060998916626, + "num_tokens": 183364781.0, + "step": 7207 + }, + { + "epoch": 0.7915660004392708, + "grad_norm": 2.245537757873535, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7133370637893677, + "num_tokens": 183389793.0, + "step": 7208 + }, + { + "epoch": 0.7916758181418845, + "grad_norm": 2.266112804412842, + "learning_rate": 1e-06, + "loss": 1.0151, + "mean_token_accuracy": 0.6916489601135254, + "num_tokens": 183415205.0, + "step": 7209 + }, + { + "epoch": 0.7917856358444981, + "grad_norm": 2.3240768909454346, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7266587615013123, + "num_tokens": 183436522.0, + "step": 7210 + }, + { + "epoch": 0.7918954535471118, + "grad_norm": 2.2047770023345947, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7224955558776855, + "num_tokens": 183461285.0, + "step": 7211 + }, + { + "epoch": 0.7920052712497254, + "grad_norm": 2.230769157409668, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7186172604560852, + "num_tokens": 183484606.0, + "step": 7212 + }, + { + "epoch": 0.7921150889523391, + "grad_norm": 2.00187611579895, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7192764282226562, + "num_tokens": 183511744.0, + "step": 7213 + }, + { + "epoch": 0.7922249066549528, + "grad_norm": 2.08547043800354, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7071573734283447, + "num_tokens": 183538026.0, + "step": 7214 + }, + { + "epoch": 0.7923347243575665, + "grad_norm": 2.417358875274658, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7141876816749573, + "num_tokens": 183558780.0, + "step": 7215 + }, + { + "epoch": 0.7924445420601801, + "grad_norm": 2.3912205696105957, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6874725818634033, + "num_tokens": 183580263.0, + "step": 7216 + }, + { + "epoch": 0.7925543597627938, + "grad_norm": 1.9386779069900513, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7179967761039734, + "num_tokens": 183608698.0, + "step": 7217 + }, + { + "epoch": 0.7926641774654074, + "grad_norm": 2.192826986312866, + "learning_rate": 1e-06, + "loss": 0.9905, + "mean_token_accuracy": 0.6961979269981384, + "num_tokens": 183632931.0, + "step": 7218 + }, + { + "epoch": 0.792773995168021, + "grad_norm": 1.8578739166259766, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6932339072227478, + "num_tokens": 183667744.0, + "step": 7219 + }, + { + "epoch": 0.7928838128706347, + "grad_norm": 2.0515689849853516, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6930323839187622, + "num_tokens": 183695604.0, + "step": 7220 + }, + { + "epoch": 0.7929936305732485, + "grad_norm": 1.9841316938400269, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7118344306945801, + "num_tokens": 183725201.0, + "step": 7221 + }, + { + "epoch": 0.7931034482758621, + "grad_norm": 2.0388224124908447, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.7032797336578369, + "num_tokens": 183755864.0, + "step": 7222 + }, + { + "epoch": 0.7932132659784757, + "grad_norm": 2.575159788131714, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7168701887130737, + "num_tokens": 183776259.0, + "step": 7223 + }, + { + "epoch": 0.7933230836810894, + "grad_norm": 2.052872896194458, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6854985356330872, + "num_tokens": 183807448.0, + "step": 7224 + }, + { + "epoch": 0.793432901383703, + "grad_norm": 2.2817251682281494, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7161561250686646, + "num_tokens": 183829723.0, + "step": 7225 + }, + { + "epoch": 0.7935427190863167, + "grad_norm": 2.511568546295166, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7140412926673889, + "num_tokens": 183849391.0, + "step": 7226 + }, + { + "epoch": 0.7936525367889303, + "grad_norm": 2.335456132888794, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.6991227865219116, + "num_tokens": 183873895.0, + "step": 7227 + }, + { + "epoch": 0.7937623544915441, + "grad_norm": 2.2889840602874756, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7049252986907959, + "num_tokens": 183899736.0, + "step": 7228 + }, + { + "epoch": 0.7938721721941577, + "grad_norm": 2.2020037174224854, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7088031768798828, + "num_tokens": 183923357.0, + "step": 7229 + }, + { + "epoch": 0.7939819898967714, + "grad_norm": 2.1693389415740967, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6914001107215881, + "num_tokens": 183949222.0, + "step": 7230 + }, + { + "epoch": 0.794091807599385, + "grad_norm": 2.565936326980591, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7265933752059937, + "num_tokens": 183968674.0, + "step": 7231 + }, + { + "epoch": 0.7942016253019987, + "grad_norm": 2.1925442218780518, + "learning_rate": 1e-06, + "loss": 1.0268, + "mean_token_accuracy": 0.7059051394462585, + "num_tokens": 183997892.0, + "step": 7232 + }, + { + "epoch": 0.7943114430046123, + "grad_norm": 2.243540048599243, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7263487577438354, + "num_tokens": 184022054.0, + "step": 7233 + }, + { + "epoch": 0.794421260707226, + "grad_norm": 2.091228723526001, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7000633478164673, + "num_tokens": 184047806.0, + "step": 7234 + }, + { + "epoch": 0.7945310784098397, + "grad_norm": 2.145796060562134, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6988997459411621, + "num_tokens": 184074544.0, + "step": 7235 + }, + { + "epoch": 0.7946408961124534, + "grad_norm": 2.154416799545288, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7226990461349487, + "num_tokens": 184102888.0, + "step": 7236 + }, + { + "epoch": 0.794750713815067, + "grad_norm": 2.446258544921875, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7167881727218628, + "num_tokens": 184126097.0, + "step": 7237 + }, + { + "epoch": 0.7948605315176807, + "grad_norm": 1.9542617797851562, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7033772468566895, + "num_tokens": 184156629.0, + "step": 7238 + }, + { + "epoch": 0.7949703492202943, + "grad_norm": 1.9923900365829468, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7174239158630371, + "num_tokens": 184186286.0, + "step": 7239 + }, + { + "epoch": 0.795080166922908, + "grad_norm": 2.121507406234741, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7090368270874023, + "num_tokens": 184213768.0, + "step": 7240 + }, + { + "epoch": 0.7951899846255216, + "grad_norm": 2.6181697845458984, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7345844507217407, + "num_tokens": 184231343.0, + "step": 7241 + }, + { + "epoch": 0.7952998023281352, + "grad_norm": 2.2753560543060303, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7235331535339355, + "num_tokens": 184253209.0, + "step": 7242 + }, + { + "epoch": 0.795409620030749, + "grad_norm": 2.2750377655029297, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7011635303497314, + "num_tokens": 184276707.0, + "step": 7243 + }, + { + "epoch": 0.7955194377333626, + "grad_norm": 2.3929479122161865, + "learning_rate": 1e-06, + "loss": 1.0574, + "mean_token_accuracy": 0.6764113306999207, + "num_tokens": 184299544.0, + "step": 7244 + }, + { + "epoch": 0.7956292554359763, + "grad_norm": 2.2275266647338867, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.714750349521637, + "num_tokens": 184325142.0, + "step": 7245 + }, + { + "epoch": 0.7957390731385899, + "grad_norm": 2.155667543411255, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7117926478385925, + "num_tokens": 184349960.0, + "step": 7246 + }, + { + "epoch": 0.7958488908412036, + "grad_norm": 2.0147275924682617, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6983298063278198, + "num_tokens": 184378713.0, + "step": 7247 + }, + { + "epoch": 0.7959587085438172, + "grad_norm": 2.124351739883423, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7271319627761841, + "num_tokens": 184403045.0, + "step": 7248 + }, + { + "epoch": 0.7960685262464309, + "grad_norm": 2.279040813446045, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7124996185302734, + "num_tokens": 184425916.0, + "step": 7249 + }, + { + "epoch": 0.7961783439490446, + "grad_norm": 2.206758975982666, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7044151425361633, + "num_tokens": 184449519.0, + "step": 7250 + }, + { + "epoch": 0.7962881616516583, + "grad_norm": 1.9968475103378296, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6963407397270203, + "num_tokens": 184479935.0, + "step": 7251 + }, + { + "epoch": 0.7963979793542719, + "grad_norm": 2.1153314113616943, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.6988297700881958, + "num_tokens": 184507114.0, + "step": 7252 + }, + { + "epoch": 0.7965077970568856, + "grad_norm": 1.9579474925994873, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.7019752860069275, + "num_tokens": 184537077.0, + "step": 7253 + }, + { + "epoch": 0.7966176147594992, + "grad_norm": 2.22961163520813, + "learning_rate": 1e-06, + "loss": 1.0288, + "mean_token_accuracy": 0.6855573654174805, + "num_tokens": 184562076.0, + "step": 7254 + }, + { + "epoch": 0.7967274324621129, + "grad_norm": 2.5579535961151123, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7066946625709534, + "num_tokens": 184581318.0, + "step": 7255 + }, + { + "epoch": 0.7968372501647265, + "grad_norm": 2.081644058227539, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7012782096862793, + "num_tokens": 184608788.0, + "step": 7256 + }, + { + "epoch": 0.7969470678673403, + "grad_norm": 2.1698665618896484, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7412111163139343, + "num_tokens": 184632384.0, + "step": 7257 + }, + { + "epoch": 0.7970568855699539, + "grad_norm": 1.8043851852416992, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6922268867492676, + "num_tokens": 184667972.0, + "step": 7258 + }, + { + "epoch": 0.7971667032725676, + "grad_norm": 2.0792248249053955, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6942195892333984, + "num_tokens": 184695626.0, + "step": 7259 + }, + { + "epoch": 0.7972765209751812, + "grad_norm": 2.222144365310669, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7071096301078796, + "num_tokens": 184721430.0, + "step": 7260 + }, + { + "epoch": 0.7973863386777948, + "grad_norm": 2.200901985168457, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7288848161697388, + "num_tokens": 184746677.0, + "step": 7261 + }, + { + "epoch": 0.7974961563804085, + "grad_norm": 2.1279327869415283, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.6988809108734131, + "num_tokens": 184772800.0, + "step": 7262 + }, + { + "epoch": 0.7976059740830221, + "grad_norm": 2.170907497406006, + "learning_rate": 1e-06, + "loss": 1.0732, + "mean_token_accuracy": 0.6776012182235718, + "num_tokens": 184800040.0, + "step": 7263 + }, + { + "epoch": 0.7977157917856359, + "grad_norm": 1.9604977369308472, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7090520262718201, + "num_tokens": 184830443.0, + "step": 7264 + }, + { + "epoch": 0.7978256094882495, + "grad_norm": 1.9407124519348145, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7154819965362549, + "num_tokens": 184859783.0, + "step": 7265 + }, + { + "epoch": 0.7979354271908632, + "grad_norm": 2.3416786193847656, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7311629056930542, + "num_tokens": 184881175.0, + "step": 7266 + }, + { + "epoch": 0.7980452448934768, + "grad_norm": 2.1113221645355225, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.6955908536911011, + "num_tokens": 184907820.0, + "step": 7267 + }, + { + "epoch": 0.7981550625960905, + "grad_norm": 2.272770404815674, + "learning_rate": 1e-06, + "loss": 1.1041, + "mean_token_accuracy": 0.6650210618972778, + "num_tokens": 184934767.0, + "step": 7268 + }, + { + "epoch": 0.7982648802987041, + "grad_norm": 2.6271755695343018, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.715364396572113, + "num_tokens": 184953483.0, + "step": 7269 + }, + { + "epoch": 0.7983746980013178, + "grad_norm": 2.1166269779205322, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.6980623006820679, + "num_tokens": 184979404.0, + "step": 7270 + }, + { + "epoch": 0.7984845157039314, + "grad_norm": 2.580040216445923, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7227232456207275, + "num_tokens": 184997463.0, + "step": 7271 + }, + { + "epoch": 0.7985943334065452, + "grad_norm": 2.497951030731201, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7363535165786743, + "num_tokens": 185016626.0, + "step": 7272 + }, + { + "epoch": 0.7987041511091588, + "grad_norm": 2.179692029953003, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7132872939109802, + "num_tokens": 185042379.0, + "step": 7273 + }, + { + "epoch": 0.7988139688117725, + "grad_norm": 2.322319984436035, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7262818813323975, + "num_tokens": 185063721.0, + "step": 7274 + }, + { + "epoch": 0.7989237865143861, + "grad_norm": 2.4087135791778564, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7267670631408691, + "num_tokens": 185084203.0, + "step": 7275 + }, + { + "epoch": 0.7990336042169998, + "grad_norm": 2.06937313079834, + "learning_rate": 1e-06, + "loss": 1.0298, + "mean_token_accuracy": 0.68527752161026, + "num_tokens": 185113672.0, + "step": 7276 + }, + { + "epoch": 0.7991434219196134, + "grad_norm": 2.0451202392578125, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7116168737411499, + "num_tokens": 185139987.0, + "step": 7277 + }, + { + "epoch": 0.799253239622227, + "grad_norm": 2.2555646896362305, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7135554552078247, + "num_tokens": 185162993.0, + "step": 7278 + }, + { + "epoch": 0.7993630573248408, + "grad_norm": 2.1786763668060303, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.6924071311950684, + "num_tokens": 185189608.0, + "step": 7279 + }, + { + "epoch": 0.7994728750274545, + "grad_norm": 2.250394821166992, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7017856240272522, + "num_tokens": 185212290.0, + "step": 7280 + }, + { + "epoch": 0.7995826927300681, + "grad_norm": 2.2336440086364746, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7235507369041443, + "num_tokens": 185234319.0, + "step": 7281 + }, + { + "epoch": 0.7996925104326817, + "grad_norm": 2.1157054901123047, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.730919361114502, + "num_tokens": 185261023.0, + "step": 7282 + }, + { + "epoch": 0.7998023281352954, + "grad_norm": 2.101046085357666, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7188394069671631, + "num_tokens": 185288683.0, + "step": 7283 + }, + { + "epoch": 0.799912145837909, + "grad_norm": 2.1484439373016357, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.706706166267395, + "num_tokens": 185315335.0, + "step": 7284 + }, + { + "epoch": 0.8000219635405227, + "grad_norm": 1.9735766649246216, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6971668004989624, + "num_tokens": 185344754.0, + "step": 7285 + }, + { + "epoch": 0.8001317812431364, + "grad_norm": 2.2614591121673584, + "learning_rate": 1e-06, + "loss": 1.0078, + "mean_token_accuracy": 0.6979819536209106, + "num_tokens": 185368593.0, + "step": 7286 + }, + { + "epoch": 0.8002415989457501, + "grad_norm": 2.3228399753570557, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.7015183568000793, + "num_tokens": 185393618.0, + "step": 7287 + }, + { + "epoch": 0.8003514166483637, + "grad_norm": 2.6542139053344727, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7078431248664856, + "num_tokens": 185411392.0, + "step": 7288 + }, + { + "epoch": 0.8004612343509774, + "grad_norm": 2.2025861740112305, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7314668297767639, + "num_tokens": 185434648.0, + "step": 7289 + }, + { + "epoch": 0.800571052053591, + "grad_norm": 2.1889195442199707, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7005099058151245, + "num_tokens": 185459164.0, + "step": 7290 + }, + { + "epoch": 0.8006808697562047, + "grad_norm": 2.1231815814971924, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7162147760391235, + "num_tokens": 185486304.0, + "step": 7291 + }, + { + "epoch": 0.8007906874588183, + "grad_norm": 1.9102253913879395, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.6924622058868408, + "num_tokens": 185518559.0, + "step": 7292 + }, + { + "epoch": 0.8009005051614321, + "grad_norm": 2.2661337852478027, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7167065739631653, + "num_tokens": 185541509.0, + "step": 7293 + }, + { + "epoch": 0.8010103228640457, + "grad_norm": 1.9643536806106567, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7049678564071655, + "num_tokens": 185574145.0, + "step": 7294 + }, + { + "epoch": 0.8011201405666594, + "grad_norm": 1.88852059841156, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.6972514390945435, + "num_tokens": 185607196.0, + "step": 7295 + }, + { + "epoch": 0.801229958269273, + "grad_norm": 2.030029296875, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7371731996536255, + "num_tokens": 185634080.0, + "step": 7296 + }, + { + "epoch": 0.8013397759718867, + "grad_norm": 2.103456974029541, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6852678060531616, + "num_tokens": 185662958.0, + "step": 7297 + }, + { + "epoch": 0.8014495936745003, + "grad_norm": 2.3667283058166504, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7234926819801331, + "num_tokens": 185684354.0, + "step": 7298 + }, + { + "epoch": 0.801559411377114, + "grad_norm": 2.109034538269043, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7028626799583435, + "num_tokens": 185709619.0, + "step": 7299 + }, + { + "epoch": 0.8016692290797276, + "grad_norm": 1.91776704788208, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6877781748771667, + "num_tokens": 185741691.0, + "step": 7300 + }, + { + "epoch": 0.8017790467823414, + "grad_norm": 2.1204030513763428, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7032122611999512, + "num_tokens": 185768874.0, + "step": 7301 + }, + { + "epoch": 0.801888864484955, + "grad_norm": 2.1621060371398926, + "learning_rate": 1e-06, + "loss": 1.0479, + "mean_token_accuracy": 0.6812678575515747, + "num_tokens": 185796078.0, + "step": 7302 + }, + { + "epoch": 0.8019986821875686, + "grad_norm": 2.328077554702759, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7061013579368591, + "num_tokens": 185819891.0, + "step": 7303 + }, + { + "epoch": 0.8021084998901823, + "grad_norm": 2.300165891647339, + "learning_rate": 1e-06, + "loss": 1.0594, + "mean_token_accuracy": 0.6817097663879395, + "num_tokens": 185843655.0, + "step": 7304 + }, + { + "epoch": 0.8022183175927959, + "grad_norm": 2.5214054584503174, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7306520938873291, + "num_tokens": 185862241.0, + "step": 7305 + }, + { + "epoch": 0.8023281352954096, + "grad_norm": 2.2283103466033936, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6886083483695984, + "num_tokens": 185886293.0, + "step": 7306 + }, + { + "epoch": 0.8024379529980232, + "grad_norm": 2.333869218826294, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7300413846969604, + "num_tokens": 185905811.0, + "step": 7307 + }, + { + "epoch": 0.802547770700637, + "grad_norm": 2.149545907974243, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7120764255523682, + "num_tokens": 185930080.0, + "step": 7308 + }, + { + "epoch": 0.8026575884032506, + "grad_norm": 2.573610782623291, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7091943621635437, + "num_tokens": 185949639.0, + "step": 7309 + }, + { + "epoch": 0.8027674061058643, + "grad_norm": 2.3772668838500977, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.721876859664917, + "num_tokens": 185969747.0, + "step": 7310 + }, + { + "epoch": 0.8028772238084779, + "grad_norm": 2.2875301837921143, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7168788909912109, + "num_tokens": 185993016.0, + "step": 7311 + }, + { + "epoch": 0.8029870415110916, + "grad_norm": 2.0516457557678223, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.71622234582901, + "num_tokens": 186020679.0, + "step": 7312 + }, + { + "epoch": 0.8030968592137052, + "grad_norm": 2.2041726112365723, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7099022269248962, + "num_tokens": 186044196.0, + "step": 7313 + }, + { + "epoch": 0.8032066769163189, + "grad_norm": 2.4012835025787354, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7138064503669739, + "num_tokens": 186065472.0, + "step": 7314 + }, + { + "epoch": 0.8033164946189326, + "grad_norm": 2.0720105171203613, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.735477089881897, + "num_tokens": 186091149.0, + "step": 7315 + }, + { + "epoch": 0.8034263123215463, + "grad_norm": 2.398517608642578, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7167909145355225, + "num_tokens": 186110892.0, + "step": 7316 + }, + { + "epoch": 0.8035361300241599, + "grad_norm": 2.162484645843506, + "learning_rate": 1e-06, + "loss": 1.0282, + "mean_token_accuracy": 0.6876262426376343, + "num_tokens": 186136142.0, + "step": 7317 + }, + { + "epoch": 0.8036459477267736, + "grad_norm": 2.19067645072937, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7205530405044556, + "num_tokens": 186160005.0, + "step": 7318 + }, + { + "epoch": 0.8037557654293872, + "grad_norm": 2.071051597595215, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7162678837776184, + "num_tokens": 186186601.0, + "step": 7319 + }, + { + "epoch": 0.8038655831320008, + "grad_norm": 2.4288675785064697, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7011854648590088, + "num_tokens": 186207766.0, + "step": 7320 + }, + { + "epoch": 0.8039754008346145, + "grad_norm": 2.358365774154663, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7317208051681519, + "num_tokens": 186228568.0, + "step": 7321 + }, + { + "epoch": 0.8040852185372283, + "grad_norm": 2.000321388244629, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.7034360766410828, + "num_tokens": 186258229.0, + "step": 7322 + }, + { + "epoch": 0.8041950362398419, + "grad_norm": 2.177717447280884, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7306646108627319, + "num_tokens": 186282405.0, + "step": 7323 + }, + { + "epoch": 0.8043048539424555, + "grad_norm": 2.193455457687378, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6997350454330444, + "num_tokens": 186308074.0, + "step": 7324 + }, + { + "epoch": 0.8044146716450692, + "grad_norm": 1.8541814088821411, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6938702464103699, + "num_tokens": 186342830.0, + "step": 7325 + }, + { + "epoch": 0.8045244893476828, + "grad_norm": 2.0379271507263184, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7228635549545288, + "num_tokens": 186370890.0, + "step": 7326 + }, + { + "epoch": 0.8046343070502965, + "grad_norm": 1.9835443496704102, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.705985426902771, + "num_tokens": 186398515.0, + "step": 7327 + }, + { + "epoch": 0.8047441247529101, + "grad_norm": 2.0770132541656494, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.704542338848114, + "num_tokens": 186424958.0, + "step": 7328 + }, + { + "epoch": 0.8048539424555239, + "grad_norm": 2.121830463409424, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.6999877691268921, + "num_tokens": 186451638.0, + "step": 7329 + }, + { + "epoch": 0.8049637601581375, + "grad_norm": 2.4962735176086426, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7216683626174927, + "num_tokens": 186471718.0, + "step": 7330 + }, + { + "epoch": 0.8050735778607512, + "grad_norm": 2.1316018104553223, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7078561782836914, + "num_tokens": 186498936.0, + "step": 7331 + }, + { + "epoch": 0.8051833955633648, + "grad_norm": 2.228803873062134, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7119228839874268, + "num_tokens": 186524745.0, + "step": 7332 + }, + { + "epoch": 0.8052932132659785, + "grad_norm": 1.9287066459655762, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7064514756202698, + "num_tokens": 186555840.0, + "step": 7333 + }, + { + "epoch": 0.8054030309685921, + "grad_norm": 2.1356940269470215, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.7030612230300903, + "num_tokens": 186584197.0, + "step": 7334 + }, + { + "epoch": 0.8055128486712058, + "grad_norm": 1.9978022575378418, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.6829717755317688, + "num_tokens": 186615352.0, + "step": 7335 + }, + { + "epoch": 0.8056226663738194, + "grad_norm": 1.9931483268737793, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.6998926401138306, + "num_tokens": 186643144.0, + "step": 7336 + }, + { + "epoch": 0.8057324840764332, + "grad_norm": 2.5109331607818604, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7350287437438965, + "num_tokens": 186661460.0, + "step": 7337 + }, + { + "epoch": 0.8058423017790468, + "grad_norm": 1.891268014907837, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.6997687816619873, + "num_tokens": 186691970.0, + "step": 7338 + }, + { + "epoch": 0.8059521194816605, + "grad_norm": 2.178785800933838, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.715705394744873, + "num_tokens": 186715919.0, + "step": 7339 + }, + { + "epoch": 0.8060619371842741, + "grad_norm": 2.2161574363708496, + "learning_rate": 1e-06, + "loss": 1.0365, + "mean_token_accuracy": 0.7013716101646423, + "num_tokens": 186739885.0, + "step": 7340 + }, + { + "epoch": 0.8061717548868877, + "grad_norm": 2.2395057678222656, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.7074045538902283, + "num_tokens": 186765087.0, + "step": 7341 + }, + { + "epoch": 0.8062815725895014, + "grad_norm": 2.1930367946624756, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7143856287002563, + "num_tokens": 186790223.0, + "step": 7342 + }, + { + "epoch": 0.806391390292115, + "grad_norm": 2.220493793487549, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7075192928314209, + "num_tokens": 186813881.0, + "step": 7343 + }, + { + "epoch": 0.8065012079947288, + "grad_norm": 2.271951675415039, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7170242071151733, + "num_tokens": 186836631.0, + "step": 7344 + }, + { + "epoch": 0.8066110256973424, + "grad_norm": 2.236889362335205, + "learning_rate": 1e-06, + "loss": 1.0364, + "mean_token_accuracy": 0.6851644515991211, + "num_tokens": 186863167.0, + "step": 7345 + }, + { + "epoch": 0.8067208433999561, + "grad_norm": 2.137176990509033, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7180005311965942, + "num_tokens": 186889050.0, + "step": 7346 + }, + { + "epoch": 0.8068306611025697, + "grad_norm": 2.0454914569854736, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6937608122825623, + "num_tokens": 186917549.0, + "step": 7347 + }, + { + "epoch": 0.8069404788051834, + "grad_norm": 1.9139586687088013, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.6889762282371521, + "num_tokens": 186951381.0, + "step": 7348 + }, + { + "epoch": 0.807050296507797, + "grad_norm": 2.2733571529388428, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.7037720680236816, + "num_tokens": 186974833.0, + "step": 7349 + }, + { + "epoch": 0.8071601142104107, + "grad_norm": 1.9604272842407227, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.7035269141197205, + "num_tokens": 187002519.0, + "step": 7350 + }, + { + "epoch": 0.8072699319130244, + "grad_norm": 2.2648494243621826, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.68745356798172, + "num_tokens": 187027875.0, + "step": 7351 + }, + { + "epoch": 0.8073797496156381, + "grad_norm": 2.1520020961761475, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6922010183334351, + "num_tokens": 187055022.0, + "step": 7352 + }, + { + "epoch": 0.8074895673182517, + "grad_norm": 2.1027395725250244, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7180456519126892, + "num_tokens": 187081372.0, + "step": 7353 + }, + { + "epoch": 0.8075993850208654, + "grad_norm": 2.226757526397705, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.6828176379203796, + "num_tokens": 187106569.0, + "step": 7354 + }, + { + "epoch": 0.807709202723479, + "grad_norm": 2.2497808933258057, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7180189490318298, + "num_tokens": 187129571.0, + "step": 7355 + }, + { + "epoch": 0.8078190204260927, + "grad_norm": 2.080674171447754, + "learning_rate": 1e-06, + "loss": 0.9919, + "mean_token_accuracy": 0.6916471123695374, + "num_tokens": 187157714.0, + "step": 7356 + }, + { + "epoch": 0.8079288381287063, + "grad_norm": 2.022618293762207, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7277519702911377, + "num_tokens": 187184052.0, + "step": 7357 + }, + { + "epoch": 0.8080386558313201, + "grad_norm": 2.2057912349700928, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7057756781578064, + "num_tokens": 187207126.0, + "step": 7358 + }, + { + "epoch": 0.8081484735339337, + "grad_norm": 2.4346091747283936, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7143344879150391, + "num_tokens": 187228741.0, + "step": 7359 + }, + { + "epoch": 0.8082582912365474, + "grad_norm": 2.2159101963043213, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7282188534736633, + "num_tokens": 187251580.0, + "step": 7360 + }, + { + "epoch": 0.808368108939161, + "grad_norm": 1.9435961246490479, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7087757587432861, + "num_tokens": 187282552.0, + "step": 7361 + }, + { + "epoch": 0.8084779266417746, + "grad_norm": 2.2925968170166016, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.7100216150283813, + "num_tokens": 187304621.0, + "step": 7362 + }, + { + "epoch": 0.8085877443443883, + "grad_norm": 2.200939416885376, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7071493864059448, + "num_tokens": 187329671.0, + "step": 7363 + }, + { + "epoch": 0.8086975620470019, + "grad_norm": 2.281651020050049, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7063542008399963, + "num_tokens": 187353435.0, + "step": 7364 + }, + { + "epoch": 0.8088073797496156, + "grad_norm": 2.2153120040893555, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.721416711807251, + "num_tokens": 187377225.0, + "step": 7365 + }, + { + "epoch": 0.8089171974522293, + "grad_norm": 1.9761383533477783, + "learning_rate": 1e-06, + "loss": 1.0195, + "mean_token_accuracy": 0.6895087361335754, + "num_tokens": 187408523.0, + "step": 7366 + }, + { + "epoch": 0.809027015154843, + "grad_norm": 2.3718044757843018, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7231494784355164, + "num_tokens": 187430205.0, + "step": 7367 + }, + { + "epoch": 0.8091368328574566, + "grad_norm": 2.417229413986206, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7042542695999146, + "num_tokens": 187452492.0, + "step": 7368 + }, + { + "epoch": 0.8092466505600703, + "grad_norm": 1.997999906539917, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.7000408172607422, + "num_tokens": 187482614.0, + "step": 7369 + }, + { + "epoch": 0.8093564682626839, + "grad_norm": 2.3280670642852783, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.710407018661499, + "num_tokens": 187507719.0, + "step": 7370 + }, + { + "epoch": 0.8094662859652976, + "grad_norm": 2.161121129989624, + "learning_rate": 1e-06, + "loss": 1.0411, + "mean_token_accuracy": 0.6910210847854614, + "num_tokens": 187535747.0, + "step": 7371 + }, + { + "epoch": 0.8095761036679112, + "grad_norm": 2.1990232467651367, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7235434055328369, + "num_tokens": 187561076.0, + "step": 7372 + }, + { + "epoch": 0.809685921370525, + "grad_norm": 2.151597499847412, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6943230628967285, + "num_tokens": 187588135.0, + "step": 7373 + }, + { + "epoch": 0.8097957390731386, + "grad_norm": 2.3888847827911377, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.708139181137085, + "num_tokens": 187610319.0, + "step": 7374 + }, + { + "epoch": 0.8099055567757523, + "grad_norm": 2.032796859741211, + "learning_rate": 1e-06, + "loss": 1.0445, + "mean_token_accuracy": 0.6953616142272949, + "num_tokens": 187638999.0, + "step": 7375 + }, + { + "epoch": 0.8100153744783659, + "grad_norm": 2.1807472705841064, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7115809917449951, + "num_tokens": 187663953.0, + "step": 7376 + }, + { + "epoch": 0.8101251921809796, + "grad_norm": 2.0520598888397217, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.7072163224220276, + "num_tokens": 187690562.0, + "step": 7377 + }, + { + "epoch": 0.8102350098835932, + "grad_norm": 2.2374231815338135, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.709815263748169, + "num_tokens": 187714248.0, + "step": 7378 + }, + { + "epoch": 0.8103448275862069, + "grad_norm": 2.245570659637451, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.6929219365119934, + "num_tokens": 187738932.0, + "step": 7379 + }, + { + "epoch": 0.8104546452888206, + "grad_norm": 2.233506441116333, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6963874101638794, + "num_tokens": 187762532.0, + "step": 7380 + }, + { + "epoch": 0.8105644629914343, + "grad_norm": 2.137529134750366, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6933251619338989, + "num_tokens": 187788621.0, + "step": 7381 + }, + { + "epoch": 0.8106742806940479, + "grad_norm": 2.2212603092193604, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6960697174072266, + "num_tokens": 187812936.0, + "step": 7382 + }, + { + "epoch": 0.8107840983966615, + "grad_norm": 2.507290840148926, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7179462909698486, + "num_tokens": 187832700.0, + "step": 7383 + }, + { + "epoch": 0.8108939160992752, + "grad_norm": 2.1265056133270264, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7079620957374573, + "num_tokens": 187858849.0, + "step": 7384 + }, + { + "epoch": 0.8110037338018888, + "grad_norm": 2.2383177280426025, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7446265816688538, + "num_tokens": 187883022.0, + "step": 7385 + }, + { + "epoch": 0.8111135515045025, + "grad_norm": 2.198944568634033, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7149344086647034, + "num_tokens": 187907174.0, + "step": 7386 + }, + { + "epoch": 0.8112233692071162, + "grad_norm": 2.2233219146728516, + "learning_rate": 1e-06, + "loss": 1.0605, + "mean_token_accuracy": 0.6882307529449463, + "num_tokens": 187933554.0, + "step": 7387 + }, + { + "epoch": 0.8113331869097299, + "grad_norm": 2.082190990447998, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7121955752372742, + "num_tokens": 187959955.0, + "step": 7388 + }, + { + "epoch": 0.8114430046123435, + "grad_norm": 2.0322868824005127, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7073596715927124, + "num_tokens": 187990304.0, + "step": 7389 + }, + { + "epoch": 0.8115528223149572, + "grad_norm": 2.1394262313842773, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6962546110153198, + "num_tokens": 188016526.0, + "step": 7390 + }, + { + "epoch": 0.8116626400175708, + "grad_norm": 2.3089210987091064, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7175604701042175, + "num_tokens": 188038763.0, + "step": 7391 + }, + { + "epoch": 0.8117724577201845, + "grad_norm": 2.0650527477264404, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.7045209407806396, + "num_tokens": 188066024.0, + "step": 7392 + }, + { + "epoch": 0.8118822754227981, + "grad_norm": 2.212468147277832, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7139819860458374, + "num_tokens": 188089968.0, + "step": 7393 + }, + { + "epoch": 0.8119920931254118, + "grad_norm": 2.533712148666382, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7195507287979126, + "num_tokens": 188109089.0, + "step": 7394 + }, + { + "epoch": 0.8121019108280255, + "grad_norm": 2.0755667686462402, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.7081714272499084, + "num_tokens": 188136390.0, + "step": 7395 + }, + { + "epoch": 0.8122117285306392, + "grad_norm": 2.2315995693206787, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7122654914855957, + "num_tokens": 188158916.0, + "step": 7396 + }, + { + "epoch": 0.8123215462332528, + "grad_norm": 2.253007411956787, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7142239809036255, + "num_tokens": 188182432.0, + "step": 7397 + }, + { + "epoch": 0.8124313639358665, + "grad_norm": 2.0789248943328857, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.703770101070404, + "num_tokens": 188209176.0, + "step": 7398 + }, + { + "epoch": 0.8125411816384801, + "grad_norm": 2.416624069213867, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.730012059211731, + "num_tokens": 188229704.0, + "step": 7399 + }, + { + "epoch": 0.8126509993410937, + "grad_norm": 2.4950850009918213, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6882208585739136, + "num_tokens": 188250920.0, + "step": 7400 + }, + { + "epoch": 0.8127608170437074, + "grad_norm": 2.4289865493774414, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7270747423171997, + "num_tokens": 188270943.0, + "step": 7401 + }, + { + "epoch": 0.8128706347463212, + "grad_norm": 2.2505722045898438, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7125178575515747, + "num_tokens": 188295250.0, + "step": 7402 + }, + { + "epoch": 0.8129804524489348, + "grad_norm": 2.109738349914551, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7118343114852905, + "num_tokens": 188321936.0, + "step": 7403 + }, + { + "epoch": 0.8130902701515484, + "grad_norm": 2.254183053970337, + "learning_rate": 1e-06, + "loss": 1.0921, + "mean_token_accuracy": 0.684394359588623, + "num_tokens": 188346861.0, + "step": 7404 + }, + { + "epoch": 0.8132000878541621, + "grad_norm": 1.955628752708435, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7110502123832703, + "num_tokens": 188377257.0, + "step": 7405 + }, + { + "epoch": 0.8133099055567757, + "grad_norm": 2.02866530418396, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7095524072647095, + "num_tokens": 188404396.0, + "step": 7406 + }, + { + "epoch": 0.8134197232593894, + "grad_norm": 1.9884917736053467, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7029095888137817, + "num_tokens": 188436423.0, + "step": 7407 + }, + { + "epoch": 0.813529540962003, + "grad_norm": 2.276846408843994, + "learning_rate": 1e-06, + "loss": 1.0514, + "mean_token_accuracy": 0.6839638948440552, + "num_tokens": 188461581.0, + "step": 7408 + }, + { + "epoch": 0.8136393586646168, + "grad_norm": 2.1911885738372803, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7049964666366577, + "num_tokens": 188486980.0, + "step": 7409 + }, + { + "epoch": 0.8137491763672304, + "grad_norm": 2.13547420501709, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7225303053855896, + "num_tokens": 188512913.0, + "step": 7410 + }, + { + "epoch": 0.8138589940698441, + "grad_norm": 2.4782066345214844, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7121193408966064, + "num_tokens": 188533329.0, + "step": 7411 + }, + { + "epoch": 0.8139688117724577, + "grad_norm": 2.2679262161254883, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7264770865440369, + "num_tokens": 188556801.0, + "step": 7412 + }, + { + "epoch": 0.8140786294750714, + "grad_norm": 2.265587329864502, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7027239203453064, + "num_tokens": 188582343.0, + "step": 7413 + }, + { + "epoch": 0.814188447177685, + "grad_norm": 2.1428515911102295, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7255810499191284, + "num_tokens": 188607793.0, + "step": 7414 + }, + { + "epoch": 0.8142982648802987, + "grad_norm": 2.338031053543091, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.703737735748291, + "num_tokens": 188631063.0, + "step": 7415 + }, + { + "epoch": 0.8144080825829124, + "grad_norm": 2.2410483360290527, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7144789695739746, + "num_tokens": 188654221.0, + "step": 7416 + }, + { + "epoch": 0.8145179002855261, + "grad_norm": 2.524913787841797, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.697801947593689, + "num_tokens": 188675228.0, + "step": 7417 + }, + { + "epoch": 0.8146277179881397, + "grad_norm": 2.601364850997925, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7065671682357788, + "num_tokens": 188695603.0, + "step": 7418 + }, + { + "epoch": 0.8147375356907534, + "grad_norm": 2.1929757595062256, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7405979037284851, + "num_tokens": 188719058.0, + "step": 7419 + }, + { + "epoch": 0.814847353393367, + "grad_norm": 1.9850010871887207, + "learning_rate": 1e-06, + "loss": 1.0075, + "mean_token_accuracy": 0.6866827011108398, + "num_tokens": 188750334.0, + "step": 7420 + }, + { + "epoch": 0.8149571710959806, + "grad_norm": 2.3636789321899414, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7371004819869995, + "num_tokens": 188772098.0, + "step": 7421 + }, + { + "epoch": 0.8150669887985943, + "grad_norm": 2.13102388381958, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6874440908432007, + "num_tokens": 188799488.0, + "step": 7422 + }, + { + "epoch": 0.8151768065012079, + "grad_norm": 2.053739070892334, + "learning_rate": 1e-06, + "loss": 1.0588, + "mean_token_accuracy": 0.6787939071655273, + "num_tokens": 188829405.0, + "step": 7423 + }, + { + "epoch": 0.8152866242038217, + "grad_norm": 2.541712522506714, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7245588898658752, + "num_tokens": 188847716.0, + "step": 7424 + }, + { + "epoch": 0.8153964419064353, + "grad_norm": 1.909825086593628, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6878820061683655, + "num_tokens": 188879010.0, + "step": 7425 + }, + { + "epoch": 0.815506259609049, + "grad_norm": 2.5358152389526367, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7283342480659485, + "num_tokens": 188897923.0, + "step": 7426 + }, + { + "epoch": 0.8156160773116626, + "grad_norm": 2.2734932899475098, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7219178676605225, + "num_tokens": 188922592.0, + "step": 7427 + }, + { + "epoch": 0.8157258950142763, + "grad_norm": 2.0217092037200928, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.6968348026275635, + "num_tokens": 188954798.0, + "step": 7428 + }, + { + "epoch": 0.8158357127168899, + "grad_norm": 2.507737159729004, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7198415994644165, + "num_tokens": 188973879.0, + "step": 7429 + }, + { + "epoch": 0.8159455304195036, + "grad_norm": 2.2084739208221436, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6930999755859375, + "num_tokens": 188998821.0, + "step": 7430 + }, + { + "epoch": 0.8160553481221173, + "grad_norm": 2.1345248222351074, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7117010354995728, + "num_tokens": 189025615.0, + "step": 7431 + }, + { + "epoch": 0.816165165824731, + "grad_norm": 2.156055450439453, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7100269794464111, + "num_tokens": 189050941.0, + "step": 7432 + }, + { + "epoch": 0.8162749835273446, + "grad_norm": 2.386046886444092, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7133884429931641, + "num_tokens": 189072715.0, + "step": 7433 + }, + { + "epoch": 0.8163848012299583, + "grad_norm": 2.0737361907958984, + "learning_rate": 1e-06, + "loss": 1.0537, + "mean_token_accuracy": 0.6835391521453857, + "num_tokens": 189100347.0, + "step": 7434 + }, + { + "epoch": 0.8164946189325719, + "grad_norm": 2.035055160522461, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7126050591468811, + "num_tokens": 189130715.0, + "step": 7435 + }, + { + "epoch": 0.8166044366351856, + "grad_norm": 1.934110403060913, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.695554256439209, + "num_tokens": 189163099.0, + "step": 7436 + }, + { + "epoch": 0.8167142543377992, + "grad_norm": 2.115314245223999, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7066977024078369, + "num_tokens": 189187354.0, + "step": 7437 + }, + { + "epoch": 0.816824072040413, + "grad_norm": 2.0471527576446533, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6925809383392334, + "num_tokens": 189217300.0, + "step": 7438 + }, + { + "epoch": 0.8169338897430266, + "grad_norm": 2.3690972328186035, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7237585186958313, + "num_tokens": 189238030.0, + "step": 7439 + }, + { + "epoch": 0.8170437074456403, + "grad_norm": 2.215888738632202, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7113137245178223, + "num_tokens": 189263928.0, + "step": 7440 + }, + { + "epoch": 0.8171535251482539, + "grad_norm": 1.9373921155929565, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6971200704574585, + "num_tokens": 189295526.0, + "step": 7441 + }, + { + "epoch": 0.8172633428508675, + "grad_norm": 2.130664348602295, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7270885109901428, + "num_tokens": 189321433.0, + "step": 7442 + }, + { + "epoch": 0.8173731605534812, + "grad_norm": 2.3607239723205566, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7072332501411438, + "num_tokens": 189342970.0, + "step": 7443 + }, + { + "epoch": 0.8174829782560948, + "grad_norm": 2.5334079265594482, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7121995687484741, + "num_tokens": 189363916.0, + "step": 7444 + }, + { + "epoch": 0.8175927959587086, + "grad_norm": 2.042041301727295, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.6980860829353333, + "num_tokens": 189393066.0, + "step": 7445 + }, + { + "epoch": 0.8177026136613222, + "grad_norm": 2.271885395050049, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7085587382316589, + "num_tokens": 189417183.0, + "step": 7446 + }, + { + "epoch": 0.8178124313639359, + "grad_norm": 2.1537814140319824, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.732864499092102, + "num_tokens": 189443346.0, + "step": 7447 + }, + { + "epoch": 0.8179222490665495, + "grad_norm": 2.228896379470825, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6827012300491333, + "num_tokens": 189468395.0, + "step": 7448 + }, + { + "epoch": 0.8180320667691632, + "grad_norm": 2.0779662132263184, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7068345546722412, + "num_tokens": 189497783.0, + "step": 7449 + }, + { + "epoch": 0.8181418844717768, + "grad_norm": 2.388671398162842, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6906701326370239, + "num_tokens": 189521642.0, + "step": 7450 + }, + { + "epoch": 0.8182517021743905, + "grad_norm": 2.1802120208740234, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7270165085792542, + "num_tokens": 189546106.0, + "step": 7451 + }, + { + "epoch": 0.8183615198770041, + "grad_norm": 1.9233999252319336, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6962918043136597, + "num_tokens": 189577496.0, + "step": 7452 + }, + { + "epoch": 0.8184713375796179, + "grad_norm": 2.255472421646118, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6973143219947815, + "num_tokens": 189604139.0, + "step": 7453 + }, + { + "epoch": 0.8185811552822315, + "grad_norm": 2.2818381786346436, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7353223562240601, + "num_tokens": 189627726.0, + "step": 7454 + }, + { + "epoch": 0.8186909729848452, + "grad_norm": 2.356818914413452, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7000536918640137, + "num_tokens": 189649773.0, + "step": 7455 + }, + { + "epoch": 0.8188007906874588, + "grad_norm": 2.008279800415039, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7071279287338257, + "num_tokens": 189677459.0, + "step": 7456 + }, + { + "epoch": 0.8189106083900725, + "grad_norm": 2.0403878688812256, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6821736693382263, + "num_tokens": 189705278.0, + "step": 7457 + }, + { + "epoch": 0.8190204260926861, + "grad_norm": 2.2225654125213623, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.7072988152503967, + "num_tokens": 189731877.0, + "step": 7458 + }, + { + "epoch": 0.8191302437952998, + "grad_norm": 2.0067830085754395, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7134053707122803, + "num_tokens": 189759849.0, + "step": 7459 + }, + { + "epoch": 0.8192400614979135, + "grad_norm": 2.2579193115234375, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7215912938117981, + "num_tokens": 189781185.0, + "step": 7460 + }, + { + "epoch": 0.8193498792005272, + "grad_norm": 2.1751255989074707, + "learning_rate": 1e-06, + "loss": 1.0487, + "mean_token_accuracy": 0.6893132925033569, + "num_tokens": 189805924.0, + "step": 7461 + }, + { + "epoch": 0.8194596969031408, + "grad_norm": 2.2598602771759033, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6886568665504456, + "num_tokens": 189831124.0, + "step": 7462 + }, + { + "epoch": 0.8195695146057544, + "grad_norm": 1.9850616455078125, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7002527713775635, + "num_tokens": 189860171.0, + "step": 7463 + }, + { + "epoch": 0.8196793323083681, + "grad_norm": 2.2627999782562256, + "learning_rate": 1e-06, + "loss": 1.0703, + "mean_token_accuracy": 0.6745656728744507, + "num_tokens": 189886312.0, + "step": 7464 + }, + { + "epoch": 0.8197891500109817, + "grad_norm": 2.182628870010376, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7116990089416504, + "num_tokens": 189909099.0, + "step": 7465 + }, + { + "epoch": 0.8198989677135954, + "grad_norm": 2.198906898498535, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.70729660987854, + "num_tokens": 189932343.0, + "step": 7466 + }, + { + "epoch": 0.8200087854162091, + "grad_norm": 2.3789873123168945, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7097931504249573, + "num_tokens": 189954660.0, + "step": 7467 + }, + { + "epoch": 0.8201186031188228, + "grad_norm": 2.2162282466888428, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7175556421279907, + "num_tokens": 189978799.0, + "step": 7468 + }, + { + "epoch": 0.8202284208214364, + "grad_norm": 2.120229721069336, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6943193674087524, + "num_tokens": 190007981.0, + "step": 7469 + }, + { + "epoch": 0.8203382385240501, + "grad_norm": 2.0695366859436035, + "learning_rate": 1e-06, + "loss": 0.9742, + "mean_token_accuracy": 0.6960994005203247, + "num_tokens": 190036468.0, + "step": 7470 + }, + { + "epoch": 0.8204480562266637, + "grad_norm": 2.36631441116333, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7133547067642212, + "num_tokens": 190059568.0, + "step": 7471 + }, + { + "epoch": 0.8205578739292774, + "grad_norm": 2.2031826972961426, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7213391065597534, + "num_tokens": 190085113.0, + "step": 7472 + }, + { + "epoch": 0.820667691631891, + "grad_norm": 2.263192653656006, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7202019691467285, + "num_tokens": 190108253.0, + "step": 7473 + }, + { + "epoch": 0.8207775093345048, + "grad_norm": 2.349759340286255, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7065732479095459, + "num_tokens": 190130469.0, + "step": 7474 + }, + { + "epoch": 0.8208873270371184, + "grad_norm": 2.4704010486602783, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7333438396453857, + "num_tokens": 190151246.0, + "step": 7475 + }, + { + "epoch": 0.8209971447397321, + "grad_norm": 2.1584620475769043, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7296934127807617, + "num_tokens": 190176230.0, + "step": 7476 + }, + { + "epoch": 0.8211069624423457, + "grad_norm": 2.238403081893921, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.7033636569976807, + "num_tokens": 190201087.0, + "step": 7477 + }, + { + "epoch": 0.8212167801449594, + "grad_norm": 2.3621068000793457, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6991897225379944, + "num_tokens": 190224006.0, + "step": 7478 + }, + { + "epoch": 0.821326597847573, + "grad_norm": 2.0906107425689697, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7022927403450012, + "num_tokens": 190251990.0, + "step": 7479 + }, + { + "epoch": 0.8214364155501866, + "grad_norm": 2.0547759532928467, + "learning_rate": 1e-06, + "loss": 1.0417, + "mean_token_accuracy": 0.6844574809074402, + "num_tokens": 190281709.0, + "step": 7480 + }, + { + "epoch": 0.8215462332528004, + "grad_norm": 1.9786714315414429, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6987522840499878, + "num_tokens": 190311071.0, + "step": 7481 + }, + { + "epoch": 0.821656050955414, + "grad_norm": 2.0755512714385986, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7097426652908325, + "num_tokens": 190338481.0, + "step": 7482 + }, + { + "epoch": 0.8217658686580277, + "grad_norm": 2.1200621128082275, + "learning_rate": 1e-06, + "loss": 1.0526, + "mean_token_accuracy": 0.6807562708854675, + "num_tokens": 190368091.0, + "step": 7483 + }, + { + "epoch": 0.8218756863606413, + "grad_norm": 2.169928789138794, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7021427750587463, + "num_tokens": 190393772.0, + "step": 7484 + }, + { + "epoch": 0.821985504063255, + "grad_norm": 2.1204099655151367, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.707331120967865, + "num_tokens": 190420671.0, + "step": 7485 + }, + { + "epoch": 0.8220953217658686, + "grad_norm": 2.1038973331451416, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6946507096290588, + "num_tokens": 190446287.0, + "step": 7486 + }, + { + "epoch": 0.8222051394684823, + "grad_norm": 2.508178234100342, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7189028263092041, + "num_tokens": 190466745.0, + "step": 7487 + }, + { + "epoch": 0.8223149571710959, + "grad_norm": 2.331895351409912, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7343559265136719, + "num_tokens": 190489944.0, + "step": 7488 + }, + { + "epoch": 0.8224247748737097, + "grad_norm": 2.2963335514068604, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.7006467580795288, + "num_tokens": 190514978.0, + "step": 7489 + }, + { + "epoch": 0.8225345925763233, + "grad_norm": 1.933911681175232, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.6894906759262085, + "num_tokens": 190545747.0, + "step": 7490 + }, + { + "epoch": 0.822644410278937, + "grad_norm": 2.0325887203216553, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7227448225021362, + "num_tokens": 190574374.0, + "step": 7491 + }, + { + "epoch": 0.8227542279815506, + "grad_norm": 2.2527902126312256, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.7011210322380066, + "num_tokens": 190597896.0, + "step": 7492 + }, + { + "epoch": 0.8228640456841643, + "grad_norm": 2.0876424312591553, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.6959363222122192, + "num_tokens": 190627296.0, + "step": 7493 + }, + { + "epoch": 0.8229738633867779, + "grad_norm": 2.2203359603881836, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7044758200645447, + "num_tokens": 190650519.0, + "step": 7494 + }, + { + "epoch": 0.8230836810893916, + "grad_norm": 2.2784669399261475, + "learning_rate": 1e-06, + "loss": 1.0515, + "mean_token_accuracy": 0.6873669028282166, + "num_tokens": 190675969.0, + "step": 7495 + }, + { + "epoch": 0.8231934987920053, + "grad_norm": 2.1605312824249268, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.7005298733711243, + "num_tokens": 190701293.0, + "step": 7496 + }, + { + "epoch": 0.823303316494619, + "grad_norm": 2.287320852279663, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7087774276733398, + "num_tokens": 190723152.0, + "step": 7497 + }, + { + "epoch": 0.8234131341972326, + "grad_norm": 2.1834158897399902, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7138557434082031, + "num_tokens": 190748933.0, + "step": 7498 + }, + { + "epoch": 0.8235229518998463, + "grad_norm": 2.257875919342041, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7249647974967957, + "num_tokens": 190770006.0, + "step": 7499 + }, + { + "epoch": 0.8236327696024599, + "grad_norm": 2.083869218826294, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7096331715583801, + "num_tokens": 190796827.0, + "step": 7500 + }, + { + "epoch": 0.8237425873050735, + "grad_norm": 2.2711844444274902, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7220703959465027, + "num_tokens": 190818975.0, + "step": 7501 + }, + { + "epoch": 0.8238524050076872, + "grad_norm": 2.3672940731048584, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7302197217941284, + "num_tokens": 190839363.0, + "step": 7502 + }, + { + "epoch": 0.823962222710301, + "grad_norm": 2.1875765323638916, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.703532874584198, + "num_tokens": 190864277.0, + "step": 7503 + }, + { + "epoch": 0.8240720404129146, + "grad_norm": 2.5229012966156006, + "learning_rate": 1e-06, + "loss": 0.8189, + "mean_token_accuracy": 0.7432746291160583, + "num_tokens": 190883311.0, + "step": 7504 + }, + { + "epoch": 0.8241818581155282, + "grad_norm": 2.2725608348846436, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7018089890480042, + "num_tokens": 190905934.0, + "step": 7505 + }, + { + "epoch": 0.8242916758181419, + "grad_norm": 2.130728244781494, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7267762422561646, + "num_tokens": 190929925.0, + "step": 7506 + }, + { + "epoch": 0.8244014935207555, + "grad_norm": 2.46197247505188, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.716339111328125, + "num_tokens": 190950243.0, + "step": 7507 + }, + { + "epoch": 0.8245113112233692, + "grad_norm": 2.0668649673461914, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7319818139076233, + "num_tokens": 190977274.0, + "step": 7508 + }, + { + "epoch": 0.8246211289259828, + "grad_norm": 2.262507438659668, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7080364227294922, + "num_tokens": 191002578.0, + "step": 7509 + }, + { + "epoch": 0.8247309466285966, + "grad_norm": 2.291058301925659, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7064052820205688, + "num_tokens": 191026047.0, + "step": 7510 + }, + { + "epoch": 0.8248407643312102, + "grad_norm": 2.121819496154785, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.6937307119369507, + "num_tokens": 191053705.0, + "step": 7511 + }, + { + "epoch": 0.8249505820338239, + "grad_norm": 2.242889404296875, + "learning_rate": 1e-06, + "loss": 1.0067, + "mean_token_accuracy": 0.6938250064849854, + "num_tokens": 191078300.0, + "step": 7512 + }, + { + "epoch": 0.8250603997364375, + "grad_norm": 2.5131638050079346, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7222665548324585, + "num_tokens": 191097637.0, + "step": 7513 + }, + { + "epoch": 0.8251702174390512, + "grad_norm": 2.129733085632324, + "learning_rate": 1e-06, + "loss": 1.0761, + "mean_token_accuracy": 0.6728371977806091, + "num_tokens": 191125788.0, + "step": 7514 + }, + { + "epoch": 0.8252800351416648, + "grad_norm": 2.5242700576782227, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7074392437934875, + "num_tokens": 191145726.0, + "step": 7515 + }, + { + "epoch": 0.8253898528442785, + "grad_norm": 2.166821002960205, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7098534107208252, + "num_tokens": 191174540.0, + "step": 7516 + }, + { + "epoch": 0.8254996705468921, + "grad_norm": 2.0336241722106934, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7070208191871643, + "num_tokens": 191202271.0, + "step": 7517 + }, + { + "epoch": 0.8256094882495059, + "grad_norm": 2.171281576156616, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6950845122337341, + "num_tokens": 191227219.0, + "step": 7518 + }, + { + "epoch": 0.8257193059521195, + "grad_norm": 2.136667013168335, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.693005383014679, + "num_tokens": 191254144.0, + "step": 7519 + }, + { + "epoch": 0.8258291236547332, + "grad_norm": 2.2283003330230713, + "learning_rate": 1e-06, + "loss": 1.0361, + "mean_token_accuracy": 0.6940662860870361, + "num_tokens": 191278080.0, + "step": 7520 + }, + { + "epoch": 0.8259389413573468, + "grad_norm": 2.14936900138855, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7040719985961914, + "num_tokens": 191304408.0, + "step": 7521 + }, + { + "epoch": 0.8260487590599604, + "grad_norm": 2.2886269092559814, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6941726207733154, + "num_tokens": 191328458.0, + "step": 7522 + }, + { + "epoch": 0.8261585767625741, + "grad_norm": 2.287810802459717, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7084611058235168, + "num_tokens": 191351920.0, + "step": 7523 + }, + { + "epoch": 0.8262683944651877, + "grad_norm": 2.2237889766693115, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.729780912399292, + "num_tokens": 191375226.0, + "step": 7524 + }, + { + "epoch": 0.8263782121678015, + "grad_norm": 2.3655571937561035, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7292641997337341, + "num_tokens": 191396579.0, + "step": 7525 + }, + { + "epoch": 0.8264880298704151, + "grad_norm": 2.1906182765960693, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7090637683868408, + "num_tokens": 191422326.0, + "step": 7526 + }, + { + "epoch": 0.8265978475730288, + "grad_norm": 2.1417040824890137, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.704828143119812, + "num_tokens": 191449531.0, + "step": 7527 + }, + { + "epoch": 0.8267076652756424, + "grad_norm": 2.3354249000549316, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.73126220703125, + "num_tokens": 191474558.0, + "step": 7528 + }, + { + "epoch": 0.8268174829782561, + "grad_norm": 1.865347981452942, + "learning_rate": 1e-06, + "loss": 1.0623, + "mean_token_accuracy": 0.6798557043075562, + "num_tokens": 191510123.0, + "step": 7529 + }, + { + "epoch": 0.8269273006808697, + "grad_norm": 2.2502620220184326, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.698928713798523, + "num_tokens": 191533520.0, + "step": 7530 + }, + { + "epoch": 0.8270371183834834, + "grad_norm": 1.9538910388946533, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7059415578842163, + "num_tokens": 191564930.0, + "step": 7531 + }, + { + "epoch": 0.8271469360860971, + "grad_norm": 2.0694425106048584, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7170521020889282, + "num_tokens": 191591709.0, + "step": 7532 + }, + { + "epoch": 0.8272567537887108, + "grad_norm": 2.1621901988983154, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6951462030410767, + "num_tokens": 191618204.0, + "step": 7533 + }, + { + "epoch": 0.8273665714913244, + "grad_norm": 2.1632320880889893, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6977782845497131, + "num_tokens": 191643550.0, + "step": 7534 + }, + { + "epoch": 0.8274763891939381, + "grad_norm": 2.6769492626190186, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7208555936813354, + "num_tokens": 191661720.0, + "step": 7535 + }, + { + "epoch": 0.8275862068965517, + "grad_norm": 2.1030542850494385, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7166603803634644, + "num_tokens": 191686921.0, + "step": 7536 + }, + { + "epoch": 0.8276960245991654, + "grad_norm": 2.694537878036499, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7310735583305359, + "num_tokens": 191704562.0, + "step": 7537 + }, + { + "epoch": 0.827805842301779, + "grad_norm": 2.1036622524261475, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7139729261398315, + "num_tokens": 191735042.0, + "step": 7538 + }, + { + "epoch": 0.8279156600043928, + "grad_norm": 2.0328667163848877, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7190288305282593, + "num_tokens": 191761988.0, + "step": 7539 + }, + { + "epoch": 0.8280254777070064, + "grad_norm": 2.0890626907348633, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7031367421150208, + "num_tokens": 191788400.0, + "step": 7540 + }, + { + "epoch": 0.82813529540962, + "grad_norm": 2.5471181869506836, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7188608050346375, + "num_tokens": 191807401.0, + "step": 7541 + }, + { + "epoch": 0.8282451131122337, + "grad_norm": 2.0944063663482666, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7218092679977417, + "num_tokens": 191832383.0, + "step": 7542 + }, + { + "epoch": 0.8283549308148473, + "grad_norm": 1.9089239835739136, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7172937989234924, + "num_tokens": 191863822.0, + "step": 7543 + }, + { + "epoch": 0.828464748517461, + "grad_norm": 2.289724349975586, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7057995200157166, + "num_tokens": 191888200.0, + "step": 7544 + }, + { + "epoch": 0.8285745662200746, + "grad_norm": 2.2244105339050293, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7103964686393738, + "num_tokens": 191912133.0, + "step": 7545 + }, + { + "epoch": 0.8286843839226883, + "grad_norm": 2.2011208534240723, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6848310232162476, + "num_tokens": 191937990.0, + "step": 7546 + }, + { + "epoch": 0.828794201625302, + "grad_norm": 1.9620778560638428, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7209383845329285, + "num_tokens": 191967493.0, + "step": 7547 + }, + { + "epoch": 0.8289040193279157, + "grad_norm": 2.2633252143859863, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.6956602931022644, + "num_tokens": 191992361.0, + "step": 7548 + }, + { + "epoch": 0.8290138370305293, + "grad_norm": 1.9771475791931152, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.6984202265739441, + "num_tokens": 192022641.0, + "step": 7549 + }, + { + "epoch": 0.829123654733143, + "grad_norm": 2.3483214378356934, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.7016603350639343, + "num_tokens": 192044855.0, + "step": 7550 + }, + { + "epoch": 0.8292334724357566, + "grad_norm": 2.28955078125, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7141350507736206, + "num_tokens": 192068986.0, + "step": 7551 + }, + { + "epoch": 0.8293432901383703, + "grad_norm": 2.2614288330078125, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7068697214126587, + "num_tokens": 192093206.0, + "step": 7552 + }, + { + "epoch": 0.8294531078409839, + "grad_norm": 2.0715279579162598, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7370082139968872, + "num_tokens": 192120957.0, + "step": 7553 + }, + { + "epoch": 0.8295629255435977, + "grad_norm": 1.9894018173217773, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7215563654899597, + "num_tokens": 192151735.0, + "step": 7554 + }, + { + "epoch": 0.8296727432462113, + "grad_norm": 2.004957914352417, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7003034949302673, + "num_tokens": 192182188.0, + "step": 7555 + }, + { + "epoch": 0.829782560948825, + "grad_norm": 2.315187931060791, + "learning_rate": 1e-06, + "loss": 0.8002, + "mean_token_accuracy": 0.74875408411026, + "num_tokens": 192203687.0, + "step": 7556 + }, + { + "epoch": 0.8298923786514386, + "grad_norm": 1.9326882362365723, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6953445672988892, + "num_tokens": 192236313.0, + "step": 7557 + }, + { + "epoch": 0.8300021963540523, + "grad_norm": 2.5111382007598877, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7258999347686768, + "num_tokens": 192255249.0, + "step": 7558 + }, + { + "epoch": 0.8301120140566659, + "grad_norm": 2.092451810836792, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7006181478500366, + "num_tokens": 192283013.0, + "step": 7559 + }, + { + "epoch": 0.8302218317592795, + "grad_norm": 2.0728020668029785, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7095205783843994, + "num_tokens": 192312916.0, + "step": 7560 + }, + { + "epoch": 0.8303316494618933, + "grad_norm": 2.3603265285491943, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7138055562973022, + "num_tokens": 192336022.0, + "step": 7561 + }, + { + "epoch": 0.830441467164507, + "grad_norm": 2.372236967086792, + "learning_rate": 1e-06, + "loss": 1.0715, + "mean_token_accuracy": 0.6773133277893066, + "num_tokens": 192359888.0, + "step": 7562 + }, + { + "epoch": 0.8305512848671206, + "grad_norm": 2.1000401973724365, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6882362961769104, + "num_tokens": 192388313.0, + "step": 7563 + }, + { + "epoch": 0.8306611025697342, + "grad_norm": 2.230977773666382, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7081491947174072, + "num_tokens": 192411162.0, + "step": 7564 + }, + { + "epoch": 0.8307709202723479, + "grad_norm": 2.1546502113342285, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7183303833007812, + "num_tokens": 192436726.0, + "step": 7565 + }, + { + "epoch": 0.8308807379749615, + "grad_norm": 2.2591395378112793, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7087112069129944, + "num_tokens": 192460497.0, + "step": 7566 + }, + { + "epoch": 0.8309905556775752, + "grad_norm": 1.9145797491073608, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6892605423927307, + "num_tokens": 192493405.0, + "step": 7567 + }, + { + "epoch": 0.8311003733801889, + "grad_norm": 2.260833501815796, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7442779541015625, + "num_tokens": 192516337.0, + "step": 7568 + }, + { + "epoch": 0.8312101910828026, + "grad_norm": 2.1242940425872803, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6896505355834961, + "num_tokens": 192543349.0, + "step": 7569 + }, + { + "epoch": 0.8313200087854162, + "grad_norm": 2.6403441429138184, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7150768637657166, + "num_tokens": 192560421.0, + "step": 7570 + }, + { + "epoch": 0.8314298264880299, + "grad_norm": 2.0532851219177246, + "learning_rate": 1e-06, + "loss": 1.0461, + "mean_token_accuracy": 0.6903420686721802, + "num_tokens": 192589935.0, + "step": 7571 + }, + { + "epoch": 0.8315396441906435, + "grad_norm": 2.2164294719696045, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6911225914955139, + "num_tokens": 192614863.0, + "step": 7572 + }, + { + "epoch": 0.8316494618932572, + "grad_norm": 2.338618755340576, + "learning_rate": 1e-06, + "loss": 1.0564, + "mean_token_accuracy": 0.691902220249176, + "num_tokens": 192639426.0, + "step": 7573 + }, + { + "epoch": 0.8317592795958708, + "grad_norm": 2.257110834121704, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7177639603614807, + "num_tokens": 192664265.0, + "step": 7574 + }, + { + "epoch": 0.8318690972984845, + "grad_norm": 2.1727137565612793, + "learning_rate": 1e-06, + "loss": 1.0263, + "mean_token_accuracy": 0.683899462223053, + "num_tokens": 192691615.0, + "step": 7575 + }, + { + "epoch": 0.8319789150010982, + "grad_norm": 2.2042112350463867, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.716600775718689, + "num_tokens": 192718160.0, + "step": 7576 + }, + { + "epoch": 0.8320887327037119, + "grad_norm": 2.211763858795166, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6975380182266235, + "num_tokens": 192742270.0, + "step": 7577 + }, + { + "epoch": 0.8321985504063255, + "grad_norm": 2.1595704555511475, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.7035250663757324, + "num_tokens": 192768626.0, + "step": 7578 + }, + { + "epoch": 0.8323083681089392, + "grad_norm": 2.142320394515991, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7131312489509583, + "num_tokens": 192795018.0, + "step": 7579 + }, + { + "epoch": 0.8324181858115528, + "grad_norm": 2.090893507003784, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7151509523391724, + "num_tokens": 192820125.0, + "step": 7580 + }, + { + "epoch": 0.8325280035141664, + "grad_norm": 2.409961462020874, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7093901038169861, + "num_tokens": 192841745.0, + "step": 7581 + }, + { + "epoch": 0.8326378212167801, + "grad_norm": 2.202458381652832, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.7017509937286377, + "num_tokens": 192867284.0, + "step": 7582 + }, + { + "epoch": 0.8327476389193939, + "grad_norm": 2.3463170528411865, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7277507781982422, + "num_tokens": 192889165.0, + "step": 7583 + }, + { + "epoch": 0.8328574566220075, + "grad_norm": 1.9901074171066284, + "learning_rate": 1e-06, + "loss": 1.1045, + "mean_token_accuracy": 0.6677378416061401, + "num_tokens": 192920646.0, + "step": 7584 + }, + { + "epoch": 0.8329672743246211, + "grad_norm": 2.5884881019592285, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7245597839355469, + "num_tokens": 192939544.0, + "step": 7585 + }, + { + "epoch": 0.8330770920272348, + "grad_norm": 2.2168147563934326, + "learning_rate": 1e-06, + "loss": 1.0234, + "mean_token_accuracy": 0.6870824098587036, + "num_tokens": 192963428.0, + "step": 7586 + }, + { + "epoch": 0.8331869097298484, + "grad_norm": 2.149667263031006, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6904681324958801, + "num_tokens": 192989675.0, + "step": 7587 + }, + { + "epoch": 0.8332967274324621, + "grad_norm": 2.0259957313537598, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.6957637071609497, + "num_tokens": 193018154.0, + "step": 7588 + }, + { + "epoch": 0.8334065451350757, + "grad_norm": 2.0384438037872314, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6975622177124023, + "num_tokens": 193046492.0, + "step": 7589 + }, + { + "epoch": 0.8335163628376895, + "grad_norm": 2.0418083667755127, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7055004239082336, + "num_tokens": 193074930.0, + "step": 7590 + }, + { + "epoch": 0.8336261805403031, + "grad_norm": 2.2213199138641357, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6972218751907349, + "num_tokens": 193100743.0, + "step": 7591 + }, + { + "epoch": 0.8337359982429168, + "grad_norm": 2.0703914165496826, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.7020966410636902, + "num_tokens": 193128909.0, + "step": 7592 + }, + { + "epoch": 0.8338458159455304, + "grad_norm": 2.2443456649780273, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7083767652511597, + "num_tokens": 193154182.0, + "step": 7593 + }, + { + "epoch": 0.8339556336481441, + "grad_norm": 2.2140324115753174, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7209703326225281, + "num_tokens": 193176934.0, + "step": 7594 + }, + { + "epoch": 0.8340654513507577, + "grad_norm": 2.181607484817505, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.7095203399658203, + "num_tokens": 193200972.0, + "step": 7595 + }, + { + "epoch": 0.8341752690533714, + "grad_norm": 1.9895302057266235, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7089838981628418, + "num_tokens": 193229626.0, + "step": 7596 + }, + { + "epoch": 0.8342850867559851, + "grad_norm": 1.8561493158340454, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7079554200172424, + "num_tokens": 193261525.0, + "step": 7597 + }, + { + "epoch": 0.8343949044585988, + "grad_norm": 2.028350830078125, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6965861320495605, + "num_tokens": 193290073.0, + "step": 7598 + }, + { + "epoch": 0.8345047221612124, + "grad_norm": 2.503227710723877, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7322260141372681, + "num_tokens": 193310038.0, + "step": 7599 + }, + { + "epoch": 0.834614539863826, + "grad_norm": 2.1140990257263184, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7057917714118958, + "num_tokens": 193337106.0, + "step": 7600 + }, + { + "epoch": 0.8347243575664397, + "grad_norm": 2.223580837249756, + "learning_rate": 1e-06, + "loss": 1.0536, + "mean_token_accuracy": 0.6876342296600342, + "num_tokens": 193362332.0, + "step": 7601 + }, + { + "epoch": 0.8348341752690533, + "grad_norm": 2.2895755767822266, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7211739420890808, + "num_tokens": 193385162.0, + "step": 7602 + }, + { + "epoch": 0.834943992971667, + "grad_norm": 2.245319366455078, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7246443033218384, + "num_tokens": 193409130.0, + "step": 7603 + }, + { + "epoch": 0.8350538106742806, + "grad_norm": 2.0241472721099854, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6971545219421387, + "num_tokens": 193438114.0, + "step": 7604 + }, + { + "epoch": 0.8351636283768944, + "grad_norm": 2.2196707725524902, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7081087827682495, + "num_tokens": 193462145.0, + "step": 7605 + }, + { + "epoch": 0.835273446079508, + "grad_norm": 2.1308865547180176, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7060564756393433, + "num_tokens": 193487283.0, + "step": 7606 + }, + { + "epoch": 0.8353832637821217, + "grad_norm": 2.1644513607025146, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6881153583526611, + "num_tokens": 193515610.0, + "step": 7607 + }, + { + "epoch": 0.8354930814847353, + "grad_norm": 2.273799180984497, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.7023878693580627, + "num_tokens": 193540745.0, + "step": 7608 + }, + { + "epoch": 0.835602899187349, + "grad_norm": 2.1644580364227295, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.6926892995834351, + "num_tokens": 193566209.0, + "step": 7609 + }, + { + "epoch": 0.8357127168899626, + "grad_norm": 2.151075601577759, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7088145017623901, + "num_tokens": 193592918.0, + "step": 7610 + }, + { + "epoch": 0.8358225345925763, + "grad_norm": 1.9978742599487305, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7129335403442383, + "num_tokens": 193620309.0, + "step": 7611 + }, + { + "epoch": 0.83593235229519, + "grad_norm": 2.4303126335144043, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7000961899757385, + "num_tokens": 193643789.0, + "step": 7612 + }, + { + "epoch": 0.8360421699978037, + "grad_norm": 2.2624170780181885, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6906763315200806, + "num_tokens": 193669074.0, + "step": 7613 + }, + { + "epoch": 0.8361519877004173, + "grad_norm": 2.4984512329101562, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7102180123329163, + "num_tokens": 193689961.0, + "step": 7614 + }, + { + "epoch": 0.836261805403031, + "grad_norm": 2.111294746398926, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7075185179710388, + "num_tokens": 193717534.0, + "step": 7615 + }, + { + "epoch": 0.8363716231056446, + "grad_norm": 2.016172409057617, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7170702219009399, + "num_tokens": 193747089.0, + "step": 7616 + }, + { + "epoch": 0.8364814408082583, + "grad_norm": 2.257256031036377, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7065269947052002, + "num_tokens": 193771969.0, + "step": 7617 + }, + { + "epoch": 0.8365912585108719, + "grad_norm": 2.2705764770507812, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7483705282211304, + "num_tokens": 193794171.0, + "step": 7618 + }, + { + "epoch": 0.8367010762134857, + "grad_norm": 2.2845191955566406, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7180372476577759, + "num_tokens": 193816679.0, + "step": 7619 + }, + { + "epoch": 0.8368108939160993, + "grad_norm": 2.1470487117767334, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7150242328643799, + "num_tokens": 193842278.0, + "step": 7620 + }, + { + "epoch": 0.836920711618713, + "grad_norm": 2.4331369400024414, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6998767256736755, + "num_tokens": 193865096.0, + "step": 7621 + }, + { + "epoch": 0.8370305293213266, + "grad_norm": 2.1408610343933105, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7158404588699341, + "num_tokens": 193890654.0, + "step": 7622 + }, + { + "epoch": 0.8371403470239402, + "grad_norm": 2.3536574840545654, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.6978327035903931, + "num_tokens": 193913790.0, + "step": 7623 + }, + { + "epoch": 0.8372501647265539, + "grad_norm": 2.559234619140625, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.721784234046936, + "num_tokens": 193934047.0, + "step": 7624 + }, + { + "epoch": 0.8373599824291675, + "grad_norm": 2.383200168609619, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7193074226379395, + "num_tokens": 193956010.0, + "step": 7625 + }, + { + "epoch": 0.8374698001317813, + "grad_norm": 2.254244565963745, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7310847043991089, + "num_tokens": 193979570.0, + "step": 7626 + }, + { + "epoch": 0.8375796178343949, + "grad_norm": 2.456233263015747, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7329493761062622, + "num_tokens": 193999180.0, + "step": 7627 + }, + { + "epoch": 0.8376894355370086, + "grad_norm": 2.226186513900757, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.6965560913085938, + "num_tokens": 194023832.0, + "step": 7628 + }, + { + "epoch": 0.8377992532396222, + "grad_norm": 2.0218117237091064, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7372541427612305, + "num_tokens": 194052467.0, + "step": 7629 + }, + { + "epoch": 0.8379090709422359, + "grad_norm": 2.334941864013672, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7132974863052368, + "num_tokens": 194074793.0, + "step": 7630 + }, + { + "epoch": 0.8380188886448495, + "grad_norm": 2.1969945430755615, + "learning_rate": 1e-06, + "loss": 1.0864, + "mean_token_accuracy": 0.6706109046936035, + "num_tokens": 194101263.0, + "step": 7631 + }, + { + "epoch": 0.8381287063474632, + "grad_norm": 2.2237327098846436, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7248088717460632, + "num_tokens": 194125104.0, + "step": 7632 + }, + { + "epoch": 0.8382385240500769, + "grad_norm": 2.0771589279174805, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7185071110725403, + "num_tokens": 194152565.0, + "step": 7633 + }, + { + "epoch": 0.8383483417526906, + "grad_norm": 2.3331522941589355, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7086366415023804, + "num_tokens": 194176212.0, + "step": 7634 + }, + { + "epoch": 0.8384581594553042, + "grad_norm": 2.218830108642578, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.7066882848739624, + "num_tokens": 194202158.0, + "step": 7635 + }, + { + "epoch": 0.8385679771579179, + "grad_norm": 1.9549418687820435, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7070386409759521, + "num_tokens": 194233129.0, + "step": 7636 + }, + { + "epoch": 0.8386777948605315, + "grad_norm": 1.99979567527771, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.706072986125946, + "num_tokens": 194262942.0, + "step": 7637 + }, + { + "epoch": 0.8387876125631452, + "grad_norm": 2.0087757110595703, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7162669897079468, + "num_tokens": 194292438.0, + "step": 7638 + }, + { + "epoch": 0.8388974302657588, + "grad_norm": 2.337737560272217, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7204412817955017, + "num_tokens": 194317651.0, + "step": 7639 + }, + { + "epoch": 0.8390072479683724, + "grad_norm": 2.341646194458008, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6916622519493103, + "num_tokens": 194339807.0, + "step": 7640 + }, + { + "epoch": 0.8391170656709862, + "grad_norm": 2.177567720413208, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7340993881225586, + "num_tokens": 194365739.0, + "step": 7641 + }, + { + "epoch": 0.8392268833735999, + "grad_norm": 2.1777186393737793, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.716827392578125, + "num_tokens": 194393658.0, + "step": 7642 + }, + { + "epoch": 0.8393367010762135, + "grad_norm": 2.2253592014312744, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.707261323928833, + "num_tokens": 194417266.0, + "step": 7643 + }, + { + "epoch": 0.8394465187788271, + "grad_norm": 2.0878522396087646, + "learning_rate": 1e-06, + "loss": 1.0005, + "mean_token_accuracy": 0.7001469135284424, + "num_tokens": 194443942.0, + "step": 7644 + }, + { + "epoch": 0.8395563364814408, + "grad_norm": 2.0475597381591797, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.7108347415924072, + "num_tokens": 194472711.0, + "step": 7645 + }, + { + "epoch": 0.8396661541840544, + "grad_norm": 2.2895920276641846, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.7091773748397827, + "num_tokens": 194495599.0, + "step": 7646 + }, + { + "epoch": 0.8397759718866681, + "grad_norm": 2.0670816898345947, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6971675157546997, + "num_tokens": 194524172.0, + "step": 7647 + }, + { + "epoch": 0.8398857895892818, + "grad_norm": 2.1860523223876953, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7070459127426147, + "num_tokens": 194550709.0, + "step": 7648 + }, + { + "epoch": 0.8399956072918955, + "grad_norm": 1.9924300909042358, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.6962020397186279, + "num_tokens": 194580739.0, + "step": 7649 + }, + { + "epoch": 0.8401054249945091, + "grad_norm": 2.3620665073394775, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7131431102752686, + "num_tokens": 194603375.0, + "step": 7650 + }, + { + "epoch": 0.8402152426971228, + "grad_norm": 2.218487024307251, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7143001556396484, + "num_tokens": 194629456.0, + "step": 7651 + }, + { + "epoch": 0.8403250603997364, + "grad_norm": 2.1699447631835938, + "learning_rate": 1e-06, + "loss": 1.038, + "mean_token_accuracy": 0.6926642656326294, + "num_tokens": 194656680.0, + "step": 7652 + }, + { + "epoch": 0.8404348781023501, + "grad_norm": 2.254523277282715, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.6976749897003174, + "num_tokens": 194679135.0, + "step": 7653 + }, + { + "epoch": 0.8405446958049637, + "grad_norm": 2.017176389694214, + "learning_rate": 1e-06, + "loss": 1.011, + "mean_token_accuracy": 0.693626880645752, + "num_tokens": 194706759.0, + "step": 7654 + }, + { + "epoch": 0.8406545135075775, + "grad_norm": 2.31876802444458, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.7489883899688721, + "num_tokens": 194728996.0, + "step": 7655 + }, + { + "epoch": 0.8407643312101911, + "grad_norm": 2.52374529838562, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.6978170871734619, + "num_tokens": 194749760.0, + "step": 7656 + }, + { + "epoch": 0.8408741489128048, + "grad_norm": 2.161825180053711, + "learning_rate": 1e-06, + "loss": 1.0426, + "mean_token_accuracy": 0.686977207660675, + "num_tokens": 194775245.0, + "step": 7657 + }, + { + "epoch": 0.8409839666154184, + "grad_norm": 2.239832878112793, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7224695086479187, + "num_tokens": 194798088.0, + "step": 7658 + }, + { + "epoch": 0.8410937843180321, + "grad_norm": 2.273737907409668, + "learning_rate": 1e-06, + "loss": 1.012, + "mean_token_accuracy": 0.6893181800842285, + "num_tokens": 194824951.0, + "step": 7659 + }, + { + "epoch": 0.8412036020206457, + "grad_norm": 2.146650791168213, + "learning_rate": 1e-06, + "loss": 1.025, + "mean_token_accuracy": 0.6895859241485596, + "num_tokens": 194852028.0, + "step": 7660 + }, + { + "epoch": 0.8413134197232593, + "grad_norm": 2.155097246170044, + "learning_rate": 1e-06, + "loss": 1.0599, + "mean_token_accuracy": 0.6819665431976318, + "num_tokens": 194878141.0, + "step": 7661 + }, + { + "epoch": 0.8414232374258731, + "grad_norm": 2.6666758060455322, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7057726383209229, + "num_tokens": 194895618.0, + "step": 7662 + }, + { + "epoch": 0.8415330551284868, + "grad_norm": 2.153212785720825, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7188887596130371, + "num_tokens": 194923812.0, + "step": 7663 + }, + { + "epoch": 0.8416428728311004, + "grad_norm": 2.290996551513672, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7122050523757935, + "num_tokens": 194945883.0, + "step": 7664 + }, + { + "epoch": 0.841752690533714, + "grad_norm": 2.10373854637146, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7070527076721191, + "num_tokens": 194972581.0, + "step": 7665 + }, + { + "epoch": 0.8418625082363277, + "grad_norm": 2.162027597427368, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6964573264122009, + "num_tokens": 194998087.0, + "step": 7666 + }, + { + "epoch": 0.8419723259389413, + "grad_norm": 2.2310328483581543, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7184579372406006, + "num_tokens": 195022403.0, + "step": 7667 + }, + { + "epoch": 0.842082143641555, + "grad_norm": 2.4135725498199463, + "learning_rate": 1e-06, + "loss": 1.0318, + "mean_token_accuracy": 0.6927626132965088, + "num_tokens": 195045763.0, + "step": 7668 + }, + { + "epoch": 0.8421919613441686, + "grad_norm": 2.0035061836242676, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7014708518981934, + "num_tokens": 195074408.0, + "step": 7669 + }, + { + "epoch": 0.8423017790467824, + "grad_norm": 2.398782253265381, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7037484049797058, + "num_tokens": 195096876.0, + "step": 7670 + }, + { + "epoch": 0.842411596749396, + "grad_norm": 2.119394063949585, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7200638055801392, + "num_tokens": 195123328.0, + "step": 7671 + }, + { + "epoch": 0.8425214144520097, + "grad_norm": 2.339505434036255, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6964172124862671, + "num_tokens": 195148018.0, + "step": 7672 + }, + { + "epoch": 0.8426312321546233, + "grad_norm": 2.1456310749053955, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6956868171691895, + "num_tokens": 195175667.0, + "step": 7673 + }, + { + "epoch": 0.842741049857237, + "grad_norm": 2.4513885974884033, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7206621170043945, + "num_tokens": 195198878.0, + "step": 7674 + }, + { + "epoch": 0.8428508675598506, + "grad_norm": 2.135228157043457, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7397830486297607, + "num_tokens": 195223569.0, + "step": 7675 + }, + { + "epoch": 0.8429606852624643, + "grad_norm": 2.2513561248779297, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6949605941772461, + "num_tokens": 195249185.0, + "step": 7676 + }, + { + "epoch": 0.843070502965078, + "grad_norm": 2.510442018508911, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6899253129959106, + "num_tokens": 195270584.0, + "step": 7677 + }, + { + "epoch": 0.8431803206676917, + "grad_norm": 2.2147815227508545, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7163262367248535, + "num_tokens": 195294834.0, + "step": 7678 + }, + { + "epoch": 0.8432901383703053, + "grad_norm": 2.091996908187866, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7115539312362671, + "num_tokens": 195320604.0, + "step": 7679 + }, + { + "epoch": 0.843399956072919, + "grad_norm": 2.0987143516540527, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7243434190750122, + "num_tokens": 195345867.0, + "step": 7680 + }, + { + "epoch": 0.8435097737755326, + "grad_norm": 2.3796582221984863, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7067567706108093, + "num_tokens": 195366039.0, + "step": 7681 + }, + { + "epoch": 0.8436195914781462, + "grad_norm": 2.572131633758545, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.753393828868866, + "num_tokens": 195382619.0, + "step": 7682 + }, + { + "epoch": 0.8437294091807599, + "grad_norm": 2.263235569000244, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7221322059631348, + "num_tokens": 195404766.0, + "step": 7683 + }, + { + "epoch": 0.8438392268833736, + "grad_norm": 1.9577828645706177, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7073060274124146, + "num_tokens": 195435230.0, + "step": 7684 + }, + { + "epoch": 0.8439490445859873, + "grad_norm": 2.297417163848877, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.6951922178268433, + "num_tokens": 195459411.0, + "step": 7685 + }, + { + "epoch": 0.8440588622886009, + "grad_norm": 2.4819986820220947, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7071784138679504, + "num_tokens": 195479868.0, + "step": 7686 + }, + { + "epoch": 0.8441686799912146, + "grad_norm": 2.1137068271636963, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.693848729133606, + "num_tokens": 195506476.0, + "step": 7687 + }, + { + "epoch": 0.8442784976938282, + "grad_norm": 2.2699851989746094, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7092785239219666, + "num_tokens": 195530309.0, + "step": 7688 + }, + { + "epoch": 0.8443883153964419, + "grad_norm": 2.244277000427246, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7325611710548401, + "num_tokens": 195555827.0, + "step": 7689 + }, + { + "epoch": 0.8444981330990555, + "grad_norm": 2.4226293563842773, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7116140723228455, + "num_tokens": 195576121.0, + "step": 7690 + }, + { + "epoch": 0.8446079508016693, + "grad_norm": 2.3360307216644287, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.6808123588562012, + "num_tokens": 195600037.0, + "step": 7691 + }, + { + "epoch": 0.8447177685042829, + "grad_norm": 1.8969234228134155, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6934531927108765, + "num_tokens": 195633271.0, + "step": 7692 + }, + { + "epoch": 0.8448275862068966, + "grad_norm": 1.994855284690857, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.6986638307571411, + "num_tokens": 195663306.0, + "step": 7693 + }, + { + "epoch": 0.8449374039095102, + "grad_norm": 2.055246591567993, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7238022089004517, + "num_tokens": 195691103.0, + "step": 7694 + }, + { + "epoch": 0.8450472216121239, + "grad_norm": 2.6100847721099854, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7257176637649536, + "num_tokens": 195708524.0, + "step": 7695 + }, + { + "epoch": 0.8451570393147375, + "grad_norm": 2.105698347091675, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7300636768341064, + "num_tokens": 195734097.0, + "step": 7696 + }, + { + "epoch": 0.8452668570173512, + "grad_norm": 2.3562517166137695, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7179653644561768, + "num_tokens": 195756724.0, + "step": 7697 + }, + { + "epoch": 0.8453766747199648, + "grad_norm": 2.1354122161865234, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7157084941864014, + "num_tokens": 195782798.0, + "step": 7698 + }, + { + "epoch": 0.8454864924225786, + "grad_norm": 2.0760905742645264, + "learning_rate": 1e-06, + "loss": 1.0412, + "mean_token_accuracy": 0.6857529282569885, + "num_tokens": 195813237.0, + "step": 7699 + }, + { + "epoch": 0.8455963101251922, + "grad_norm": 2.1052491664886475, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7067533135414124, + "num_tokens": 195839499.0, + "step": 7700 + }, + { + "epoch": 0.8457061278278059, + "grad_norm": 2.042203903198242, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7000224590301514, + "num_tokens": 195868477.0, + "step": 7701 + }, + { + "epoch": 0.8458159455304195, + "grad_norm": 2.2008230686187744, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7239207029342651, + "num_tokens": 195892290.0, + "step": 7702 + }, + { + "epoch": 0.8459257632330331, + "grad_norm": 2.1207633018493652, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7213441133499146, + "num_tokens": 195917833.0, + "step": 7703 + }, + { + "epoch": 0.8460355809356468, + "grad_norm": 2.3190486431121826, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6987487077713013, + "num_tokens": 195941217.0, + "step": 7704 + }, + { + "epoch": 0.8461453986382604, + "grad_norm": 1.8466025590896606, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7067490816116333, + "num_tokens": 195974844.0, + "step": 7705 + }, + { + "epoch": 0.8462552163408742, + "grad_norm": 1.7752866744995117, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6921270489692688, + "num_tokens": 196012443.0, + "step": 7706 + }, + { + "epoch": 0.8463650340434878, + "grad_norm": 2.2001330852508545, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7047834396362305, + "num_tokens": 196035609.0, + "step": 7707 + }, + { + "epoch": 0.8464748517461015, + "grad_norm": 1.9887892007827759, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7081878185272217, + "num_tokens": 196064554.0, + "step": 7708 + }, + { + "epoch": 0.8465846694487151, + "grad_norm": 2.2776951789855957, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7182676196098328, + "num_tokens": 196087233.0, + "step": 7709 + }, + { + "epoch": 0.8466944871513288, + "grad_norm": 2.1732428073883057, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.709714949131012, + "num_tokens": 196113181.0, + "step": 7710 + }, + { + "epoch": 0.8468043048539424, + "grad_norm": 2.189700126647949, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7160937786102295, + "num_tokens": 196137844.0, + "step": 7711 + }, + { + "epoch": 0.8469141225565561, + "grad_norm": 1.8474081754684448, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6980149745941162, + "num_tokens": 196173018.0, + "step": 7712 + }, + { + "epoch": 0.8470239402591698, + "grad_norm": 2.211083173751831, + "learning_rate": 1e-06, + "loss": 1.0166, + "mean_token_accuracy": 0.6991901993751526, + "num_tokens": 196198262.0, + "step": 7713 + }, + { + "epoch": 0.8471337579617835, + "grad_norm": 2.1440184116363525, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7128284573554993, + "num_tokens": 196222513.0, + "step": 7714 + }, + { + "epoch": 0.8472435756643971, + "grad_norm": 2.267814874649048, + "learning_rate": 1e-06, + "loss": 1.0111, + "mean_token_accuracy": 0.6937428712844849, + "num_tokens": 196248046.0, + "step": 7715 + }, + { + "epoch": 0.8473533933670108, + "grad_norm": 1.9858570098876953, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7070106267929077, + "num_tokens": 196277694.0, + "step": 7716 + }, + { + "epoch": 0.8474632110696244, + "grad_norm": 2.3280913829803467, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.696678638458252, + "num_tokens": 196302935.0, + "step": 7717 + }, + { + "epoch": 0.8475730287722381, + "grad_norm": 2.134016752243042, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6881735324859619, + "num_tokens": 196329639.0, + "step": 7718 + }, + { + "epoch": 0.8476828464748517, + "grad_norm": 2.0517613887786865, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7234586477279663, + "num_tokens": 196357856.0, + "step": 7719 + }, + { + "epoch": 0.8477926641774655, + "grad_norm": 2.100348472595215, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7227834463119507, + "num_tokens": 196383672.0, + "step": 7720 + }, + { + "epoch": 0.8479024818800791, + "grad_norm": 1.9810433387756348, + "learning_rate": 1e-06, + "loss": 1.0145, + "mean_token_accuracy": 0.6891628503799438, + "num_tokens": 196414363.0, + "step": 7721 + }, + { + "epoch": 0.8480122995826928, + "grad_norm": 2.3179244995117188, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7227878570556641, + "num_tokens": 196435698.0, + "step": 7722 + }, + { + "epoch": 0.8481221172853064, + "grad_norm": 2.201589822769165, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.692396879196167, + "num_tokens": 196462669.0, + "step": 7723 + }, + { + "epoch": 0.84823193498792, + "grad_norm": 2.079615354537964, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.7021316289901733, + "num_tokens": 196488351.0, + "step": 7724 + }, + { + "epoch": 0.8483417526905337, + "grad_norm": 2.193773031234741, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7019762396812439, + "num_tokens": 196514740.0, + "step": 7725 + }, + { + "epoch": 0.8484515703931473, + "grad_norm": 2.3594858646392822, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7012697458267212, + "num_tokens": 196537811.0, + "step": 7726 + }, + { + "epoch": 0.848561388095761, + "grad_norm": 2.1442019939422607, + "learning_rate": 1e-06, + "loss": 1.0489, + "mean_token_accuracy": 0.6804227828979492, + "num_tokens": 196566051.0, + "step": 7727 + }, + { + "epoch": 0.8486712057983747, + "grad_norm": 2.0316948890686035, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7034860849380493, + "num_tokens": 196595742.0, + "step": 7728 + }, + { + "epoch": 0.8487810235009884, + "grad_norm": 2.1619858741760254, + "learning_rate": 1e-06, + "loss": 1.0172, + "mean_token_accuracy": 0.6995199918746948, + "num_tokens": 196622367.0, + "step": 7729 + }, + { + "epoch": 0.848890841203602, + "grad_norm": 2.012545347213745, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7102667689323425, + "num_tokens": 196650867.0, + "step": 7730 + }, + { + "epoch": 0.8490006589062157, + "grad_norm": 2.029428005218506, + "learning_rate": 1e-06, + "loss": 1.0074, + "mean_token_accuracy": 0.6874185800552368, + "num_tokens": 196680110.0, + "step": 7731 + }, + { + "epoch": 0.8491104766088293, + "grad_norm": 2.262591600418091, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7200164794921875, + "num_tokens": 196702797.0, + "step": 7732 + }, + { + "epoch": 0.849220294311443, + "grad_norm": 2.494760751724243, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7062031626701355, + "num_tokens": 196723027.0, + "step": 7733 + }, + { + "epoch": 0.8493301120140566, + "grad_norm": 2.619703769683838, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7177537679672241, + "num_tokens": 196740555.0, + "step": 7734 + }, + { + "epoch": 0.8494399297166704, + "grad_norm": 2.1581809520721436, + "learning_rate": 1e-06, + "loss": 1.0763, + "mean_token_accuracy": 0.6810839772224426, + "num_tokens": 196768043.0, + "step": 7735 + }, + { + "epoch": 0.849549747419284, + "grad_norm": 2.091977596282959, + "learning_rate": 1e-06, + "loss": 1.0012, + "mean_token_accuracy": 0.6962631940841675, + "num_tokens": 196794998.0, + "step": 7736 + }, + { + "epoch": 0.8496595651218977, + "grad_norm": 2.1762468814849854, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.6946521997451782, + "num_tokens": 196820769.0, + "step": 7737 + }, + { + "epoch": 0.8497693828245113, + "grad_norm": 2.4417731761932373, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7330167889595032, + "num_tokens": 196840197.0, + "step": 7738 + }, + { + "epoch": 0.849879200527125, + "grad_norm": 2.257110834121704, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7360787391662598, + "num_tokens": 196861910.0, + "step": 7739 + }, + { + "epoch": 0.8499890182297386, + "grad_norm": 1.9978229999542236, + "learning_rate": 1e-06, + "loss": 1.0365, + "mean_token_accuracy": 0.6939906477928162, + "num_tokens": 196891366.0, + "step": 7740 + }, + { + "epoch": 0.8500988359323522, + "grad_norm": 2.334409475326538, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7058308720588684, + "num_tokens": 196914532.0, + "step": 7741 + }, + { + "epoch": 0.850208653634966, + "grad_norm": 2.2534689903259277, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.730663001537323, + "num_tokens": 196936056.0, + "step": 7742 + }, + { + "epoch": 0.8503184713375797, + "grad_norm": 2.003211736679077, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7179481387138367, + "num_tokens": 196965167.0, + "step": 7743 + }, + { + "epoch": 0.8504282890401933, + "grad_norm": 2.016756534576416, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.72525554895401, + "num_tokens": 196996159.0, + "step": 7744 + }, + { + "epoch": 0.8505381067428069, + "grad_norm": 2.312664747238159, + "learning_rate": 1e-06, + "loss": 1.0606, + "mean_token_accuracy": 0.6860071420669556, + "num_tokens": 197020113.0, + "step": 7745 + }, + { + "epoch": 0.8506479244454206, + "grad_norm": 1.8965563774108887, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6909714937210083, + "num_tokens": 197051461.0, + "step": 7746 + }, + { + "epoch": 0.8507577421480342, + "grad_norm": 2.3799052238464355, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7060057520866394, + "num_tokens": 197072338.0, + "step": 7747 + }, + { + "epoch": 0.8508675598506479, + "grad_norm": 2.21286678314209, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7278400659561157, + "num_tokens": 197094642.0, + "step": 7748 + }, + { + "epoch": 0.8509773775532616, + "grad_norm": 2.5923819541931152, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7098538875579834, + "num_tokens": 197112754.0, + "step": 7749 + }, + { + "epoch": 0.8510871952558753, + "grad_norm": 1.9671778678894043, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7126331329345703, + "num_tokens": 197142461.0, + "step": 7750 + }, + { + "epoch": 0.8511970129584889, + "grad_norm": 2.205470085144043, + "learning_rate": 1e-06, + "loss": 1.0226, + "mean_token_accuracy": 0.6933172941207886, + "num_tokens": 197169884.0, + "step": 7751 + }, + { + "epoch": 0.8513068306611026, + "grad_norm": 2.294599771499634, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.6998811960220337, + "num_tokens": 197193930.0, + "step": 7752 + }, + { + "epoch": 0.8514166483637162, + "grad_norm": 1.9151439666748047, + "learning_rate": 1e-06, + "loss": 1.048, + "mean_token_accuracy": 0.6810067296028137, + "num_tokens": 197227294.0, + "step": 7753 + }, + { + "epoch": 0.8515264660663299, + "grad_norm": 2.26556134223938, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7014044523239136, + "num_tokens": 197252643.0, + "step": 7754 + }, + { + "epoch": 0.8516362837689435, + "grad_norm": 2.588132381439209, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7277712821960449, + "num_tokens": 197271289.0, + "step": 7755 + }, + { + "epoch": 0.8517461014715572, + "grad_norm": 2.3735334873199463, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7219760417938232, + "num_tokens": 197292049.0, + "step": 7756 + }, + { + "epoch": 0.8518559191741709, + "grad_norm": 2.268401622772217, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7229335904121399, + "num_tokens": 197316852.0, + "step": 7757 + }, + { + "epoch": 0.8519657368767846, + "grad_norm": 2.085650682449341, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7024147510528564, + "num_tokens": 197345051.0, + "step": 7758 + }, + { + "epoch": 0.8520755545793982, + "grad_norm": 2.5164201259613037, + "learning_rate": 1e-06, + "loss": 1.0345, + "mean_token_accuracy": 0.6981235146522522, + "num_tokens": 197367266.0, + "step": 7759 + }, + { + "epoch": 0.8521853722820119, + "grad_norm": 2.5186352729797363, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7027007341384888, + "num_tokens": 197397007.0, + "step": 7760 + }, + { + "epoch": 0.8522951899846255, + "grad_norm": 2.1684844493865967, + "learning_rate": 1e-06, + "loss": 1.0035, + "mean_token_accuracy": 0.6902342438697815, + "num_tokens": 197423397.0, + "step": 7761 + }, + { + "epoch": 0.8524050076872391, + "grad_norm": 2.296344518661499, + "learning_rate": 1e-06, + "loss": 1.0605, + "mean_token_accuracy": 0.673862874507904, + "num_tokens": 197448366.0, + "step": 7762 + }, + { + "epoch": 0.8525148253898528, + "grad_norm": 2.151153087615967, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7072513103485107, + "num_tokens": 197472924.0, + "step": 7763 + }, + { + "epoch": 0.8526246430924665, + "grad_norm": 2.2047181129455566, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.7007845044136047, + "num_tokens": 197497507.0, + "step": 7764 + }, + { + "epoch": 0.8527344607950802, + "grad_norm": 2.157407760620117, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7039147615432739, + "num_tokens": 197522337.0, + "step": 7765 + }, + { + "epoch": 0.8528442784976938, + "grad_norm": 2.2897841930389404, + "learning_rate": 1e-06, + "loss": 1.0706, + "mean_token_accuracy": 0.6794309616088867, + "num_tokens": 197547496.0, + "step": 7766 + }, + { + "epoch": 0.8529540962003075, + "grad_norm": 2.488363027572632, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6954104900360107, + "num_tokens": 197568804.0, + "step": 7767 + }, + { + "epoch": 0.8530639139029211, + "grad_norm": 2.214379072189331, + "learning_rate": 1e-06, + "loss": 1.0538, + "mean_token_accuracy": 0.676279604434967, + "num_tokens": 197597158.0, + "step": 7768 + }, + { + "epoch": 0.8531737316055348, + "grad_norm": 2.4258530139923096, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7176378965377808, + "num_tokens": 197618820.0, + "step": 7769 + }, + { + "epoch": 0.8532835493081484, + "grad_norm": 2.177314519882202, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7209511995315552, + "num_tokens": 197642972.0, + "step": 7770 + }, + { + "epoch": 0.8533933670107622, + "grad_norm": 2.2766025066375732, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7193488478660583, + "num_tokens": 197666019.0, + "step": 7771 + }, + { + "epoch": 0.8535031847133758, + "grad_norm": 2.067896842956543, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.709987998008728, + "num_tokens": 197696047.0, + "step": 7772 + }, + { + "epoch": 0.8536130024159895, + "grad_norm": 2.084134578704834, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7334530353546143, + "num_tokens": 197723081.0, + "step": 7773 + }, + { + "epoch": 0.8537228201186031, + "grad_norm": 2.31366229057312, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7108290791511536, + "num_tokens": 197747196.0, + "step": 7774 + }, + { + "epoch": 0.8538326378212168, + "grad_norm": 2.052220582962036, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7218497395515442, + "num_tokens": 197776365.0, + "step": 7775 + }, + { + "epoch": 0.8539424555238304, + "grad_norm": 2.2964284420013428, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.6961756944656372, + "num_tokens": 197799707.0, + "step": 7776 + }, + { + "epoch": 0.8540522732264441, + "grad_norm": 2.169191598892212, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7135177850723267, + "num_tokens": 197825146.0, + "step": 7777 + }, + { + "epoch": 0.8541620909290578, + "grad_norm": 2.2584407329559326, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7480128407478333, + "num_tokens": 197846772.0, + "step": 7778 + }, + { + "epoch": 0.8542719086316715, + "grad_norm": 2.280951738357544, + "learning_rate": 1e-06, + "loss": 1.0188, + "mean_token_accuracy": 0.6843148469924927, + "num_tokens": 197870783.0, + "step": 7779 + }, + { + "epoch": 0.8543817263342851, + "grad_norm": 1.941087007522583, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7039877772331238, + "num_tokens": 197902617.0, + "step": 7780 + }, + { + "epoch": 0.8544915440368988, + "grad_norm": 2.118769645690918, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6991885304450989, + "num_tokens": 197930343.0, + "step": 7781 + }, + { + "epoch": 0.8546013617395124, + "grad_norm": 1.9859787225723267, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.6998543739318848, + "num_tokens": 197957553.0, + "step": 7782 + }, + { + "epoch": 0.854711179442126, + "grad_norm": 1.9884121417999268, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7128870487213135, + "num_tokens": 197984825.0, + "step": 7783 + }, + { + "epoch": 0.8548209971447397, + "grad_norm": 2.0606677532196045, + "learning_rate": 1e-06, + "loss": 1.0015, + "mean_token_accuracy": 0.6920430660247803, + "num_tokens": 198012720.0, + "step": 7784 + }, + { + "epoch": 0.8549308148473534, + "grad_norm": 2.3605034351348877, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7221948504447937, + "num_tokens": 198033400.0, + "step": 7785 + }, + { + "epoch": 0.8550406325499671, + "grad_norm": 2.5053915977478027, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7016526460647583, + "num_tokens": 198054805.0, + "step": 7786 + }, + { + "epoch": 0.8551504502525807, + "grad_norm": 2.1142783164978027, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.7052772045135498, + "num_tokens": 198080960.0, + "step": 7787 + }, + { + "epoch": 0.8552602679551944, + "grad_norm": 2.006037712097168, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7069231271743774, + "num_tokens": 198111183.0, + "step": 7788 + }, + { + "epoch": 0.855370085657808, + "grad_norm": 2.111816883087158, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.711056113243103, + "num_tokens": 198137557.0, + "step": 7789 + }, + { + "epoch": 0.8554799033604217, + "grad_norm": 2.090970754623413, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7094083428382874, + "num_tokens": 198163145.0, + "step": 7790 + }, + { + "epoch": 0.8555897210630353, + "grad_norm": 2.124950885772705, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7176162004470825, + "num_tokens": 198186834.0, + "step": 7791 + }, + { + "epoch": 0.855699538765649, + "grad_norm": 2.3188092708587646, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7209881544113159, + "num_tokens": 198208861.0, + "step": 7792 + }, + { + "epoch": 0.8558093564682627, + "grad_norm": 2.102994680404663, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7076636552810669, + "num_tokens": 198235522.0, + "step": 7793 + }, + { + "epoch": 0.8559191741708764, + "grad_norm": 2.1703577041625977, + "learning_rate": 1e-06, + "loss": 1.0338, + "mean_token_accuracy": 0.6912400722503662, + "num_tokens": 198259196.0, + "step": 7794 + }, + { + "epoch": 0.85602899187349, + "grad_norm": 1.8923434019088745, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7087626457214355, + "num_tokens": 198291738.0, + "step": 7795 + }, + { + "epoch": 0.8561388095761037, + "grad_norm": 2.3927440643310547, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.6997274160385132, + "num_tokens": 198315073.0, + "step": 7796 + }, + { + "epoch": 0.8562486272787173, + "grad_norm": 2.6626148223876953, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6996845006942749, + "num_tokens": 198333843.0, + "step": 7797 + }, + { + "epoch": 0.856358444981331, + "grad_norm": 2.3917436599731445, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7187021374702454, + "num_tokens": 198354509.0, + "step": 7798 + }, + { + "epoch": 0.8564682626839446, + "grad_norm": 2.603468179702759, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7209296226501465, + "num_tokens": 198372845.0, + "step": 7799 + }, + { + "epoch": 0.8565780803865584, + "grad_norm": 2.5402238368988037, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.729694128036499, + "num_tokens": 198390282.0, + "step": 7800 + }, + { + "epoch": 0.856687898089172, + "grad_norm": 2.3423683643341064, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7241698503494263, + "num_tokens": 198412664.0, + "step": 7801 + }, + { + "epoch": 0.8567977157917857, + "grad_norm": 2.0727858543395996, + "learning_rate": 1e-06, + "loss": 1.053, + "mean_token_accuracy": 0.6822856664657593, + "num_tokens": 198441110.0, + "step": 7802 + }, + { + "epoch": 0.8569075334943993, + "grad_norm": 2.3723795413970947, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7087407112121582, + "num_tokens": 198463411.0, + "step": 7803 + }, + { + "epoch": 0.8570173511970129, + "grad_norm": 2.1707825660705566, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7172662019729614, + "num_tokens": 198490077.0, + "step": 7804 + }, + { + "epoch": 0.8571271688996266, + "grad_norm": 2.2236826419830322, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.69024258852005, + "num_tokens": 198514412.0, + "step": 7805 + }, + { + "epoch": 0.8572369866022402, + "grad_norm": 1.9934581518173218, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.7032502889633179, + "num_tokens": 198543035.0, + "step": 7806 + }, + { + "epoch": 0.857346804304854, + "grad_norm": 2.106802463531494, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7001641988754272, + "num_tokens": 198568951.0, + "step": 7807 + }, + { + "epoch": 0.8574566220074676, + "grad_norm": 2.1505236625671387, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7028895616531372, + "num_tokens": 198596388.0, + "step": 7808 + }, + { + "epoch": 0.8575664397100813, + "grad_norm": 1.9819084405899048, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6952215433120728, + "num_tokens": 198626921.0, + "step": 7809 + }, + { + "epoch": 0.8576762574126949, + "grad_norm": 2.089996337890625, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7237095832824707, + "num_tokens": 198653376.0, + "step": 7810 + }, + { + "epoch": 0.8577860751153086, + "grad_norm": 2.408381700515747, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7158019542694092, + "num_tokens": 198674955.0, + "step": 7811 + }, + { + "epoch": 0.8578958928179222, + "grad_norm": 2.250319242477417, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7118529081344604, + "num_tokens": 198698850.0, + "step": 7812 + }, + { + "epoch": 0.8580057105205359, + "grad_norm": 2.2823500633239746, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7230451703071594, + "num_tokens": 198721681.0, + "step": 7813 + }, + { + "epoch": 0.8581155282231496, + "grad_norm": 2.2502007484436035, + "learning_rate": 1e-06, + "loss": 1.039, + "mean_token_accuracy": 0.6889411807060242, + "num_tokens": 198750100.0, + "step": 7814 + }, + { + "epoch": 0.8582253459257633, + "grad_norm": 2.2615232467651367, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6979940533638, + "num_tokens": 198773371.0, + "step": 7815 + }, + { + "epoch": 0.8583351636283769, + "grad_norm": 2.0153558254241943, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.698672890663147, + "num_tokens": 198804293.0, + "step": 7816 + }, + { + "epoch": 0.8584449813309906, + "grad_norm": 2.127656936645508, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.6990876197814941, + "num_tokens": 198829437.0, + "step": 7817 + }, + { + "epoch": 0.8585547990336042, + "grad_norm": 2.092094659805298, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7458639740943909, + "num_tokens": 198854306.0, + "step": 7818 + }, + { + "epoch": 0.8586646167362179, + "grad_norm": 2.211885452270508, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7134386301040649, + "num_tokens": 198880403.0, + "step": 7819 + }, + { + "epoch": 0.8587744344388315, + "grad_norm": 2.1362953186035156, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7164158225059509, + "num_tokens": 198907238.0, + "step": 7820 + }, + { + "epoch": 0.8588842521414451, + "grad_norm": 2.34598445892334, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7022453546524048, + "num_tokens": 198929613.0, + "step": 7821 + }, + { + "epoch": 0.8589940698440589, + "grad_norm": 2.0917325019836426, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6856898665428162, + "num_tokens": 198956480.0, + "step": 7822 + }, + { + "epoch": 0.8591038875466726, + "grad_norm": 2.1764883995056152, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7173497676849365, + "num_tokens": 198981493.0, + "step": 7823 + }, + { + "epoch": 0.8592137052492862, + "grad_norm": 2.2702107429504395, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7004867196083069, + "num_tokens": 199004312.0, + "step": 7824 + }, + { + "epoch": 0.8593235229518998, + "grad_norm": 2.1406054496765137, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.7063866853713989, + "num_tokens": 199029680.0, + "step": 7825 + }, + { + "epoch": 0.8594333406545135, + "grad_norm": 2.2756919860839844, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.7008944153785706, + "num_tokens": 199054649.0, + "step": 7826 + }, + { + "epoch": 0.8595431583571271, + "grad_norm": 1.980002522468567, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7149524688720703, + "num_tokens": 199082032.0, + "step": 7827 + }, + { + "epoch": 0.8596529760597408, + "grad_norm": 2.052462339401245, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7091410160064697, + "num_tokens": 199108100.0, + "step": 7828 + }, + { + "epoch": 0.8597627937623545, + "grad_norm": 2.342435836791992, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7470303177833557, + "num_tokens": 199128468.0, + "step": 7829 + }, + { + "epoch": 0.8598726114649682, + "grad_norm": 2.2432971000671387, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7049914598464966, + "num_tokens": 199152522.0, + "step": 7830 + }, + { + "epoch": 0.8599824291675818, + "grad_norm": 2.143799304962158, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7233182191848755, + "num_tokens": 199178321.0, + "step": 7831 + }, + { + "epoch": 0.8600922468701955, + "grad_norm": 2.0268378257751465, + "learning_rate": 1e-06, + "loss": 1.0137, + "mean_token_accuracy": 0.6921542882919312, + "num_tokens": 199208254.0, + "step": 7832 + }, + { + "epoch": 0.8602020645728091, + "grad_norm": 2.107050895690918, + "learning_rate": 1e-06, + "loss": 1.0362, + "mean_token_accuracy": 0.689681887626648, + "num_tokens": 199235302.0, + "step": 7833 + }, + { + "epoch": 0.8603118822754228, + "grad_norm": 2.289426326751709, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7069036364555359, + "num_tokens": 199258754.0, + "step": 7834 + }, + { + "epoch": 0.8604216999780364, + "grad_norm": 2.2339961528778076, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.700406551361084, + "num_tokens": 199282384.0, + "step": 7835 + }, + { + "epoch": 0.8605315176806502, + "grad_norm": 2.127078056335449, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7081249952316284, + "num_tokens": 199308460.0, + "step": 7836 + }, + { + "epoch": 0.8606413353832638, + "grad_norm": 2.1685118675231934, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7037209272384644, + "num_tokens": 199334108.0, + "step": 7837 + }, + { + "epoch": 0.8607511530858775, + "grad_norm": 2.170759439468384, + "learning_rate": 1e-06, + "loss": 1.0655, + "mean_token_accuracy": 0.676959216594696, + "num_tokens": 199363484.0, + "step": 7838 + }, + { + "epoch": 0.8608609707884911, + "grad_norm": 1.963333010673523, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7026084661483765, + "num_tokens": 199393981.0, + "step": 7839 + }, + { + "epoch": 0.8609707884911048, + "grad_norm": 1.939987301826477, + "learning_rate": 1e-06, + "loss": 1.0627, + "mean_token_accuracy": 0.6784103512763977, + "num_tokens": 199426571.0, + "step": 7840 + }, + { + "epoch": 0.8610806061937184, + "grad_norm": 2.0642006397247314, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6957815885543823, + "num_tokens": 199455194.0, + "step": 7841 + }, + { + "epoch": 0.861190423896332, + "grad_norm": 2.1553516387939453, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7112964987754822, + "num_tokens": 199482691.0, + "step": 7842 + }, + { + "epoch": 0.8613002415989458, + "grad_norm": 2.3662807941436768, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7392092943191528, + "num_tokens": 199503512.0, + "step": 7843 + }, + { + "epoch": 0.8614100593015594, + "grad_norm": 2.0070841312408447, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7011399865150452, + "num_tokens": 199531726.0, + "step": 7844 + }, + { + "epoch": 0.8615198770041731, + "grad_norm": 2.306001663208008, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7107590436935425, + "num_tokens": 199553504.0, + "step": 7845 + }, + { + "epoch": 0.8616296947067867, + "grad_norm": 2.3779401779174805, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7206622958183289, + "num_tokens": 199574831.0, + "step": 7846 + }, + { + "epoch": 0.8617395124094004, + "grad_norm": 2.459341526031494, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.713295042514801, + "num_tokens": 199595261.0, + "step": 7847 + }, + { + "epoch": 0.861849330112014, + "grad_norm": 2.1514923572540283, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6988129019737244, + "num_tokens": 199622477.0, + "step": 7848 + }, + { + "epoch": 0.8619591478146277, + "grad_norm": 2.3920159339904785, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.70819091796875, + "num_tokens": 199644302.0, + "step": 7849 + }, + { + "epoch": 0.8620689655172413, + "grad_norm": 1.9627231359481812, + "learning_rate": 1e-06, + "loss": 1.009, + "mean_token_accuracy": 0.7007023096084595, + "num_tokens": 199675575.0, + "step": 7850 + }, + { + "epoch": 0.8621787832198551, + "grad_norm": 1.9165112972259521, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6971418261528015, + "num_tokens": 199706488.0, + "step": 7851 + }, + { + "epoch": 0.8622886009224687, + "grad_norm": 2.390608072280884, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7174392938613892, + "num_tokens": 199728380.0, + "step": 7852 + }, + { + "epoch": 0.8623984186250824, + "grad_norm": 2.2420623302459717, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7211325168609619, + "num_tokens": 199751799.0, + "step": 7853 + }, + { + "epoch": 0.862508236327696, + "grad_norm": 2.218874454498291, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7065664529800415, + "num_tokens": 199776877.0, + "step": 7854 + }, + { + "epoch": 0.8626180540303097, + "grad_norm": 2.324556350708008, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7241873145103455, + "num_tokens": 199798787.0, + "step": 7855 + }, + { + "epoch": 0.8627278717329233, + "grad_norm": 2.291243076324463, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7142431735992432, + "num_tokens": 199821289.0, + "step": 7856 + }, + { + "epoch": 0.862837689435537, + "grad_norm": 2.1512577533721924, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7104145884513855, + "num_tokens": 199847352.0, + "step": 7857 + }, + { + "epoch": 0.8629475071381507, + "grad_norm": 2.1887848377227783, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7099704742431641, + "num_tokens": 199871108.0, + "step": 7858 + }, + { + "epoch": 0.8630573248407644, + "grad_norm": 2.5014960765838623, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.709243655204773, + "num_tokens": 199893431.0, + "step": 7859 + }, + { + "epoch": 0.863167142543378, + "grad_norm": 1.9574341773986816, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7158956527709961, + "num_tokens": 199920904.0, + "step": 7860 + }, + { + "epoch": 0.8632769602459917, + "grad_norm": 2.4583773612976074, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7288718223571777, + "num_tokens": 199941402.0, + "step": 7861 + }, + { + "epoch": 0.8633867779486053, + "grad_norm": 2.5217642784118652, + "learning_rate": 1e-06, + "loss": 1.0092, + "mean_token_accuracy": 0.6927279829978943, + "num_tokens": 199961743.0, + "step": 7862 + }, + { + "epoch": 0.863496595651219, + "grad_norm": 2.4243414402008057, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7260311841964722, + "num_tokens": 199984026.0, + "step": 7863 + }, + { + "epoch": 0.8636064133538326, + "grad_norm": 2.3120131492614746, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7293379306793213, + "num_tokens": 200005859.0, + "step": 7864 + }, + { + "epoch": 0.8637162310564463, + "grad_norm": 2.1571459770202637, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6950536966323853, + "num_tokens": 200031858.0, + "step": 7865 + }, + { + "epoch": 0.86382604875906, + "grad_norm": 2.1532835960388184, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.6950669884681702, + "num_tokens": 200058336.0, + "step": 7866 + }, + { + "epoch": 0.8639358664616736, + "grad_norm": 2.2352209091186523, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7154866456985474, + "num_tokens": 200081882.0, + "step": 7867 + }, + { + "epoch": 0.8640456841642873, + "grad_norm": 2.7822375297546387, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6887305974960327, + "num_tokens": 200099017.0, + "step": 7868 + }, + { + "epoch": 0.8641555018669009, + "grad_norm": 2.20444917678833, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6948032975196838, + "num_tokens": 200126042.0, + "step": 7869 + }, + { + "epoch": 0.8642653195695146, + "grad_norm": 2.139021158218384, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6907576322555542, + "num_tokens": 200152652.0, + "step": 7870 + }, + { + "epoch": 0.8643751372721282, + "grad_norm": 2.111860990524292, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7324659824371338, + "num_tokens": 200177856.0, + "step": 7871 + }, + { + "epoch": 0.864484954974742, + "grad_norm": 2.1717514991760254, + "learning_rate": 1e-06, + "loss": 0.7706, + "mean_token_accuracy": 0.7474874258041382, + "num_tokens": 200200609.0, + "step": 7872 + }, + { + "epoch": 0.8645947726773556, + "grad_norm": 2.0644001960754395, + "learning_rate": 1e-06, + "loss": 1.0372, + "mean_token_accuracy": 0.6909224987030029, + "num_tokens": 200227643.0, + "step": 7873 + }, + { + "epoch": 0.8647045903799693, + "grad_norm": 2.037794351577759, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7199420928955078, + "num_tokens": 200253383.0, + "step": 7874 + }, + { + "epoch": 0.8648144080825829, + "grad_norm": 2.4570181369781494, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6965570449829102, + "num_tokens": 200272529.0, + "step": 7875 + }, + { + "epoch": 0.8649242257851966, + "grad_norm": 2.2639219760894775, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7129079103469849, + "num_tokens": 200295763.0, + "step": 7876 + }, + { + "epoch": 0.8650340434878102, + "grad_norm": 2.1665589809417725, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.724428117275238, + "num_tokens": 200320673.0, + "step": 7877 + }, + { + "epoch": 0.8651438611904239, + "grad_norm": 2.232276201248169, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7275984287261963, + "num_tokens": 200343845.0, + "step": 7878 + }, + { + "epoch": 0.8652536788930375, + "grad_norm": 2.2444796562194824, + "learning_rate": 1e-06, + "loss": 1.0393, + "mean_token_accuracy": 0.6823908090591431, + "num_tokens": 200370113.0, + "step": 7879 + }, + { + "epoch": 0.8653634965956513, + "grad_norm": 2.2089345455169678, + "learning_rate": 1e-06, + "loss": 1.0283, + "mean_token_accuracy": 0.6837645769119263, + "num_tokens": 200398069.0, + "step": 7880 + }, + { + "epoch": 0.8654733142982649, + "grad_norm": 2.2617006301879883, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.7027337551116943, + "num_tokens": 200422221.0, + "step": 7881 + }, + { + "epoch": 0.8655831320008786, + "grad_norm": 2.033313035964966, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7011116743087769, + "num_tokens": 200449835.0, + "step": 7882 + }, + { + "epoch": 0.8656929497034922, + "grad_norm": 2.1459901332855225, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6991457939147949, + "num_tokens": 200476190.0, + "step": 7883 + }, + { + "epoch": 0.8658027674061058, + "grad_norm": 2.528224229812622, + "learning_rate": 1e-06, + "loss": 1.0534, + "mean_token_accuracy": 0.6796224117279053, + "num_tokens": 200496807.0, + "step": 7884 + }, + { + "epoch": 0.8659125851087195, + "grad_norm": 1.8841807842254639, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7067338228225708, + "num_tokens": 200528411.0, + "step": 7885 + }, + { + "epoch": 0.8660224028113331, + "grad_norm": 2.0462729930877686, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.6962172985076904, + "num_tokens": 200557238.0, + "step": 7886 + }, + { + "epoch": 0.8661322205139469, + "grad_norm": 2.10298490524292, + "learning_rate": 1e-06, + "loss": 1.059, + "mean_token_accuracy": 0.6796915531158447, + "num_tokens": 200586716.0, + "step": 7887 + }, + { + "epoch": 0.8662420382165605, + "grad_norm": 1.9880915880203247, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.6998345851898193, + "num_tokens": 200618930.0, + "step": 7888 + }, + { + "epoch": 0.8663518559191742, + "grad_norm": 1.975825548171997, + "learning_rate": 1e-06, + "loss": 1.0874, + "mean_token_accuracy": 0.6789112091064453, + "num_tokens": 200649417.0, + "step": 7889 + }, + { + "epoch": 0.8664616736217878, + "grad_norm": 2.279386520385742, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7269042730331421, + "num_tokens": 200672256.0, + "step": 7890 + }, + { + "epoch": 0.8665714913244015, + "grad_norm": 2.2560243606567383, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.6944586038589478, + "num_tokens": 200697244.0, + "step": 7891 + }, + { + "epoch": 0.8666813090270151, + "grad_norm": 1.9671744108200073, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7296348810195923, + "num_tokens": 200725199.0, + "step": 7892 + }, + { + "epoch": 0.8667911267296288, + "grad_norm": 2.211468458175659, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7107970714569092, + "num_tokens": 200748229.0, + "step": 7893 + }, + { + "epoch": 0.8669009444322425, + "grad_norm": 2.058403491973877, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.7014168500900269, + "num_tokens": 200778152.0, + "step": 7894 + }, + { + "epoch": 0.8670107621348562, + "grad_norm": 2.244152069091797, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7241382002830505, + "num_tokens": 200802362.0, + "step": 7895 + }, + { + "epoch": 0.8671205798374698, + "grad_norm": 2.258690118789673, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6936267614364624, + "num_tokens": 200826991.0, + "step": 7896 + }, + { + "epoch": 0.8672303975400835, + "grad_norm": 1.943343997001648, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.6933401823043823, + "num_tokens": 200857446.0, + "step": 7897 + }, + { + "epoch": 0.8673402152426971, + "grad_norm": 1.9616166353225708, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6961103677749634, + "num_tokens": 200891624.0, + "step": 7898 + }, + { + "epoch": 0.8674500329453108, + "grad_norm": 2.2666285037994385, + "learning_rate": 1e-06, + "loss": 1.0358, + "mean_token_accuracy": 0.6803818941116333, + "num_tokens": 200916472.0, + "step": 7899 + }, + { + "epoch": 0.8675598506479244, + "grad_norm": 2.2009541988372803, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7179660797119141, + "num_tokens": 200940831.0, + "step": 7900 + }, + { + "epoch": 0.8676696683505382, + "grad_norm": 2.2646825313568115, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7210176587104797, + "num_tokens": 200965155.0, + "step": 7901 + }, + { + "epoch": 0.8677794860531518, + "grad_norm": 2.3745574951171875, + "learning_rate": 1e-06, + "loss": 1.0332, + "mean_token_accuracy": 0.7003880739212036, + "num_tokens": 200987909.0, + "step": 7902 + }, + { + "epoch": 0.8678893037557655, + "grad_norm": 2.3052356243133545, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7235220670700073, + "num_tokens": 201011343.0, + "step": 7903 + }, + { + "epoch": 0.8679991214583791, + "grad_norm": 1.9340510368347168, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6949993371963501, + "num_tokens": 201042879.0, + "step": 7904 + }, + { + "epoch": 0.8681089391609927, + "grad_norm": 2.347280263900757, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.6945987939834595, + "num_tokens": 201066711.0, + "step": 7905 + }, + { + "epoch": 0.8682187568636064, + "grad_norm": 2.388132333755493, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6998951435089111, + "num_tokens": 201089159.0, + "step": 7906 + }, + { + "epoch": 0.86832857456622, + "grad_norm": 2.022247552871704, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.6875178813934326, + "num_tokens": 201119297.0, + "step": 7907 + }, + { + "epoch": 0.8684383922688337, + "grad_norm": 1.8918675184249878, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7122284770011902, + "num_tokens": 201151390.0, + "step": 7908 + }, + { + "epoch": 0.8685482099714474, + "grad_norm": 2.176392078399658, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.6873416900634766, + "num_tokens": 201177027.0, + "step": 7909 + }, + { + "epoch": 0.8686580276740611, + "grad_norm": 2.6872658729553223, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.7041833400726318, + "num_tokens": 201197574.0, + "step": 7910 + }, + { + "epoch": 0.8687678453766747, + "grad_norm": 2.1049184799194336, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7138625383377075, + "num_tokens": 201223422.0, + "step": 7911 + }, + { + "epoch": 0.8688776630792884, + "grad_norm": 2.433807611465454, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7117369174957275, + "num_tokens": 201244051.0, + "step": 7912 + }, + { + "epoch": 0.868987480781902, + "grad_norm": 1.9708685874938965, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7006939649581909, + "num_tokens": 201274459.0, + "step": 7913 + }, + { + "epoch": 0.8690972984845157, + "grad_norm": 2.26556134223938, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7061160802841187, + "num_tokens": 201299245.0, + "step": 7914 + }, + { + "epoch": 0.8692071161871293, + "grad_norm": 2.3361032009124756, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7222738265991211, + "num_tokens": 201323785.0, + "step": 7915 + }, + { + "epoch": 0.8693169338897431, + "grad_norm": 1.9725120067596436, + "learning_rate": 1e-06, + "loss": 1.0754, + "mean_token_accuracy": 0.6739432215690613, + "num_tokens": 201354661.0, + "step": 7916 + }, + { + "epoch": 0.8694267515923567, + "grad_norm": 2.459146499633789, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.701621413230896, + "num_tokens": 201379631.0, + "step": 7917 + }, + { + "epoch": 0.8695365692949704, + "grad_norm": 1.9453566074371338, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.6983683705329895, + "num_tokens": 201410782.0, + "step": 7918 + }, + { + "epoch": 0.869646386997584, + "grad_norm": 2.116516351699829, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.7044380903244019, + "num_tokens": 201438612.0, + "step": 7919 + }, + { + "epoch": 0.8697562047001977, + "grad_norm": 2.159696340560913, + "learning_rate": 1e-06, + "loss": 1.0024, + "mean_token_accuracy": 0.6935163736343384, + "num_tokens": 201465178.0, + "step": 7920 + }, + { + "epoch": 0.8698660224028113, + "grad_norm": 2.0786638259887695, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7273482084274292, + "num_tokens": 201491652.0, + "step": 7921 + }, + { + "epoch": 0.869975840105425, + "grad_norm": 2.0706067085266113, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.704086184501648, + "num_tokens": 201518931.0, + "step": 7922 + }, + { + "epoch": 0.8700856578080387, + "grad_norm": 2.248918056488037, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7113308906555176, + "num_tokens": 201542156.0, + "step": 7923 + }, + { + "epoch": 0.8701954755106523, + "grad_norm": 2.3683271408081055, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7191436290740967, + "num_tokens": 201564095.0, + "step": 7924 + }, + { + "epoch": 0.870305293213266, + "grad_norm": 2.1377103328704834, + "learning_rate": 1e-06, + "loss": 1.0587, + "mean_token_accuracy": 0.6857776641845703, + "num_tokens": 201591051.0, + "step": 7925 + }, + { + "epoch": 0.8704151109158796, + "grad_norm": 2.345424175262451, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7268645763397217, + "num_tokens": 201614638.0, + "step": 7926 + }, + { + "epoch": 0.8705249286184933, + "grad_norm": 2.364429473876953, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.7027180194854736, + "num_tokens": 201638758.0, + "step": 7927 + }, + { + "epoch": 0.8706347463211069, + "grad_norm": 1.9927836656570435, + "learning_rate": 1e-06, + "loss": 1.0058, + "mean_token_accuracy": 0.6967881321907043, + "num_tokens": 201668224.0, + "step": 7928 + }, + { + "epoch": 0.8707445640237206, + "grad_norm": 2.320579767227173, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7060144543647766, + "num_tokens": 201689667.0, + "step": 7929 + }, + { + "epoch": 0.8708543817263343, + "grad_norm": 2.3551695346832275, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7199996113777161, + "num_tokens": 201712661.0, + "step": 7930 + }, + { + "epoch": 0.870964199428948, + "grad_norm": 2.0795297622680664, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7110684514045715, + "num_tokens": 201740055.0, + "step": 7931 + }, + { + "epoch": 0.8710740171315616, + "grad_norm": 2.4825127124786377, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7214961051940918, + "num_tokens": 201760071.0, + "step": 7932 + }, + { + "epoch": 0.8711838348341753, + "grad_norm": 1.9964113235473633, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7095177173614502, + "num_tokens": 201792891.0, + "step": 7933 + }, + { + "epoch": 0.8712936525367889, + "grad_norm": 2.402235507965088, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7267405986785889, + "num_tokens": 201813687.0, + "step": 7934 + }, + { + "epoch": 0.8714034702394026, + "grad_norm": 2.117229461669922, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7166317701339722, + "num_tokens": 201840810.0, + "step": 7935 + }, + { + "epoch": 0.8715132879420162, + "grad_norm": 2.0571231842041016, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6974002122879028, + "num_tokens": 201870921.0, + "step": 7936 + }, + { + "epoch": 0.87162310564463, + "grad_norm": 2.3522837162017822, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.715009331703186, + "num_tokens": 201893041.0, + "step": 7937 + }, + { + "epoch": 0.8717329233472436, + "grad_norm": 2.6285293102264404, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7124358415603638, + "num_tokens": 201911680.0, + "step": 7938 + }, + { + "epoch": 0.8718427410498573, + "grad_norm": 1.9844800233840942, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7161811590194702, + "num_tokens": 201941383.0, + "step": 7939 + }, + { + "epoch": 0.8719525587524709, + "grad_norm": 2.1386847496032715, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7223609089851379, + "num_tokens": 201966762.0, + "step": 7940 + }, + { + "epoch": 0.8720623764550846, + "grad_norm": 2.198460102081299, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6952167749404907, + "num_tokens": 201992594.0, + "step": 7941 + }, + { + "epoch": 0.8721721941576982, + "grad_norm": 2.2872886657714844, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6949031949043274, + "num_tokens": 202017873.0, + "step": 7942 + }, + { + "epoch": 0.8722820118603118, + "grad_norm": 2.3260741233825684, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7103897929191589, + "num_tokens": 202038903.0, + "step": 7943 + }, + { + "epoch": 0.8723918295629255, + "grad_norm": 2.0490245819091797, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7221937775611877, + "num_tokens": 202064162.0, + "step": 7944 + }, + { + "epoch": 0.8725016472655392, + "grad_norm": 2.0681753158569336, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.6977100968360901, + "num_tokens": 202093012.0, + "step": 7945 + }, + { + "epoch": 0.8726114649681529, + "grad_norm": 2.0480902194976807, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.698698103427887, + "num_tokens": 202122708.0, + "step": 7946 + }, + { + "epoch": 0.8727212826707665, + "grad_norm": 2.1073317527770996, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6870017051696777, + "num_tokens": 202150853.0, + "step": 7947 + }, + { + "epoch": 0.8728311003733802, + "grad_norm": 2.311075448989868, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7157318592071533, + "num_tokens": 202172862.0, + "step": 7948 + }, + { + "epoch": 0.8729409180759938, + "grad_norm": 2.1594464778900146, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7137541770935059, + "num_tokens": 202199219.0, + "step": 7949 + }, + { + "epoch": 0.8730507357786075, + "grad_norm": 2.1207327842712402, + "learning_rate": 1e-06, + "loss": 0.9954, + "mean_token_accuracy": 0.6995595693588257, + "num_tokens": 202228267.0, + "step": 7950 + }, + { + "epoch": 0.8731605534812211, + "grad_norm": 2.4145164489746094, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7133389711380005, + "num_tokens": 202249402.0, + "step": 7951 + }, + { + "epoch": 0.8732703711838349, + "grad_norm": 1.9916255474090576, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7044857740402222, + "num_tokens": 202278977.0, + "step": 7952 + }, + { + "epoch": 0.8733801888864485, + "grad_norm": 2.0054454803466797, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.6997833251953125, + "num_tokens": 202310196.0, + "step": 7953 + }, + { + "epoch": 0.8734900065890622, + "grad_norm": 2.278695821762085, + "learning_rate": 1e-06, + "loss": 1.0768, + "mean_token_accuracy": 0.693024754524231, + "num_tokens": 202334638.0, + "step": 7954 + }, + { + "epoch": 0.8735998242916758, + "grad_norm": 2.20333194732666, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7015572190284729, + "num_tokens": 202359533.0, + "step": 7955 + }, + { + "epoch": 0.8737096419942895, + "grad_norm": 2.028400182723999, + "learning_rate": 1e-06, + "loss": 1.0509, + "mean_token_accuracy": 0.6849532127380371, + "num_tokens": 202390263.0, + "step": 7956 + }, + { + "epoch": 0.8738194596969031, + "grad_norm": 2.3648345470428467, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.6921229362487793, + "num_tokens": 202414449.0, + "step": 7957 + }, + { + "epoch": 0.8739292773995168, + "grad_norm": 2.002937078475952, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7146191000938416, + "num_tokens": 202442377.0, + "step": 7958 + }, + { + "epoch": 0.8740390951021305, + "grad_norm": 1.910534381866455, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7092546224594116, + "num_tokens": 202473319.0, + "step": 7959 + }, + { + "epoch": 0.8741489128047442, + "grad_norm": 2.094655752182007, + "learning_rate": 1e-06, + "loss": 1.0851, + "mean_token_accuracy": 0.6776423454284668, + "num_tokens": 202503187.0, + "step": 7960 + }, + { + "epoch": 0.8742587305073578, + "grad_norm": 2.4239823818206787, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7171072959899902, + "num_tokens": 202524551.0, + "step": 7961 + }, + { + "epoch": 0.8743685482099715, + "grad_norm": 2.085024833679199, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.7092938423156738, + "num_tokens": 202552676.0, + "step": 7962 + }, + { + "epoch": 0.8744783659125851, + "grad_norm": 2.054905652999878, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7038550972938538, + "num_tokens": 202579540.0, + "step": 7963 + }, + { + "epoch": 0.8745881836151987, + "grad_norm": 2.6009395122528076, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7219109535217285, + "num_tokens": 202600326.0, + "step": 7964 + }, + { + "epoch": 0.8746980013178124, + "grad_norm": 2.164445638656616, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7241652011871338, + "num_tokens": 202624399.0, + "step": 7965 + }, + { + "epoch": 0.8748078190204261, + "grad_norm": 2.0809009075164795, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.713001549243927, + "num_tokens": 202651755.0, + "step": 7966 + }, + { + "epoch": 0.8749176367230398, + "grad_norm": 2.137953042984009, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7014685869216919, + "num_tokens": 202676659.0, + "step": 7967 + }, + { + "epoch": 0.8750274544256534, + "grad_norm": 1.9481514692306519, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.712926983833313, + "num_tokens": 202709452.0, + "step": 7968 + }, + { + "epoch": 0.8751372721282671, + "grad_norm": 2.3385281562805176, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7195408344268799, + "num_tokens": 202730605.0, + "step": 7969 + }, + { + "epoch": 0.8752470898308807, + "grad_norm": 2.4920055866241455, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7222596406936646, + "num_tokens": 202750529.0, + "step": 7970 + }, + { + "epoch": 0.8753569075334944, + "grad_norm": 2.1683809757232666, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7182574272155762, + "num_tokens": 202777544.0, + "step": 7971 + }, + { + "epoch": 0.875466725236108, + "grad_norm": 2.216038942337036, + "learning_rate": 1e-06, + "loss": 1.0525, + "mean_token_accuracy": 0.6963082551956177, + "num_tokens": 202805042.0, + "step": 7972 + }, + { + "epoch": 0.8755765429387217, + "grad_norm": 1.9879186153411865, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.705420732498169, + "num_tokens": 202835452.0, + "step": 7973 + }, + { + "epoch": 0.8756863606413354, + "grad_norm": 2.1764566898345947, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7021239995956421, + "num_tokens": 202863035.0, + "step": 7974 + }, + { + "epoch": 0.8757961783439491, + "grad_norm": 2.2772939205169678, + "learning_rate": 1e-06, + "loss": 1.0342, + "mean_token_accuracy": 0.6852855086326599, + "num_tokens": 202886642.0, + "step": 7975 + }, + { + "epoch": 0.8759059960465627, + "grad_norm": 2.038870334625244, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7228269577026367, + "num_tokens": 202914651.0, + "step": 7976 + }, + { + "epoch": 0.8760158137491764, + "grad_norm": 2.313093423843384, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7100380659103394, + "num_tokens": 202938769.0, + "step": 7977 + }, + { + "epoch": 0.87612563145179, + "grad_norm": 2.5573220252990723, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7118165493011475, + "num_tokens": 202957005.0, + "step": 7978 + }, + { + "epoch": 0.8762354491544037, + "grad_norm": 2.2747321128845215, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.6913785934448242, + "num_tokens": 202981054.0, + "step": 7979 + }, + { + "epoch": 0.8763452668570173, + "grad_norm": 2.309323310852051, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7197822332382202, + "num_tokens": 203007789.0, + "step": 7980 + }, + { + "epoch": 0.8764550845596311, + "grad_norm": 2.0320088863372803, + "learning_rate": 1e-06, + "loss": 1.0496, + "mean_token_accuracy": 0.6861251592636108, + "num_tokens": 203036749.0, + "step": 7981 + }, + { + "epoch": 0.8765649022622447, + "grad_norm": 2.1742193698883057, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7237042188644409, + "num_tokens": 203061166.0, + "step": 7982 + }, + { + "epoch": 0.8766747199648584, + "grad_norm": 2.052196502685547, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7110942602157593, + "num_tokens": 203090142.0, + "step": 7983 + }, + { + "epoch": 0.876784537667472, + "grad_norm": 2.299290180206299, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.6998820900917053, + "num_tokens": 203112853.0, + "step": 7984 + }, + { + "epoch": 0.8768943553700856, + "grad_norm": 2.0029613971710205, + "learning_rate": 1e-06, + "loss": 1.0291, + "mean_token_accuracy": 0.6924653649330139, + "num_tokens": 203143338.0, + "step": 7985 + }, + { + "epoch": 0.8770041730726993, + "grad_norm": 2.0776429176330566, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7071916460990906, + "num_tokens": 203168864.0, + "step": 7986 + }, + { + "epoch": 0.8771139907753129, + "grad_norm": 2.372684955596924, + "learning_rate": 1e-06, + "loss": 1.0669, + "mean_token_accuracy": 0.6912999153137207, + "num_tokens": 203192894.0, + "step": 7987 + }, + { + "epoch": 0.8772238084779267, + "grad_norm": 2.0192320346832275, + "learning_rate": 1e-06, + "loss": 1.0108, + "mean_token_accuracy": 0.6934518814086914, + "num_tokens": 203222941.0, + "step": 7988 + }, + { + "epoch": 0.8773336261805403, + "grad_norm": 1.9597444534301758, + "learning_rate": 1e-06, + "loss": 1.0535, + "mean_token_accuracy": 0.6831879615783691, + "num_tokens": 203254352.0, + "step": 7989 + }, + { + "epoch": 0.877443443883154, + "grad_norm": 2.3154850006103516, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7158259153366089, + "num_tokens": 203276381.0, + "step": 7990 + }, + { + "epoch": 0.8775532615857676, + "grad_norm": 2.279129981994629, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6975862979888916, + "num_tokens": 203299071.0, + "step": 7991 + }, + { + "epoch": 0.8776630792883813, + "grad_norm": 2.2385661602020264, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7113807201385498, + "num_tokens": 203324469.0, + "step": 7992 + }, + { + "epoch": 0.8777728969909949, + "grad_norm": 2.1143722534179688, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7270292043685913, + "num_tokens": 203348631.0, + "step": 7993 + }, + { + "epoch": 0.8778827146936086, + "grad_norm": 1.9341062307357788, + "learning_rate": 1e-06, + "loss": 1.0436, + "mean_token_accuracy": 0.6793509721755981, + "num_tokens": 203383553.0, + "step": 7994 + }, + { + "epoch": 0.8779925323962223, + "grad_norm": 2.2436773777008057, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6962589025497437, + "num_tokens": 203408616.0, + "step": 7995 + }, + { + "epoch": 0.878102350098836, + "grad_norm": 1.9387130737304688, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6899183392524719, + "num_tokens": 203442435.0, + "step": 7996 + }, + { + "epoch": 0.8782121678014496, + "grad_norm": 2.2052602767944336, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7163431644439697, + "num_tokens": 203466636.0, + "step": 7997 + }, + { + "epoch": 0.8783219855040633, + "grad_norm": 2.1408708095550537, + "learning_rate": 1e-06, + "loss": 1.0021, + "mean_token_accuracy": 0.6909144520759583, + "num_tokens": 203491511.0, + "step": 7998 + }, + { + "epoch": 0.8784318032066769, + "grad_norm": 2.0903725624084473, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6908824443817139, + "num_tokens": 203520660.0, + "step": 7999 + }, + { + "epoch": 0.8785416209092906, + "grad_norm": 2.393462657928467, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7225196361541748, + "num_tokens": 203540760.0, + "step": 8000 + }, + { + "epoch": 0.8786514386119042, + "grad_norm": 2.009030818939209, + "learning_rate": 1e-06, + "loss": 0.9664, + "mean_token_accuracy": 0.6983062028884888, + "num_tokens": 203570537.0, + "step": 8001 + }, + { + "epoch": 0.8787612563145178, + "grad_norm": 2.345224618911743, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7154457569122314, + "num_tokens": 203592417.0, + "step": 8002 + }, + { + "epoch": 0.8788710740171316, + "grad_norm": 2.6685447692871094, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7105801105499268, + "num_tokens": 203610516.0, + "step": 8003 + }, + { + "epoch": 0.8789808917197452, + "grad_norm": 2.05849552154541, + "learning_rate": 1e-06, + "loss": 1.0612, + "mean_token_accuracy": 0.6846481561660767, + "num_tokens": 203639940.0, + "step": 8004 + }, + { + "epoch": 0.8790907094223589, + "grad_norm": 2.147770643234253, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7217328548431396, + "num_tokens": 203664749.0, + "step": 8005 + }, + { + "epoch": 0.8792005271249725, + "grad_norm": 2.5844008922576904, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7354917526245117, + "num_tokens": 203682561.0, + "step": 8006 + }, + { + "epoch": 0.8793103448275862, + "grad_norm": 2.264359474182129, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6963544487953186, + "num_tokens": 203706479.0, + "step": 8007 + }, + { + "epoch": 0.8794201625301998, + "grad_norm": 2.1067187786102295, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7116883993148804, + "num_tokens": 203732849.0, + "step": 8008 + }, + { + "epoch": 0.8795299802328135, + "grad_norm": 2.1992242336273193, + "learning_rate": 1e-06, + "loss": 0.8243, + "mean_token_accuracy": 0.7421295642852783, + "num_tokens": 203755857.0, + "step": 8009 + }, + { + "epoch": 0.8796397979354272, + "grad_norm": 2.0731868743896484, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7108343839645386, + "num_tokens": 203781697.0, + "step": 8010 + }, + { + "epoch": 0.8797496156380409, + "grad_norm": 2.3004536628723145, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7329778671264648, + "num_tokens": 203804656.0, + "step": 8011 + }, + { + "epoch": 0.8798594333406545, + "grad_norm": 2.180663585662842, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7228581309318542, + "num_tokens": 203831561.0, + "step": 8012 + }, + { + "epoch": 0.8799692510432682, + "grad_norm": 2.2274835109710693, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7081704139709473, + "num_tokens": 203855634.0, + "step": 8013 + }, + { + "epoch": 0.8800790687458818, + "grad_norm": 2.138105630874634, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7094758749008179, + "num_tokens": 203883881.0, + "step": 8014 + }, + { + "epoch": 0.8801888864484955, + "grad_norm": 2.229250192642212, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.7014307379722595, + "num_tokens": 203908546.0, + "step": 8015 + }, + { + "epoch": 0.8802987041511091, + "grad_norm": 2.026939630508423, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7032802104949951, + "num_tokens": 203937205.0, + "step": 8016 + }, + { + "epoch": 0.8804085218537229, + "grad_norm": 2.554067373275757, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.717106819152832, + "num_tokens": 203958215.0, + "step": 8017 + }, + { + "epoch": 0.8805183395563365, + "grad_norm": 2.162102699279785, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6926465034484863, + "num_tokens": 203983116.0, + "step": 8018 + }, + { + "epoch": 0.8806281572589502, + "grad_norm": 2.214621067047119, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7061858177185059, + "num_tokens": 204008498.0, + "step": 8019 + }, + { + "epoch": 0.8807379749615638, + "grad_norm": 2.0546061992645264, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7120972275733948, + "num_tokens": 204037624.0, + "step": 8020 + }, + { + "epoch": 0.8808477926641775, + "grad_norm": 2.1779379844665527, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7097976803779602, + "num_tokens": 204060970.0, + "step": 8021 + }, + { + "epoch": 0.8809576103667911, + "grad_norm": 2.452275514602661, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7339775562286377, + "num_tokens": 204081418.0, + "step": 8022 + }, + { + "epoch": 0.8810674280694047, + "grad_norm": 2.1343953609466553, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7082560062408447, + "num_tokens": 204108940.0, + "step": 8023 + }, + { + "epoch": 0.8811772457720185, + "grad_norm": 2.0122857093811035, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7004505395889282, + "num_tokens": 204138566.0, + "step": 8024 + }, + { + "epoch": 0.8812870634746321, + "grad_norm": 2.0694870948791504, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7153716087341309, + "num_tokens": 204167296.0, + "step": 8025 + }, + { + "epoch": 0.8813968811772458, + "grad_norm": 2.2529847621917725, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.713714063167572, + "num_tokens": 204191518.0, + "step": 8026 + }, + { + "epoch": 0.8815066988798594, + "grad_norm": 1.8818120956420898, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.6979221105575562, + "num_tokens": 204221538.0, + "step": 8027 + }, + { + "epoch": 0.8816165165824731, + "grad_norm": 2.0727052688598633, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7203642129898071, + "num_tokens": 204247734.0, + "step": 8028 + }, + { + "epoch": 0.8817263342850867, + "grad_norm": 2.217862129211426, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7112957835197449, + "num_tokens": 204272072.0, + "step": 8029 + }, + { + "epoch": 0.8818361519877004, + "grad_norm": 1.966461181640625, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7158231139183044, + "num_tokens": 204301613.0, + "step": 8030 + }, + { + "epoch": 0.881945969690314, + "grad_norm": 2.285567283630371, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7205994725227356, + "num_tokens": 204323573.0, + "step": 8031 + }, + { + "epoch": 0.8820557873929278, + "grad_norm": 2.720566987991333, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7204287648200989, + "num_tokens": 204342022.0, + "step": 8032 + }, + { + "epoch": 0.8821656050955414, + "grad_norm": 2.1197497844696045, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7195720076560974, + "num_tokens": 204368397.0, + "step": 8033 + }, + { + "epoch": 0.8822754227981551, + "grad_norm": 2.29695463180542, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7209839224815369, + "num_tokens": 204391579.0, + "step": 8034 + }, + { + "epoch": 0.8823852405007687, + "grad_norm": 2.203519105911255, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7175416946411133, + "num_tokens": 204415131.0, + "step": 8035 + }, + { + "epoch": 0.8824950582033824, + "grad_norm": 2.3918991088867188, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.7033249735832214, + "num_tokens": 204436031.0, + "step": 8036 + }, + { + "epoch": 0.882604875905996, + "grad_norm": 1.9029219150543213, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7069591283798218, + "num_tokens": 204467436.0, + "step": 8037 + }, + { + "epoch": 0.8827146936086097, + "grad_norm": 2.2720916271209717, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7059112191200256, + "num_tokens": 204489912.0, + "step": 8038 + }, + { + "epoch": 0.8828245113112234, + "grad_norm": 2.0193238258361816, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7031450867652893, + "num_tokens": 204519600.0, + "step": 8039 + }, + { + "epoch": 0.8829343290138371, + "grad_norm": 2.099741220474243, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7120567560195923, + "num_tokens": 204547026.0, + "step": 8040 + }, + { + "epoch": 0.8830441467164507, + "grad_norm": 2.027782917022705, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7131099700927734, + "num_tokens": 204574490.0, + "step": 8041 + }, + { + "epoch": 0.8831539644190644, + "grad_norm": 2.091792583465576, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6915658712387085, + "num_tokens": 204603123.0, + "step": 8042 + }, + { + "epoch": 0.883263782121678, + "grad_norm": 3.4582431316375732, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7227364778518677, + "num_tokens": 204624772.0, + "step": 8043 + }, + { + "epoch": 0.8833735998242916, + "grad_norm": 2.0830533504486084, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7231466174125671, + "num_tokens": 204650261.0, + "step": 8044 + }, + { + "epoch": 0.8834834175269053, + "grad_norm": 2.293827772140503, + "learning_rate": 1e-06, + "loss": 1.0149, + "mean_token_accuracy": 0.6945882439613342, + "num_tokens": 204674240.0, + "step": 8045 + }, + { + "epoch": 0.883593235229519, + "grad_norm": 2.394845962524414, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.7092591524124146, + "num_tokens": 204698042.0, + "step": 8046 + }, + { + "epoch": 0.8837030529321327, + "grad_norm": 2.118441104888916, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7029165029525757, + "num_tokens": 204722611.0, + "step": 8047 + }, + { + "epoch": 0.8838128706347463, + "grad_norm": 1.9434313774108887, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7238341569900513, + "num_tokens": 204752702.0, + "step": 8048 + }, + { + "epoch": 0.88392268833736, + "grad_norm": 2.3441214561462402, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7015141248703003, + "num_tokens": 204776885.0, + "step": 8049 + }, + { + "epoch": 0.8840325060399736, + "grad_norm": 2.0922744274139404, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7093864679336548, + "num_tokens": 204803203.0, + "step": 8050 + }, + { + "epoch": 0.8841423237425873, + "grad_norm": 2.44751238822937, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7429369688034058, + "num_tokens": 204822623.0, + "step": 8051 + }, + { + "epoch": 0.8842521414452009, + "grad_norm": 2.079862356185913, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7192221879959106, + "num_tokens": 204849024.0, + "step": 8052 + }, + { + "epoch": 0.8843619591478147, + "grad_norm": 2.194812536239624, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.7004050016403198, + "num_tokens": 204874886.0, + "step": 8053 + }, + { + "epoch": 0.8844717768504283, + "grad_norm": 2.2856082916259766, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7101261615753174, + "num_tokens": 204899840.0, + "step": 8054 + }, + { + "epoch": 0.884581594553042, + "grad_norm": 2.3059420585632324, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7046513557434082, + "num_tokens": 204922964.0, + "step": 8055 + }, + { + "epoch": 0.8846914122556556, + "grad_norm": 2.213907241821289, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.717284083366394, + "num_tokens": 204947504.0, + "step": 8056 + }, + { + "epoch": 0.8848012299582693, + "grad_norm": 2.170642375946045, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.7043647766113281, + "num_tokens": 204973104.0, + "step": 8057 + }, + { + "epoch": 0.8849110476608829, + "grad_norm": 1.789757251739502, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.6986508369445801, + "num_tokens": 205009291.0, + "step": 8058 + }, + { + "epoch": 0.8850208653634966, + "grad_norm": 2.2371747493743896, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7077076435089111, + "num_tokens": 205033818.0, + "step": 8059 + }, + { + "epoch": 0.8851306830661102, + "grad_norm": 1.9715542793273926, + "learning_rate": 1e-06, + "loss": 1.021, + "mean_token_accuracy": 0.690396785736084, + "num_tokens": 205064527.0, + "step": 8060 + }, + { + "epoch": 0.885240500768724, + "grad_norm": 2.3123443126678467, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.6931382417678833, + "num_tokens": 205089433.0, + "step": 8061 + }, + { + "epoch": 0.8853503184713376, + "grad_norm": 2.11985445022583, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7068191766738892, + "num_tokens": 205116260.0, + "step": 8062 + }, + { + "epoch": 0.8854601361739513, + "grad_norm": 1.9305888414382935, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7144396901130676, + "num_tokens": 205145132.0, + "step": 8063 + }, + { + "epoch": 0.8855699538765649, + "grad_norm": 2.1683688163757324, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7126600742340088, + "num_tokens": 205171285.0, + "step": 8064 + }, + { + "epoch": 0.8856797715791785, + "grad_norm": 2.163140058517456, + "learning_rate": 1e-06, + "loss": 1.0267, + "mean_token_accuracy": 0.6852185726165771, + "num_tokens": 205198025.0, + "step": 8065 + }, + { + "epoch": 0.8857895892817922, + "grad_norm": 2.0982439517974854, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7108545303344727, + "num_tokens": 205224570.0, + "step": 8066 + }, + { + "epoch": 0.8858994069844058, + "grad_norm": 2.044567346572876, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7011487483978271, + "num_tokens": 205252715.0, + "step": 8067 + }, + { + "epoch": 0.8860092246870196, + "grad_norm": 2.245399236679077, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7348483204841614, + "num_tokens": 205276443.0, + "step": 8068 + }, + { + "epoch": 0.8861190423896332, + "grad_norm": 2.8745439052581787, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.6918331980705261, + "num_tokens": 205294782.0, + "step": 8069 + }, + { + "epoch": 0.8862288600922469, + "grad_norm": 2.3128960132598877, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.732558012008667, + "num_tokens": 205316683.0, + "step": 8070 + }, + { + "epoch": 0.8863386777948605, + "grad_norm": 2.2211334705352783, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7363611459732056, + "num_tokens": 205343243.0, + "step": 8071 + }, + { + "epoch": 0.8864484954974742, + "grad_norm": 2.4869353771209717, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7108574509620667, + "num_tokens": 205363853.0, + "step": 8072 + }, + { + "epoch": 0.8865583132000878, + "grad_norm": 2.088395118713379, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7212629318237305, + "num_tokens": 205391327.0, + "step": 8073 + }, + { + "epoch": 0.8866681309027015, + "grad_norm": 2.1634557247161865, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.733102560043335, + "num_tokens": 205418180.0, + "step": 8074 + }, + { + "epoch": 0.8867779486053152, + "grad_norm": 2.1245853900909424, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7028800249099731, + "num_tokens": 205445186.0, + "step": 8075 + }, + { + "epoch": 0.8868877663079289, + "grad_norm": 2.087496519088745, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7274125814437866, + "num_tokens": 205470357.0, + "step": 8076 + }, + { + "epoch": 0.8869975840105425, + "grad_norm": 2.3121144771575928, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7014322280883789, + "num_tokens": 205494567.0, + "step": 8077 + }, + { + "epoch": 0.8871074017131562, + "grad_norm": 1.9804579019546509, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6950672268867493, + "num_tokens": 205525342.0, + "step": 8078 + }, + { + "epoch": 0.8872172194157698, + "grad_norm": 2.4365038871765137, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7080782651901245, + "num_tokens": 205546607.0, + "step": 8079 + }, + { + "epoch": 0.8873270371183835, + "grad_norm": 2.384166955947876, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.700151801109314, + "num_tokens": 205567056.0, + "step": 8080 + }, + { + "epoch": 0.8874368548209971, + "grad_norm": 2.354203701019287, + "learning_rate": 1e-06, + "loss": 1.0814, + "mean_token_accuracy": 0.6733311414718628, + "num_tokens": 205588854.0, + "step": 8081 + }, + { + "epoch": 0.8875466725236109, + "grad_norm": 2.4982500076293945, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7121131420135498, + "num_tokens": 205609129.0, + "step": 8082 + }, + { + "epoch": 0.8876564902262245, + "grad_norm": 2.329345703125, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7075129747390747, + "num_tokens": 205631764.0, + "step": 8083 + }, + { + "epoch": 0.8877663079288381, + "grad_norm": 2.126310110092163, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7052962779998779, + "num_tokens": 205658523.0, + "step": 8084 + }, + { + "epoch": 0.8878761256314518, + "grad_norm": 2.2658064365386963, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7082990407943726, + "num_tokens": 205681926.0, + "step": 8085 + }, + { + "epoch": 0.8879859433340654, + "grad_norm": 2.206037998199463, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7108796834945679, + "num_tokens": 205707624.0, + "step": 8086 + }, + { + "epoch": 0.8880957610366791, + "grad_norm": 2.089886426925659, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6929703950881958, + "num_tokens": 205736084.0, + "step": 8087 + }, + { + "epoch": 0.8882055787392927, + "grad_norm": 2.4014370441436768, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.7124930024147034, + "num_tokens": 205760445.0, + "step": 8088 + }, + { + "epoch": 0.8883153964419065, + "grad_norm": 2.4411678314208984, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7171421647071838, + "num_tokens": 205781350.0, + "step": 8089 + }, + { + "epoch": 0.8884252141445201, + "grad_norm": 1.9919379949569702, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.698470950126648, + "num_tokens": 205812907.0, + "step": 8090 + }, + { + "epoch": 0.8885350318471338, + "grad_norm": 1.7955596446990967, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.6851614117622375, + "num_tokens": 205848380.0, + "step": 8091 + }, + { + "epoch": 0.8886448495497474, + "grad_norm": 2.1385204792022705, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7018568515777588, + "num_tokens": 205874779.0, + "step": 8092 + }, + { + "epoch": 0.8887546672523611, + "grad_norm": 2.1571013927459717, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7011298537254333, + "num_tokens": 205902705.0, + "step": 8093 + }, + { + "epoch": 0.8888644849549747, + "grad_norm": 1.9176568984985352, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7134242057800293, + "num_tokens": 205933628.0, + "step": 8094 + }, + { + "epoch": 0.8889743026575884, + "grad_norm": 2.039224863052368, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7037205100059509, + "num_tokens": 205962641.0, + "step": 8095 + }, + { + "epoch": 0.889084120360202, + "grad_norm": 2.3598310947418213, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7134956121444702, + "num_tokens": 205985950.0, + "step": 8096 + }, + { + "epoch": 0.8891939380628158, + "grad_norm": 2.124767303466797, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7187793254852295, + "num_tokens": 206012180.0, + "step": 8097 + }, + { + "epoch": 0.8893037557654294, + "grad_norm": 2.180864095687866, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.6913836002349854, + "num_tokens": 206038589.0, + "step": 8098 + }, + { + "epoch": 0.8894135734680431, + "grad_norm": 2.187594413757324, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7500579357147217, + "num_tokens": 206061536.0, + "step": 8099 + }, + { + "epoch": 0.8895233911706567, + "grad_norm": 2.005967855453491, + "learning_rate": 1e-06, + "loss": 1.0211, + "mean_token_accuracy": 0.68351811170578, + "num_tokens": 206095887.0, + "step": 8100 + }, + { + "epoch": 0.8896332088732704, + "grad_norm": 2.030780553817749, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.7016557455062866, + "num_tokens": 206123765.0, + "step": 8101 + }, + { + "epoch": 0.889743026575884, + "grad_norm": 2.3612852096557617, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.6947581768035889, + "num_tokens": 206147565.0, + "step": 8102 + }, + { + "epoch": 0.8898528442784976, + "grad_norm": 1.908072829246521, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6934897899627686, + "num_tokens": 206180933.0, + "step": 8103 + }, + { + "epoch": 0.8899626619811114, + "grad_norm": 2.2360212802886963, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7031440734863281, + "num_tokens": 206205409.0, + "step": 8104 + }, + { + "epoch": 0.890072479683725, + "grad_norm": 2.0674808025360107, + "learning_rate": 1e-06, + "loss": 1.0802, + "mean_token_accuracy": 0.6760017275810242, + "num_tokens": 206236149.0, + "step": 8105 + }, + { + "epoch": 0.8901822973863387, + "grad_norm": 1.9642494916915894, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.6977342367172241, + "num_tokens": 206264582.0, + "step": 8106 + }, + { + "epoch": 0.8902921150889523, + "grad_norm": 2.1979939937591553, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7132636904716492, + "num_tokens": 206289894.0, + "step": 8107 + }, + { + "epoch": 0.890401932791566, + "grad_norm": 2.5781733989715576, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7236460447311401, + "num_tokens": 206308241.0, + "step": 8108 + }, + { + "epoch": 0.8905117504941796, + "grad_norm": 1.9470733404159546, + "learning_rate": 1e-06, + "loss": 1.0484, + "mean_token_accuracy": 0.690814733505249, + "num_tokens": 206338798.0, + "step": 8109 + }, + { + "epoch": 0.8906215681967933, + "grad_norm": 2.1385247707366943, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7075793147087097, + "num_tokens": 206365157.0, + "step": 8110 + }, + { + "epoch": 0.890731385899407, + "grad_norm": 2.336648464202881, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7035270929336548, + "num_tokens": 206390338.0, + "step": 8111 + }, + { + "epoch": 0.8908412036020207, + "grad_norm": 2.0747954845428467, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7256627082824707, + "num_tokens": 206416729.0, + "step": 8112 + }, + { + "epoch": 0.8909510213046343, + "grad_norm": 2.5717902183532715, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7243430018424988, + "num_tokens": 206435355.0, + "step": 8113 + }, + { + "epoch": 0.891060839007248, + "grad_norm": 2.5931203365325928, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7171719074249268, + "num_tokens": 206454497.0, + "step": 8114 + }, + { + "epoch": 0.8911706567098616, + "grad_norm": 1.8498286008834839, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.6963106393814087, + "num_tokens": 206486615.0, + "step": 8115 + }, + { + "epoch": 0.8912804744124753, + "grad_norm": 1.958516001701355, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7332620620727539, + "num_tokens": 206514286.0, + "step": 8116 + }, + { + "epoch": 0.8913902921150889, + "grad_norm": 2.3133490085601807, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7300766110420227, + "num_tokens": 206536073.0, + "step": 8117 + }, + { + "epoch": 0.8915001098177027, + "grad_norm": 2.253648042678833, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.6964240074157715, + "num_tokens": 206559882.0, + "step": 8118 + }, + { + "epoch": 0.8916099275203163, + "grad_norm": 2.187991142272949, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7169369459152222, + "num_tokens": 206586197.0, + "step": 8119 + }, + { + "epoch": 0.89171974522293, + "grad_norm": 2.1962053775787354, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7093818783760071, + "num_tokens": 206612308.0, + "step": 8120 + }, + { + "epoch": 0.8918295629255436, + "grad_norm": 2.033907651901245, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.7015502452850342, + "num_tokens": 206641593.0, + "step": 8121 + }, + { + "epoch": 0.8919393806281573, + "grad_norm": 2.065478801727295, + "learning_rate": 1e-06, + "loss": 1.0119, + "mean_token_accuracy": 0.7044185996055603, + "num_tokens": 206669070.0, + "step": 8122 + }, + { + "epoch": 0.8920491983307709, + "grad_norm": 2.2070934772491455, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.7000125646591187, + "num_tokens": 206693024.0, + "step": 8123 + }, + { + "epoch": 0.8921590160333845, + "grad_norm": 2.5417377948760986, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7135651111602783, + "num_tokens": 206711187.0, + "step": 8124 + }, + { + "epoch": 0.8922688337359982, + "grad_norm": 2.1232802867889404, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7276712656021118, + "num_tokens": 206737184.0, + "step": 8125 + }, + { + "epoch": 0.892378651438612, + "grad_norm": 2.288027286529541, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7129883766174316, + "num_tokens": 206759226.0, + "step": 8126 + }, + { + "epoch": 0.8924884691412256, + "grad_norm": 2.2518746852874756, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7203682661056519, + "num_tokens": 206781431.0, + "step": 8127 + }, + { + "epoch": 0.8925982868438392, + "grad_norm": 2.3984291553497314, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7187683582305908, + "num_tokens": 206802894.0, + "step": 8128 + }, + { + "epoch": 0.8927081045464529, + "grad_norm": 2.5311830043792725, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7405641674995422, + "num_tokens": 206822408.0, + "step": 8129 + }, + { + "epoch": 0.8928179222490665, + "grad_norm": 2.045172691345215, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7127408981323242, + "num_tokens": 206849478.0, + "step": 8130 + }, + { + "epoch": 0.8929277399516802, + "grad_norm": 2.629899501800537, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6957864761352539, + "num_tokens": 206869177.0, + "step": 8131 + }, + { + "epoch": 0.8930375576542938, + "grad_norm": 1.7364763021469116, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.7035910487174988, + "num_tokens": 206908124.0, + "step": 8132 + }, + { + "epoch": 0.8931473753569076, + "grad_norm": 2.3729374408721924, + "learning_rate": 1e-06, + "loss": 1.097, + "mean_token_accuracy": 0.672858715057373, + "num_tokens": 206934488.0, + "step": 8133 + }, + { + "epoch": 0.8932571930595212, + "grad_norm": 2.059295177459717, + "learning_rate": 1e-06, + "loss": 1.0456, + "mean_token_accuracy": 0.6833446621894836, + "num_tokens": 206963718.0, + "step": 8134 + }, + { + "epoch": 0.8933670107621349, + "grad_norm": 2.0833492279052734, + "learning_rate": 1e-06, + "loss": 1.0409, + "mean_token_accuracy": 0.6836016774177551, + "num_tokens": 206991595.0, + "step": 8135 + }, + { + "epoch": 0.8934768284647485, + "grad_norm": 2.236302375793457, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7187313437461853, + "num_tokens": 207016901.0, + "step": 8136 + }, + { + "epoch": 0.8935866461673622, + "grad_norm": 2.0807507038116455, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7094488143920898, + "num_tokens": 207044985.0, + "step": 8137 + }, + { + "epoch": 0.8936964638699758, + "grad_norm": 1.963415503501892, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7186655402183533, + "num_tokens": 207074655.0, + "step": 8138 + }, + { + "epoch": 0.8938062815725895, + "grad_norm": 2.0863735675811768, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7036066651344299, + "num_tokens": 207102624.0, + "step": 8139 + }, + { + "epoch": 0.8939160992752032, + "grad_norm": 2.38161563873291, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7108389139175415, + "num_tokens": 207125677.0, + "step": 8140 + }, + { + "epoch": 0.8940259169778169, + "grad_norm": 2.264221668243408, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.6990101337432861, + "num_tokens": 207149195.0, + "step": 8141 + }, + { + "epoch": 0.8941357346804305, + "grad_norm": 2.4305436611175537, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7085248231887817, + "num_tokens": 207169749.0, + "step": 8142 + }, + { + "epoch": 0.8942455523830442, + "grad_norm": 2.404611825942993, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7076253890991211, + "num_tokens": 207192304.0, + "step": 8143 + }, + { + "epoch": 0.8943553700856578, + "grad_norm": 1.8377028703689575, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6914495825767517, + "num_tokens": 207228581.0, + "step": 8144 + }, + { + "epoch": 0.8944651877882714, + "grad_norm": 1.8760067224502563, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6959722638130188, + "num_tokens": 207261013.0, + "step": 8145 + }, + { + "epoch": 0.8945750054908851, + "grad_norm": 1.954055905342102, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7161906957626343, + "num_tokens": 207289749.0, + "step": 8146 + }, + { + "epoch": 0.8946848231934988, + "grad_norm": 2.2047576904296875, + "learning_rate": 1e-06, + "loss": 1.0731, + "mean_token_accuracy": 0.681755781173706, + "num_tokens": 207318474.0, + "step": 8147 + }, + { + "epoch": 0.8947946408961125, + "grad_norm": 2.4441959857940674, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7136040925979614, + "num_tokens": 207339457.0, + "step": 8148 + }, + { + "epoch": 0.8949044585987261, + "grad_norm": 2.6657180786132812, + "learning_rate": 1e-06, + "loss": 0.9512, + "mean_token_accuracy": 0.7005119919776917, + "num_tokens": 207357582.0, + "step": 8149 + }, + { + "epoch": 0.8950142763013398, + "grad_norm": 2.2815723419189453, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.7044275403022766, + "num_tokens": 207381320.0, + "step": 8150 + }, + { + "epoch": 0.8951240940039534, + "grad_norm": 2.0558674335479736, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7056725025177002, + "num_tokens": 207408896.0, + "step": 8151 + }, + { + "epoch": 0.8952339117065671, + "grad_norm": 2.1642048358917236, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7085311412811279, + "num_tokens": 207436136.0, + "step": 8152 + }, + { + "epoch": 0.8953437294091807, + "grad_norm": 2.091731071472168, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7110542058944702, + "num_tokens": 207464835.0, + "step": 8153 + }, + { + "epoch": 0.8954535471117944, + "grad_norm": 2.0883889198303223, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7108160853385925, + "num_tokens": 207491498.0, + "step": 8154 + }, + { + "epoch": 0.8955633648144081, + "grad_norm": 2.07177734375, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7111814022064209, + "num_tokens": 207516914.0, + "step": 8155 + }, + { + "epoch": 0.8956731825170218, + "grad_norm": 2.096909761428833, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7240932583808899, + "num_tokens": 207541222.0, + "step": 8156 + }, + { + "epoch": 0.8957830002196354, + "grad_norm": 2.1150217056274414, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7128100395202637, + "num_tokens": 207567187.0, + "step": 8157 + }, + { + "epoch": 0.8958928179222491, + "grad_norm": 2.109989643096924, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7189965844154358, + "num_tokens": 207591311.0, + "step": 8158 + }, + { + "epoch": 0.8960026356248627, + "grad_norm": 2.342384099960327, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7170465588569641, + "num_tokens": 207615141.0, + "step": 8159 + }, + { + "epoch": 0.8961124533274764, + "grad_norm": 2.1299874782562256, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.6991112232208252, + "num_tokens": 207641620.0, + "step": 8160 + }, + { + "epoch": 0.89622227103009, + "grad_norm": 2.3581156730651855, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.7033070921897888, + "num_tokens": 207662659.0, + "step": 8161 + }, + { + "epoch": 0.8963320887327038, + "grad_norm": 2.0809545516967773, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7064843773841858, + "num_tokens": 207689627.0, + "step": 8162 + }, + { + "epoch": 0.8964419064353174, + "grad_norm": 2.5831856727600098, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7211482524871826, + "num_tokens": 207708415.0, + "step": 8163 + }, + { + "epoch": 0.896551724137931, + "grad_norm": 2.2839298248291016, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7102242112159729, + "num_tokens": 207730312.0, + "step": 8164 + }, + { + "epoch": 0.8966615418405447, + "grad_norm": 2.0862860679626465, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6942508816719055, + "num_tokens": 207757038.0, + "step": 8165 + }, + { + "epoch": 0.8967713595431583, + "grad_norm": 2.2199833393096924, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7259416580200195, + "num_tokens": 207782661.0, + "step": 8166 + }, + { + "epoch": 0.896881177245772, + "grad_norm": 2.021589756011963, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7112300395965576, + "num_tokens": 207810918.0, + "step": 8167 + }, + { + "epoch": 0.8969909949483856, + "grad_norm": 1.9270685911178589, + "learning_rate": 1e-06, + "loss": 0.9883, + "mean_token_accuracy": 0.7026090621948242, + "num_tokens": 207842808.0, + "step": 8168 + }, + { + "epoch": 0.8971008126509994, + "grad_norm": 2.216949939727783, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7113189697265625, + "num_tokens": 207866237.0, + "step": 8169 + }, + { + "epoch": 0.897210630353613, + "grad_norm": 2.0964577198028564, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7141358852386475, + "num_tokens": 207891088.0, + "step": 8170 + }, + { + "epoch": 0.8973204480562267, + "grad_norm": 2.10262393951416, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6888148188591003, + "num_tokens": 207921040.0, + "step": 8171 + }, + { + "epoch": 0.8974302657588403, + "grad_norm": 2.232023239135742, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7315255403518677, + "num_tokens": 207944059.0, + "step": 8172 + }, + { + "epoch": 0.897540083461454, + "grad_norm": 2.4500601291656494, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.6994473338127136, + "num_tokens": 207964905.0, + "step": 8173 + }, + { + "epoch": 0.8976499011640676, + "grad_norm": 2.219254970550537, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7171570062637329, + "num_tokens": 207989667.0, + "step": 8174 + }, + { + "epoch": 0.8977597188666813, + "grad_norm": 2.1615681648254395, + "learning_rate": 1e-06, + "loss": 1.0279, + "mean_token_accuracy": 0.6882426738739014, + "num_tokens": 208016269.0, + "step": 8175 + }, + { + "epoch": 0.897869536569295, + "grad_norm": 2.296043872833252, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6990535259246826, + "num_tokens": 208039347.0, + "step": 8176 + }, + { + "epoch": 0.8979793542719087, + "grad_norm": 2.0679962635040283, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7074391841888428, + "num_tokens": 208065741.0, + "step": 8177 + }, + { + "epoch": 0.8980891719745223, + "grad_norm": 2.2146360874176025, + "learning_rate": 1e-06, + "loss": 1.0301, + "mean_token_accuracy": 0.6889817714691162, + "num_tokens": 208092711.0, + "step": 8178 + }, + { + "epoch": 0.898198989677136, + "grad_norm": 2.4165635108947754, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7088598012924194, + "num_tokens": 208113382.0, + "step": 8179 + }, + { + "epoch": 0.8983088073797496, + "grad_norm": 2.1001949310302734, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7045288681983948, + "num_tokens": 208140636.0, + "step": 8180 + }, + { + "epoch": 0.8984186250823633, + "grad_norm": 2.3702216148376465, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7369144558906555, + "num_tokens": 208163398.0, + "step": 8181 + }, + { + "epoch": 0.8985284427849769, + "grad_norm": 2.266798257827759, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.6982318162918091, + "num_tokens": 208188792.0, + "step": 8182 + }, + { + "epoch": 0.8986382604875905, + "grad_norm": 2.0672109127044678, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6928336024284363, + "num_tokens": 208217729.0, + "step": 8183 + }, + { + "epoch": 0.8987480781902043, + "grad_norm": 2.011063814163208, + "learning_rate": 1e-06, + "loss": 1.0574, + "mean_token_accuracy": 0.6902338266372681, + "num_tokens": 208248448.0, + "step": 8184 + }, + { + "epoch": 0.898857895892818, + "grad_norm": 2.406331777572632, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.7051540017127991, + "num_tokens": 208271487.0, + "step": 8185 + }, + { + "epoch": 0.8989677135954316, + "grad_norm": 1.990477204322815, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.688089907169342, + "num_tokens": 208305703.0, + "step": 8186 + }, + { + "epoch": 0.8990775312980452, + "grad_norm": 2.432880401611328, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7082329988479614, + "num_tokens": 208327784.0, + "step": 8187 + }, + { + "epoch": 0.8991873490006589, + "grad_norm": 2.038529872894287, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.7005667686462402, + "num_tokens": 208357900.0, + "step": 8188 + }, + { + "epoch": 0.8992971667032725, + "grad_norm": 2.1517679691314697, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.7010698914527893, + "num_tokens": 208386228.0, + "step": 8189 + }, + { + "epoch": 0.8994069844058862, + "grad_norm": 2.1681785583496094, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6988040208816528, + "num_tokens": 208414037.0, + "step": 8190 + }, + { + "epoch": 0.8995168021084999, + "grad_norm": 2.247382402420044, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7282566428184509, + "num_tokens": 208437386.0, + "step": 8191 + }, + { + "epoch": 0.8996266198111136, + "grad_norm": 2.1036760807037354, + "learning_rate": 1e-06, + "loss": 1.0247, + "mean_token_accuracy": 0.6935443878173828, + "num_tokens": 208465270.0, + "step": 8192 + }, + { + "epoch": 0.8997364375137272, + "grad_norm": 2.141651153564453, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7380326986312866, + "num_tokens": 208490288.0, + "step": 8193 + }, + { + "epoch": 0.8998462552163409, + "grad_norm": 2.106189727783203, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.6973991394042969, + "num_tokens": 208518635.0, + "step": 8194 + }, + { + "epoch": 0.8999560729189545, + "grad_norm": 2.111633777618408, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7086558938026428, + "num_tokens": 208545507.0, + "step": 8195 + }, + { + "epoch": 0.9000658906215682, + "grad_norm": 2.3019895553588867, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6969695091247559, + "num_tokens": 208572520.0, + "step": 8196 + }, + { + "epoch": 0.9001757083241818, + "grad_norm": 2.3900146484375, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.7147713303565979, + "num_tokens": 208599445.0, + "step": 8197 + }, + { + "epoch": 0.9002855260267956, + "grad_norm": 2.2775917053222656, + "learning_rate": 1e-06, + "loss": 1.0257, + "mean_token_accuracy": 0.6917039155960083, + "num_tokens": 208624144.0, + "step": 8198 + }, + { + "epoch": 0.9003953437294092, + "grad_norm": 1.932207465171814, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7083002924919128, + "num_tokens": 208653932.0, + "step": 8199 + }, + { + "epoch": 0.9005051614320229, + "grad_norm": 2.115816116333008, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6958678960800171, + "num_tokens": 208681383.0, + "step": 8200 + }, + { + "epoch": 0.9006149791346365, + "grad_norm": 2.272416114807129, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.699724555015564, + "num_tokens": 208705634.0, + "step": 8201 + }, + { + "epoch": 0.9007247968372502, + "grad_norm": 2.304816961288452, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.710439920425415, + "num_tokens": 208729777.0, + "step": 8202 + }, + { + "epoch": 0.9008346145398638, + "grad_norm": 2.0998384952545166, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6938155293464661, + "num_tokens": 208757890.0, + "step": 8203 + }, + { + "epoch": 0.9009444322424774, + "grad_norm": 2.07070255279541, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7137497663497925, + "num_tokens": 208785901.0, + "step": 8204 + }, + { + "epoch": 0.9010542499450912, + "grad_norm": 2.428534507751465, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7328318953514099, + "num_tokens": 208805360.0, + "step": 8205 + }, + { + "epoch": 0.9011640676477048, + "grad_norm": 2.13000750541687, + "learning_rate": 1e-06, + "loss": 0.9694, + "mean_token_accuracy": 0.7042065262794495, + "num_tokens": 208830159.0, + "step": 8206 + }, + { + "epoch": 0.9012738853503185, + "grad_norm": 2.110461711883545, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7256778478622437, + "num_tokens": 208857655.0, + "step": 8207 + }, + { + "epoch": 0.9013837030529321, + "grad_norm": 2.3085379600524902, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7190700769424438, + "num_tokens": 208880104.0, + "step": 8208 + }, + { + "epoch": 0.9014935207555458, + "grad_norm": 1.9497565031051636, + "learning_rate": 1e-06, + "loss": 1.0675, + "mean_token_accuracy": 0.6868240237236023, + "num_tokens": 208912599.0, + "step": 8209 + }, + { + "epoch": 0.9016033384581594, + "grad_norm": 2.0761380195617676, + "learning_rate": 1e-06, + "loss": 1.0047, + "mean_token_accuracy": 0.7076584100723267, + "num_tokens": 208940523.0, + "step": 8210 + }, + { + "epoch": 0.9017131561607731, + "grad_norm": 1.9574964046478271, + "learning_rate": 1e-06, + "loss": 1.0346, + "mean_token_accuracy": 0.694349467754364, + "num_tokens": 208970226.0, + "step": 8211 + }, + { + "epoch": 0.9018229738633867, + "grad_norm": 2.411033868789673, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7209329605102539, + "num_tokens": 208990513.0, + "step": 8212 + }, + { + "epoch": 0.9019327915660005, + "grad_norm": 3.249237060546875, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.7406085729598999, + "num_tokens": 209003192.0, + "step": 8213 + }, + { + "epoch": 0.9020426092686141, + "grad_norm": 2.4826714992523193, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6865688562393188, + "num_tokens": 209025536.0, + "step": 8214 + }, + { + "epoch": 0.9021524269712278, + "grad_norm": 2.4712774753570557, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7249273061752319, + "num_tokens": 209045151.0, + "step": 8215 + }, + { + "epoch": 0.9022622446738414, + "grad_norm": 2.149852991104126, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7123633623123169, + "num_tokens": 209070557.0, + "step": 8216 + }, + { + "epoch": 0.9023720623764551, + "grad_norm": 2.164612054824829, + "learning_rate": 1e-06, + "loss": 1.0275, + "mean_token_accuracy": 0.6867923736572266, + "num_tokens": 209096086.0, + "step": 8217 + }, + { + "epoch": 0.9024818800790687, + "grad_norm": 2.080136299133301, + "learning_rate": 1e-06, + "loss": 0.9964, + "mean_token_accuracy": 0.6914583444595337, + "num_tokens": 209123660.0, + "step": 8218 + }, + { + "epoch": 0.9025916977816824, + "grad_norm": 2.403654098510742, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7260003089904785, + "num_tokens": 209143977.0, + "step": 8219 + }, + { + "epoch": 0.9027015154842961, + "grad_norm": 2.1197493076324463, + "learning_rate": 1e-06, + "loss": 1.0565, + "mean_token_accuracy": 0.6850875616073608, + "num_tokens": 209170696.0, + "step": 8220 + }, + { + "epoch": 0.9028113331869098, + "grad_norm": 2.181661605834961, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.732770562171936, + "num_tokens": 209195670.0, + "step": 8221 + }, + { + "epoch": 0.9029211508895234, + "grad_norm": 2.3938846588134766, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7250015735626221, + "num_tokens": 209216316.0, + "step": 8222 + }, + { + "epoch": 0.903030968592137, + "grad_norm": 1.8982164859771729, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7000741958618164, + "num_tokens": 209244976.0, + "step": 8223 + }, + { + "epoch": 0.9031407862947507, + "grad_norm": 2.199846029281616, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7262961268424988, + "num_tokens": 209268879.0, + "step": 8224 + }, + { + "epoch": 0.9032506039973643, + "grad_norm": 2.38521671295166, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.716592013835907, + "num_tokens": 209290231.0, + "step": 8225 + }, + { + "epoch": 0.903360421699978, + "grad_norm": 2.32403564453125, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7050939798355103, + "num_tokens": 209313622.0, + "step": 8226 + }, + { + "epoch": 0.9034702394025917, + "grad_norm": 2.0879642963409424, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6923325657844543, + "num_tokens": 209342564.0, + "step": 8227 + }, + { + "epoch": 0.9035800571052054, + "grad_norm": 2.2097041606903076, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7034232020378113, + "num_tokens": 209368975.0, + "step": 8228 + }, + { + "epoch": 0.903689874807819, + "grad_norm": 2.266542911529541, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.712103009223938, + "num_tokens": 209392957.0, + "step": 8229 + }, + { + "epoch": 0.9037996925104327, + "grad_norm": 2.1312952041625977, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7032025456428528, + "num_tokens": 209420829.0, + "step": 8230 + }, + { + "epoch": 0.9039095102130463, + "grad_norm": 2.4592127799987793, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7012825012207031, + "num_tokens": 209442473.0, + "step": 8231 + }, + { + "epoch": 0.90401932791566, + "grad_norm": 2.034255266189575, + "learning_rate": 1e-06, + "loss": 1.0628, + "mean_token_accuracy": 0.6803877353668213, + "num_tokens": 209471454.0, + "step": 8232 + }, + { + "epoch": 0.9041291456182736, + "grad_norm": 2.1090874671936035, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7095226049423218, + "num_tokens": 209497622.0, + "step": 8233 + }, + { + "epoch": 0.9042389633208874, + "grad_norm": 2.2961573600769043, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7334851026535034, + "num_tokens": 209519169.0, + "step": 8234 + }, + { + "epoch": 0.904348781023501, + "grad_norm": 2.11775803565979, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.6968786716461182, + "num_tokens": 209543958.0, + "step": 8235 + }, + { + "epoch": 0.9044585987261147, + "grad_norm": 2.2506377696990967, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7162363529205322, + "num_tokens": 209567387.0, + "step": 8236 + }, + { + "epoch": 0.9045684164287283, + "grad_norm": 2.3724939823150635, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7096306681632996, + "num_tokens": 209590807.0, + "step": 8237 + }, + { + "epoch": 0.904678234131342, + "grad_norm": 2.0749311447143555, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.698205292224884, + "num_tokens": 209618669.0, + "step": 8238 + }, + { + "epoch": 0.9047880518339556, + "grad_norm": 2.273160457611084, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7108319997787476, + "num_tokens": 209642940.0, + "step": 8239 + }, + { + "epoch": 0.9048978695365693, + "grad_norm": 1.9350881576538086, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7015185356140137, + "num_tokens": 209676579.0, + "step": 8240 + }, + { + "epoch": 0.9050076872391829, + "grad_norm": 2.1465232372283936, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7035270929336548, + "num_tokens": 209704998.0, + "step": 8241 + }, + { + "epoch": 0.9051175049417967, + "grad_norm": 2.287508964538574, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6865631341934204, + "num_tokens": 209731712.0, + "step": 8242 + }, + { + "epoch": 0.9052273226444103, + "grad_norm": 2.296762466430664, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7181688547134399, + "num_tokens": 209756224.0, + "step": 8243 + }, + { + "epoch": 0.905337140347024, + "grad_norm": 2.548938274383545, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7183545827865601, + "num_tokens": 209775807.0, + "step": 8244 + }, + { + "epoch": 0.9054469580496376, + "grad_norm": 2.534546136856079, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.6994470357894897, + "num_tokens": 209796889.0, + "step": 8245 + }, + { + "epoch": 0.9055567757522512, + "grad_norm": 2.1158359050750732, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.6949409246444702, + "num_tokens": 209823848.0, + "step": 8246 + }, + { + "epoch": 0.9056665934548649, + "grad_norm": 2.125851631164551, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7422915101051331, + "num_tokens": 209849930.0, + "step": 8247 + }, + { + "epoch": 0.9057764111574785, + "grad_norm": 2.1988418102264404, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7160119414329529, + "num_tokens": 209875068.0, + "step": 8248 + }, + { + "epoch": 0.9058862288600923, + "grad_norm": 2.2079477310180664, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7168853282928467, + "num_tokens": 209898121.0, + "step": 8249 + }, + { + "epoch": 0.9059960465627059, + "grad_norm": 2.3388166427612305, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7028516530990601, + "num_tokens": 209920338.0, + "step": 8250 + }, + { + "epoch": 0.9061058642653196, + "grad_norm": 2.093014717102051, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7159628868103027, + "num_tokens": 209947352.0, + "step": 8251 + }, + { + "epoch": 0.9062156819679332, + "grad_norm": 2.239006280899048, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7090038657188416, + "num_tokens": 209970889.0, + "step": 8252 + }, + { + "epoch": 0.9063254996705469, + "grad_norm": 2.072361946105957, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7006782293319702, + "num_tokens": 210000517.0, + "step": 8253 + }, + { + "epoch": 0.9064353173731605, + "grad_norm": 2.2326316833496094, + "learning_rate": 1e-06, + "loss": 1.0091, + "mean_token_accuracy": 0.6897940635681152, + "num_tokens": 210026063.0, + "step": 8254 + }, + { + "epoch": 0.9065451350757742, + "grad_norm": 1.9947727918624878, + "learning_rate": 1e-06, + "loss": 1.0501, + "mean_token_accuracy": 0.686119019985199, + "num_tokens": 210055700.0, + "step": 8255 + }, + { + "epoch": 0.9066549527783879, + "grad_norm": 2.053481340408325, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7110298871994019, + "num_tokens": 210080625.0, + "step": 8256 + }, + { + "epoch": 0.9067647704810016, + "grad_norm": 2.421898126602173, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7203190922737122, + "num_tokens": 210102400.0, + "step": 8257 + }, + { + "epoch": 0.9068745881836152, + "grad_norm": 1.9727277755737305, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7073403000831604, + "num_tokens": 210131898.0, + "step": 8258 + }, + { + "epoch": 0.9069844058862289, + "grad_norm": 2.4793145656585693, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7095359563827515, + "num_tokens": 210153362.0, + "step": 8259 + }, + { + "epoch": 0.9070942235888425, + "grad_norm": 2.555055856704712, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7305516004562378, + "num_tokens": 210173053.0, + "step": 8260 + }, + { + "epoch": 0.9072040412914562, + "grad_norm": 2.2676146030426025, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7200922966003418, + "num_tokens": 210193861.0, + "step": 8261 + }, + { + "epoch": 0.9073138589940698, + "grad_norm": 2.3617732524871826, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7344856262207031, + "num_tokens": 210215563.0, + "step": 8262 + }, + { + "epoch": 0.9074236766966836, + "grad_norm": 2.1243128776550293, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7100769281387329, + "num_tokens": 210241488.0, + "step": 8263 + }, + { + "epoch": 0.9075334943992972, + "grad_norm": 2.265399694442749, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.7492265105247498, + "num_tokens": 210263390.0, + "step": 8264 + }, + { + "epoch": 0.9076433121019108, + "grad_norm": 2.0168495178222656, + "learning_rate": 1e-06, + "loss": 1.1576, + "mean_token_accuracy": 0.665839672088623, + "num_tokens": 210293183.0, + "step": 8265 + }, + { + "epoch": 0.9077531298045245, + "grad_norm": 2.2586517333984375, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.7012603878974915, + "num_tokens": 210316858.0, + "step": 8266 + }, + { + "epoch": 0.9078629475071381, + "grad_norm": 1.85434091091156, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6936867237091064, + "num_tokens": 210351043.0, + "step": 8267 + }, + { + "epoch": 0.9079727652097518, + "grad_norm": 2.3463494777679443, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7040437459945679, + "num_tokens": 210373212.0, + "step": 8268 + }, + { + "epoch": 0.9080825829123654, + "grad_norm": 2.270833969116211, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7178497314453125, + "num_tokens": 210397314.0, + "step": 8269 + }, + { + "epoch": 0.9081924006149792, + "grad_norm": 2.312744617462158, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7122817039489746, + "num_tokens": 210419903.0, + "step": 8270 + }, + { + "epoch": 0.9083022183175928, + "grad_norm": 2.111590623855591, + "learning_rate": 1e-06, + "loss": 1.0164, + "mean_token_accuracy": 0.7003762125968933, + "num_tokens": 210448143.0, + "step": 8271 + }, + { + "epoch": 0.9084120360202065, + "grad_norm": 2.156665802001953, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.6963162422180176, + "num_tokens": 210474952.0, + "step": 8272 + }, + { + "epoch": 0.9085218537228201, + "grad_norm": 2.123281955718994, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.704939067363739, + "num_tokens": 210500778.0, + "step": 8273 + }, + { + "epoch": 0.9086316714254338, + "grad_norm": 2.2007923126220703, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7209259271621704, + "num_tokens": 210524323.0, + "step": 8274 + }, + { + "epoch": 0.9087414891280474, + "grad_norm": 2.0405869483947754, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7122018337249756, + "num_tokens": 210553274.0, + "step": 8275 + }, + { + "epoch": 0.9088513068306611, + "grad_norm": 2.0420327186584473, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.6960430145263672, + "num_tokens": 210584744.0, + "step": 8276 + }, + { + "epoch": 0.9089611245332747, + "grad_norm": 2.3484907150268555, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.720952033996582, + "num_tokens": 210607754.0, + "step": 8277 + }, + { + "epoch": 0.9090709422358885, + "grad_norm": 2.058185338973999, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7041133046150208, + "num_tokens": 210635594.0, + "step": 8278 + }, + { + "epoch": 0.9091807599385021, + "grad_norm": 2.574592113494873, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.706604540348053, + "num_tokens": 210656808.0, + "step": 8279 + }, + { + "epoch": 0.9092905776411158, + "grad_norm": 2.320683240890503, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.7051129341125488, + "num_tokens": 210679923.0, + "step": 8280 + }, + { + "epoch": 0.9094003953437294, + "grad_norm": 2.1163713932037354, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.7044562697410583, + "num_tokens": 210707650.0, + "step": 8281 + }, + { + "epoch": 0.909510213046343, + "grad_norm": 1.9358328580856323, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7031346559524536, + "num_tokens": 210736351.0, + "step": 8282 + }, + { + "epoch": 0.9096200307489567, + "grad_norm": 2.568730115890503, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7103710174560547, + "num_tokens": 210755648.0, + "step": 8283 + }, + { + "epoch": 0.9097298484515703, + "grad_norm": 2.1751556396484375, + "learning_rate": 1e-06, + "loss": 0.9849, + "mean_token_accuracy": 0.694961428642273, + "num_tokens": 210781847.0, + "step": 8284 + }, + { + "epoch": 0.9098396661541841, + "grad_norm": 2.441279888153076, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7241984009742737, + "num_tokens": 210802386.0, + "step": 8285 + }, + { + "epoch": 0.9099494838567977, + "grad_norm": 2.5083329677581787, + "learning_rate": 1e-06, + "loss": 0.7697, + "mean_token_accuracy": 0.7530532479286194, + "num_tokens": 210820779.0, + "step": 8286 + }, + { + "epoch": 0.9100593015594114, + "grad_norm": 1.8872214555740356, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6981202363967896, + "num_tokens": 210853500.0, + "step": 8287 + }, + { + "epoch": 0.910169119262025, + "grad_norm": 2.059765338897705, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.6946958303451538, + "num_tokens": 210880622.0, + "step": 8288 + }, + { + "epoch": 0.9102789369646387, + "grad_norm": 2.2850992679595947, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7227672934532166, + "num_tokens": 210903650.0, + "step": 8289 + }, + { + "epoch": 0.9103887546672523, + "grad_norm": 2.0874128341674805, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7189761400222778, + "num_tokens": 210930843.0, + "step": 8290 + }, + { + "epoch": 0.910498572369866, + "grad_norm": 2.3808467388153076, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7138375639915466, + "num_tokens": 210952936.0, + "step": 8291 + }, + { + "epoch": 0.9106083900724797, + "grad_norm": 2.063023805618286, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7155544757843018, + "num_tokens": 210979710.0, + "step": 8292 + }, + { + "epoch": 0.9107182077750934, + "grad_norm": 2.245286703109741, + "learning_rate": 1e-06, + "loss": 1.0714, + "mean_token_accuracy": 0.673198938369751, + "num_tokens": 211006927.0, + "step": 8293 + }, + { + "epoch": 0.910828025477707, + "grad_norm": 2.3444693088531494, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7132269144058228, + "num_tokens": 211031361.0, + "step": 8294 + }, + { + "epoch": 0.9109378431803207, + "grad_norm": 1.9802210330963135, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6949759125709534, + "num_tokens": 211059980.0, + "step": 8295 + }, + { + "epoch": 0.9110476608829343, + "grad_norm": 2.113294839859009, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7059637308120728, + "num_tokens": 211085603.0, + "step": 8296 + }, + { + "epoch": 0.911157478585548, + "grad_norm": 2.4412784576416016, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7202969193458557, + "num_tokens": 211105723.0, + "step": 8297 + }, + { + "epoch": 0.9112672962881616, + "grad_norm": 3.1054763793945312, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7083221673965454, + "num_tokens": 211120178.0, + "step": 8298 + }, + { + "epoch": 0.9113771139907754, + "grad_norm": 2.1734726428985596, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7213350534439087, + "num_tokens": 211143588.0, + "step": 8299 + }, + { + "epoch": 0.911486931693389, + "grad_norm": 2.6817352771759033, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7127009630203247, + "num_tokens": 211163606.0, + "step": 8300 + }, + { + "epoch": 0.9115967493960027, + "grad_norm": 2.2278311252593994, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7101132273674011, + "num_tokens": 211189783.0, + "step": 8301 + }, + { + "epoch": 0.9117065670986163, + "grad_norm": 2.14892578125, + "learning_rate": 1e-06, + "loss": 0.9918, + "mean_token_accuracy": 0.6982035040855408, + "num_tokens": 211216251.0, + "step": 8302 + }, + { + "epoch": 0.91181638480123, + "grad_norm": 2.2883877754211426, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7159568071365356, + "num_tokens": 211238042.0, + "step": 8303 + }, + { + "epoch": 0.9119262025038436, + "grad_norm": 2.012866258621216, + "learning_rate": 1e-06, + "loss": 1.0386, + "mean_token_accuracy": 0.681580662727356, + "num_tokens": 211268967.0, + "step": 8304 + }, + { + "epoch": 0.9120360202064572, + "grad_norm": 2.147653579711914, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7038832902908325, + "num_tokens": 211297131.0, + "step": 8305 + }, + { + "epoch": 0.9121458379090709, + "grad_norm": 2.180455446243286, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7214974164962769, + "num_tokens": 211321907.0, + "step": 8306 + }, + { + "epoch": 0.9122556556116846, + "grad_norm": 2.1492786407470703, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.723399817943573, + "num_tokens": 211345535.0, + "step": 8307 + }, + { + "epoch": 0.9123654733142983, + "grad_norm": 2.081744432449341, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6950836181640625, + "num_tokens": 211374003.0, + "step": 8308 + }, + { + "epoch": 0.9124752910169119, + "grad_norm": 1.9302395582199097, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7144590616226196, + "num_tokens": 211405080.0, + "step": 8309 + }, + { + "epoch": 0.9125851087195256, + "grad_norm": 2.5289323329925537, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7096613049507141, + "num_tokens": 211425946.0, + "step": 8310 + }, + { + "epoch": 0.9126949264221392, + "grad_norm": 2.057145357131958, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.6966376304626465, + "num_tokens": 211455015.0, + "step": 8311 + }, + { + "epoch": 0.9128047441247529, + "grad_norm": 2.1090962886810303, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6912244558334351, + "num_tokens": 211484911.0, + "step": 8312 + }, + { + "epoch": 0.9129145618273665, + "grad_norm": 2.3194944858551025, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7255479097366333, + "num_tokens": 211507003.0, + "step": 8313 + }, + { + "epoch": 0.9130243795299803, + "grad_norm": 2.1284403800964355, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7114046812057495, + "num_tokens": 211532799.0, + "step": 8314 + }, + { + "epoch": 0.9131341972325939, + "grad_norm": 2.1606712341308594, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7077165842056274, + "num_tokens": 211559544.0, + "step": 8315 + }, + { + "epoch": 0.9132440149352076, + "grad_norm": 2.2455766201019287, + "learning_rate": 1e-06, + "loss": 1.0429, + "mean_token_accuracy": 0.6849130392074585, + "num_tokens": 211586440.0, + "step": 8316 + }, + { + "epoch": 0.9133538326378212, + "grad_norm": 2.1238808631896973, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6900659799575806, + "num_tokens": 211614389.0, + "step": 8317 + }, + { + "epoch": 0.9134636503404349, + "grad_norm": 2.010295867919922, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7119884490966797, + "num_tokens": 211642090.0, + "step": 8318 + }, + { + "epoch": 0.9135734680430485, + "grad_norm": 2.0546441078186035, + "learning_rate": 1e-06, + "loss": 1.0458, + "mean_token_accuracy": 0.6805378198623657, + "num_tokens": 211670550.0, + "step": 8319 + }, + { + "epoch": 0.9136832857456622, + "grad_norm": 2.3308820724487305, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7104668617248535, + "num_tokens": 211693800.0, + "step": 8320 + }, + { + "epoch": 0.9137931034482759, + "grad_norm": 2.2420644760131836, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7079660892486572, + "num_tokens": 211717532.0, + "step": 8321 + }, + { + "epoch": 0.9139029211508896, + "grad_norm": 2.0308103561401367, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7190170288085938, + "num_tokens": 211745962.0, + "step": 8322 + }, + { + "epoch": 0.9140127388535032, + "grad_norm": 2.2775301933288574, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7297065258026123, + "num_tokens": 211769758.0, + "step": 8323 + }, + { + "epoch": 0.9141225565561168, + "grad_norm": 2.2804477214813232, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7348935604095459, + "num_tokens": 211792225.0, + "step": 8324 + }, + { + "epoch": 0.9142323742587305, + "grad_norm": 2.086268663406372, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7202556133270264, + "num_tokens": 211820396.0, + "step": 8325 + }, + { + "epoch": 0.9143421919613441, + "grad_norm": 2.103644609451294, + "learning_rate": 1e-06, + "loss": 1.0356, + "mean_token_accuracy": 0.6890290975570679, + "num_tokens": 211848529.0, + "step": 8326 + }, + { + "epoch": 0.9144520096639578, + "grad_norm": 2.08828067779541, + "learning_rate": 1e-06, + "loss": 1.055, + "mean_token_accuracy": 0.683010458946228, + "num_tokens": 211876026.0, + "step": 8327 + }, + { + "epoch": 0.9145618273665715, + "grad_norm": 2.599641799926758, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7143768668174744, + "num_tokens": 211895578.0, + "step": 8328 + }, + { + "epoch": 0.9146716450691852, + "grad_norm": 2.033802032470703, + "learning_rate": 1e-06, + "loss": 1.0262, + "mean_token_accuracy": 0.6900160908699036, + "num_tokens": 211925026.0, + "step": 8329 + }, + { + "epoch": 0.9147814627717988, + "grad_norm": 2.0871331691741943, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7046617865562439, + "num_tokens": 211952123.0, + "step": 8330 + }, + { + "epoch": 0.9148912804744125, + "grad_norm": 2.2350354194641113, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7020505666732788, + "num_tokens": 211978209.0, + "step": 8331 + }, + { + "epoch": 0.9150010981770261, + "grad_norm": 1.9255820512771606, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.7023878693580627, + "num_tokens": 212009306.0, + "step": 8332 + }, + { + "epoch": 0.9151109158796398, + "grad_norm": 1.9365068674087524, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7123713493347168, + "num_tokens": 212038308.0, + "step": 8333 + }, + { + "epoch": 0.9152207335822534, + "grad_norm": 2.1503143310546875, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.719246506690979, + "num_tokens": 212062604.0, + "step": 8334 + }, + { + "epoch": 0.9153305512848671, + "grad_norm": 1.9146058559417725, + "learning_rate": 1e-06, + "loss": 1.0337, + "mean_token_accuracy": 0.6860091686248779, + "num_tokens": 212095038.0, + "step": 8335 + }, + { + "epoch": 0.9154403689874808, + "grad_norm": 2.248944044113159, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6970667839050293, + "num_tokens": 212119121.0, + "step": 8336 + }, + { + "epoch": 0.9155501866900945, + "grad_norm": 2.449103832244873, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7023389935493469, + "num_tokens": 212140736.0, + "step": 8337 + }, + { + "epoch": 0.9156600043927081, + "grad_norm": 1.892065167427063, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.703471839427948, + "num_tokens": 212172657.0, + "step": 8338 + }, + { + "epoch": 0.9157698220953218, + "grad_norm": 2.3815667629241943, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.6999354958534241, + "num_tokens": 212194984.0, + "step": 8339 + }, + { + "epoch": 0.9158796397979354, + "grad_norm": 1.8275203704833984, + "learning_rate": 1e-06, + "loss": 1.016, + "mean_token_accuracy": 0.6977394819259644, + "num_tokens": 212230757.0, + "step": 8340 + }, + { + "epoch": 0.915989457500549, + "grad_norm": 2.0259735584259033, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7053056359291077, + "num_tokens": 212260574.0, + "step": 8341 + }, + { + "epoch": 0.9160992752031627, + "grad_norm": 1.9842746257781982, + "learning_rate": 1e-06, + "loss": 1.0169, + "mean_token_accuracy": 0.693212628364563, + "num_tokens": 212289878.0, + "step": 8342 + }, + { + "epoch": 0.9162090929057765, + "grad_norm": 2.1065778732299805, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6877778768539429, + "num_tokens": 212318012.0, + "step": 8343 + }, + { + "epoch": 0.9163189106083901, + "grad_norm": 2.0338656902313232, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7059090733528137, + "num_tokens": 212347198.0, + "step": 8344 + }, + { + "epoch": 0.9164287283110037, + "grad_norm": 2.201836347579956, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.700040876865387, + "num_tokens": 212370990.0, + "step": 8345 + }, + { + "epoch": 0.9165385460136174, + "grad_norm": 2.3209176063537598, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7210098505020142, + "num_tokens": 212392968.0, + "step": 8346 + }, + { + "epoch": 0.916648363716231, + "grad_norm": 2.304152488708496, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6906217336654663, + "num_tokens": 212418398.0, + "step": 8347 + }, + { + "epoch": 0.9167581814188447, + "grad_norm": 2.1191651821136475, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7125012874603271, + "num_tokens": 212443862.0, + "step": 8348 + }, + { + "epoch": 0.9168679991214583, + "grad_norm": 2.1254804134368896, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7139338254928589, + "num_tokens": 212471230.0, + "step": 8349 + }, + { + "epoch": 0.9169778168240721, + "grad_norm": 2.365941286087036, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7060604095458984, + "num_tokens": 212493211.0, + "step": 8350 + }, + { + "epoch": 0.9170876345266857, + "grad_norm": 2.036404848098755, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.712457001209259, + "num_tokens": 212524589.0, + "step": 8351 + }, + { + "epoch": 0.9171974522292994, + "grad_norm": 2.3965065479278564, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6956843137741089, + "num_tokens": 212546167.0, + "step": 8352 + }, + { + "epoch": 0.917307269931913, + "grad_norm": 2.536048412322998, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7200722694396973, + "num_tokens": 212566624.0, + "step": 8353 + }, + { + "epoch": 0.9174170876345267, + "grad_norm": 2.4212167263031006, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6942592859268188, + "num_tokens": 212588532.0, + "step": 8354 + }, + { + "epoch": 0.9175269053371403, + "grad_norm": 2.4186646938323975, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7369567155838013, + "num_tokens": 212608077.0, + "step": 8355 + }, + { + "epoch": 0.917636723039754, + "grad_norm": 2.208177328109741, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7154701948165894, + "num_tokens": 212634580.0, + "step": 8356 + }, + { + "epoch": 0.9177465407423677, + "grad_norm": 1.9011400938034058, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.707339882850647, + "num_tokens": 212666267.0, + "step": 8357 + }, + { + "epoch": 0.9178563584449814, + "grad_norm": 2.328744888305664, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7214481830596924, + "num_tokens": 212689037.0, + "step": 8358 + }, + { + "epoch": 0.917966176147595, + "grad_norm": 1.938600778579712, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.698585033416748, + "num_tokens": 212718000.0, + "step": 8359 + }, + { + "epoch": 0.9180759938502087, + "grad_norm": 2.4839532375335693, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7176403403282166, + "num_tokens": 212739326.0, + "step": 8360 + }, + { + "epoch": 0.9181858115528223, + "grad_norm": 2.235564947128296, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7078345417976379, + "num_tokens": 212765050.0, + "step": 8361 + }, + { + "epoch": 0.918295629255436, + "grad_norm": 2.2402701377868652, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.6987761855125427, + "num_tokens": 212788195.0, + "step": 8362 + }, + { + "epoch": 0.9184054469580496, + "grad_norm": 2.4451403617858887, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7321106791496277, + "num_tokens": 212807779.0, + "step": 8363 + }, + { + "epoch": 0.9185152646606632, + "grad_norm": 2.4589619636535645, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7133964896202087, + "num_tokens": 212827928.0, + "step": 8364 + }, + { + "epoch": 0.918625082363277, + "grad_norm": 2.281926155090332, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7096432447433472, + "num_tokens": 212850149.0, + "step": 8365 + }, + { + "epoch": 0.9187349000658906, + "grad_norm": 2.249424934387207, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.708307147026062, + "num_tokens": 212873127.0, + "step": 8366 + }, + { + "epoch": 0.9188447177685043, + "grad_norm": 2.5944714546203613, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7222957015037537, + "num_tokens": 212891841.0, + "step": 8367 + }, + { + "epoch": 0.9189545354711179, + "grad_norm": 1.992997407913208, + "learning_rate": 1e-06, + "loss": 0.9744, + "mean_token_accuracy": 0.7006297707557678, + "num_tokens": 212920774.0, + "step": 8368 + }, + { + "epoch": 0.9190643531737316, + "grad_norm": 2.1722640991210938, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7342256903648376, + "num_tokens": 212944829.0, + "step": 8369 + }, + { + "epoch": 0.9191741708763452, + "grad_norm": 1.8742284774780273, + "learning_rate": 1e-06, + "loss": 1.0805, + "mean_token_accuracy": 0.6785857081413269, + "num_tokens": 212976681.0, + "step": 8370 + }, + { + "epoch": 0.9192839885789589, + "grad_norm": 2.3092713356018066, + "learning_rate": 1e-06, + "loss": 0.7791, + "mean_token_accuracy": 0.7485859394073486, + "num_tokens": 212997108.0, + "step": 8371 + }, + { + "epoch": 0.9193938062815726, + "grad_norm": 2.1746811866760254, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.705933690071106, + "num_tokens": 213021451.0, + "step": 8372 + }, + { + "epoch": 0.9195036239841863, + "grad_norm": 2.1174092292785645, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7359098196029663, + "num_tokens": 213047696.0, + "step": 8373 + }, + { + "epoch": 0.9196134416867999, + "grad_norm": 2.4651339054107666, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7100405693054199, + "num_tokens": 213067686.0, + "step": 8374 + }, + { + "epoch": 0.9197232593894136, + "grad_norm": 2.221898078918457, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7255226969718933, + "num_tokens": 213091946.0, + "step": 8375 + }, + { + "epoch": 0.9198330770920272, + "grad_norm": 1.924630045890808, + "learning_rate": 1e-06, + "loss": 1.019, + "mean_token_accuracy": 0.6927123069763184, + "num_tokens": 213124545.0, + "step": 8376 + }, + { + "epoch": 0.9199428947946409, + "grad_norm": 2.434741497039795, + "learning_rate": 1e-06, + "loss": 1.0117, + "mean_token_accuracy": 0.6973147392272949, + "num_tokens": 213146371.0, + "step": 8377 + }, + { + "epoch": 0.9200527124972545, + "grad_norm": 1.9616329669952393, + "learning_rate": 1e-06, + "loss": 1.0384, + "mean_token_accuracy": 0.6891980171203613, + "num_tokens": 213179229.0, + "step": 8378 + }, + { + "epoch": 0.9201625301998683, + "grad_norm": 2.5465826988220215, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7280652523040771, + "num_tokens": 213198709.0, + "step": 8379 + }, + { + "epoch": 0.9202723479024819, + "grad_norm": 2.096818208694458, + "learning_rate": 1e-06, + "loss": 1.0037, + "mean_token_accuracy": 0.6932730674743652, + "num_tokens": 213224602.0, + "step": 8380 + }, + { + "epoch": 0.9203821656050956, + "grad_norm": 2.0898404121398926, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7227023839950562, + "num_tokens": 213250018.0, + "step": 8381 + }, + { + "epoch": 0.9204919833077092, + "grad_norm": 2.360807418823242, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7234821915626526, + "num_tokens": 213273327.0, + "step": 8382 + }, + { + "epoch": 0.9206018010103229, + "grad_norm": 2.358016014099121, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.7016748189926147, + "num_tokens": 213296329.0, + "step": 8383 + }, + { + "epoch": 0.9207116187129365, + "grad_norm": 2.1656527519226074, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7220278382301331, + "num_tokens": 213320845.0, + "step": 8384 + }, + { + "epoch": 0.9208214364155501, + "grad_norm": 2.0199944972991943, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7291005849838257, + "num_tokens": 213349090.0, + "step": 8385 + }, + { + "epoch": 0.9209312541181639, + "grad_norm": 2.2340948581695557, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7067970633506775, + "num_tokens": 213373604.0, + "step": 8386 + }, + { + "epoch": 0.9210410718207775, + "grad_norm": 2.1546428203582764, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7167047262191772, + "num_tokens": 213401299.0, + "step": 8387 + }, + { + "epoch": 0.9211508895233912, + "grad_norm": 2.4592955112457275, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6981618404388428, + "num_tokens": 213422687.0, + "step": 8388 + }, + { + "epoch": 0.9212607072260048, + "grad_norm": 2.007009267807007, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7364171743392944, + "num_tokens": 213450850.0, + "step": 8389 + }, + { + "epoch": 0.9213705249286185, + "grad_norm": 2.2126710414886475, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7272230386734009, + "num_tokens": 213474965.0, + "step": 8390 + }, + { + "epoch": 0.9214803426312321, + "grad_norm": 2.102635622024536, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7109538316726685, + "num_tokens": 213502366.0, + "step": 8391 + }, + { + "epoch": 0.9215901603338458, + "grad_norm": 2.212498903274536, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7314554452896118, + "num_tokens": 213525924.0, + "step": 8392 + }, + { + "epoch": 0.9216999780364594, + "grad_norm": 2.1747164726257324, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7289156913757324, + "num_tokens": 213550173.0, + "step": 8393 + }, + { + "epoch": 0.9218097957390732, + "grad_norm": 2.1495063304901123, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7091460227966309, + "num_tokens": 213578424.0, + "step": 8394 + }, + { + "epoch": 0.9219196134416868, + "grad_norm": 2.247588872909546, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.72071373462677, + "num_tokens": 213601678.0, + "step": 8395 + }, + { + "epoch": 0.9220294311443005, + "grad_norm": 2.170933485031128, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7103521823883057, + "num_tokens": 213629329.0, + "step": 8396 + }, + { + "epoch": 0.9221392488469141, + "grad_norm": 2.0822343826293945, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7065550088882446, + "num_tokens": 213657557.0, + "step": 8397 + }, + { + "epoch": 0.9222490665495278, + "grad_norm": 2.292940378189087, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7268631458282471, + "num_tokens": 213679128.0, + "step": 8398 + }, + { + "epoch": 0.9223588842521414, + "grad_norm": 2.234569549560547, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6947996616363525, + "num_tokens": 213704807.0, + "step": 8399 + }, + { + "epoch": 0.922468701954755, + "grad_norm": 1.8999946117401123, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6934128999710083, + "num_tokens": 213737341.0, + "step": 8400 + }, + { + "epoch": 0.9225785196573688, + "grad_norm": 1.9859111309051514, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.688936710357666, + "num_tokens": 213766426.0, + "step": 8401 + }, + { + "epoch": 0.9226883373599825, + "grad_norm": 2.0762698650360107, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6998114585876465, + "num_tokens": 213796605.0, + "step": 8402 + }, + { + "epoch": 0.9227981550625961, + "grad_norm": 2.160949945449829, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7164379954338074, + "num_tokens": 213820834.0, + "step": 8403 + }, + { + "epoch": 0.9229079727652097, + "grad_norm": 2.0875823497772217, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.7031725645065308, + "num_tokens": 213848631.0, + "step": 8404 + }, + { + "epoch": 0.9230177904678234, + "grad_norm": 2.273125171661377, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7030396461486816, + "num_tokens": 213873656.0, + "step": 8405 + }, + { + "epoch": 0.923127608170437, + "grad_norm": 2.086601495742798, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7189669609069824, + "num_tokens": 213899722.0, + "step": 8406 + }, + { + "epoch": 0.9232374258730507, + "grad_norm": 2.0733659267425537, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7039761543273926, + "num_tokens": 213925359.0, + "step": 8407 + }, + { + "epoch": 0.9233472435756644, + "grad_norm": 2.2333321571350098, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7147419452667236, + "num_tokens": 213950155.0, + "step": 8408 + }, + { + "epoch": 0.9234570612782781, + "grad_norm": 2.323819875717163, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7298721075057983, + "num_tokens": 213971986.0, + "step": 8409 + }, + { + "epoch": 0.9235668789808917, + "grad_norm": 2.028506278991699, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6968038082122803, + "num_tokens": 213999819.0, + "step": 8410 + }, + { + "epoch": 0.9236766966835054, + "grad_norm": 2.1008691787719727, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6930212378501892, + "num_tokens": 214027995.0, + "step": 8411 + }, + { + "epoch": 0.923786514386119, + "grad_norm": 2.154137372970581, + "learning_rate": 1e-06, + "loss": 1.03, + "mean_token_accuracy": 0.68971848487854, + "num_tokens": 214059756.0, + "step": 8412 + }, + { + "epoch": 0.9238963320887327, + "grad_norm": 2.243969678878784, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7276877760887146, + "num_tokens": 214082909.0, + "step": 8413 + }, + { + "epoch": 0.9240061497913463, + "grad_norm": 2.0175068378448486, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7149947285652161, + "num_tokens": 214111278.0, + "step": 8414 + }, + { + "epoch": 0.9241159674939601, + "grad_norm": 1.9013187885284424, + "learning_rate": 1e-06, + "loss": 1.0515, + "mean_token_accuracy": 0.6810178160667419, + "num_tokens": 214144457.0, + "step": 8415 + }, + { + "epoch": 0.9242257851965737, + "grad_norm": 1.9339497089385986, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7061882615089417, + "num_tokens": 214176129.0, + "step": 8416 + }, + { + "epoch": 0.9243356028991874, + "grad_norm": 2.4050002098083496, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7239161133766174, + "num_tokens": 214197459.0, + "step": 8417 + }, + { + "epoch": 0.924445420601801, + "grad_norm": 2.223275899887085, + "learning_rate": 1e-06, + "loss": 1.01, + "mean_token_accuracy": 0.6853033900260925, + "num_tokens": 214223355.0, + "step": 8418 + }, + { + "epoch": 0.9245552383044147, + "grad_norm": 2.384925365447998, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7137843370437622, + "num_tokens": 214245061.0, + "step": 8419 + }, + { + "epoch": 0.9246650560070283, + "grad_norm": 2.313566207885742, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7074437141418457, + "num_tokens": 214267920.0, + "step": 8420 + }, + { + "epoch": 0.924774873709642, + "grad_norm": 2.9136457443237305, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7173800468444824, + "num_tokens": 214285781.0, + "step": 8421 + }, + { + "epoch": 0.9248846914122557, + "grad_norm": 2.222478151321411, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7055984139442444, + "num_tokens": 214310918.0, + "step": 8422 + }, + { + "epoch": 0.9249945091148694, + "grad_norm": 2.1431045532226562, + "learning_rate": 1e-06, + "loss": 1.1091, + "mean_token_accuracy": 0.6832605600357056, + "num_tokens": 214340055.0, + "step": 8423 + }, + { + "epoch": 0.925104326817483, + "grad_norm": 2.075450897216797, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6915900111198425, + "num_tokens": 214371369.0, + "step": 8424 + }, + { + "epoch": 0.9252141445200966, + "grad_norm": 2.3837525844573975, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7051160335540771, + "num_tokens": 214393401.0, + "step": 8425 + }, + { + "epoch": 0.9253239622227103, + "grad_norm": 2.2609314918518066, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7034891247749329, + "num_tokens": 214416861.0, + "step": 8426 + }, + { + "epoch": 0.9254337799253239, + "grad_norm": 2.08892560005188, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7053830027580261, + "num_tokens": 214442498.0, + "step": 8427 + }, + { + "epoch": 0.9255435976279376, + "grad_norm": 1.9764961004257202, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7124488353729248, + "num_tokens": 214472327.0, + "step": 8428 + }, + { + "epoch": 0.9256534153305512, + "grad_norm": 2.2469375133514404, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6985734701156616, + "num_tokens": 214496757.0, + "step": 8429 + }, + { + "epoch": 0.925763233033165, + "grad_norm": 2.0678369998931885, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7085261344909668, + "num_tokens": 214525520.0, + "step": 8430 + }, + { + "epoch": 0.9258730507357786, + "grad_norm": 2.130326509475708, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.700308620929718, + "num_tokens": 214553359.0, + "step": 8431 + }, + { + "epoch": 0.9259828684383923, + "grad_norm": 2.0193443298339844, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.695041835308075, + "num_tokens": 214586343.0, + "step": 8432 + }, + { + "epoch": 0.9260926861410059, + "grad_norm": 2.2263901233673096, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7256256937980652, + "num_tokens": 214609930.0, + "step": 8433 + }, + { + "epoch": 0.9262025038436196, + "grad_norm": 2.0608632564544678, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7242388725280762, + "num_tokens": 214636846.0, + "step": 8434 + }, + { + "epoch": 0.9263123215462332, + "grad_norm": 2.0417349338531494, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7040250301361084, + "num_tokens": 214664868.0, + "step": 8435 + }, + { + "epoch": 0.9264221392488469, + "grad_norm": 2.1313283443450928, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6916995048522949, + "num_tokens": 214690153.0, + "step": 8436 + }, + { + "epoch": 0.9265319569514606, + "grad_norm": 2.206666946411133, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7094756364822388, + "num_tokens": 214716239.0, + "step": 8437 + }, + { + "epoch": 0.9266417746540743, + "grad_norm": 1.8069061040878296, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.6937222480773926, + "num_tokens": 214752294.0, + "step": 8438 + }, + { + "epoch": 0.9267515923566879, + "grad_norm": 2.2090086936950684, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.6968972682952881, + "num_tokens": 214776572.0, + "step": 8439 + }, + { + "epoch": 0.9268614100593016, + "grad_norm": 2.3322315216064453, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7401793003082275, + "num_tokens": 214798983.0, + "step": 8440 + }, + { + "epoch": 0.9269712277619152, + "grad_norm": 2.037597894668579, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6896716356277466, + "num_tokens": 214827358.0, + "step": 8441 + }, + { + "epoch": 0.9270810454645289, + "grad_norm": 2.310096263885498, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7106881141662598, + "num_tokens": 214849184.0, + "step": 8442 + }, + { + "epoch": 0.9271908631671425, + "grad_norm": 2.0754878520965576, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7288500070571899, + "num_tokens": 214872379.0, + "step": 8443 + }, + { + "epoch": 0.9273006808697563, + "grad_norm": 2.1653685569763184, + "learning_rate": 1e-06, + "loss": 1.0065, + "mean_token_accuracy": 0.6898059844970703, + "num_tokens": 214901132.0, + "step": 8444 + }, + { + "epoch": 0.9274104985723699, + "grad_norm": 1.880888819694519, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.6893632411956787, + "num_tokens": 214932600.0, + "step": 8445 + }, + { + "epoch": 0.9275203162749835, + "grad_norm": 2.5567338466644287, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7155033349990845, + "num_tokens": 214951757.0, + "step": 8446 + }, + { + "epoch": 0.9276301339775972, + "grad_norm": 2.185309648513794, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7022613286972046, + "num_tokens": 214979537.0, + "step": 8447 + }, + { + "epoch": 0.9277399516802108, + "grad_norm": 2.4332520961761475, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7273358702659607, + "num_tokens": 214999270.0, + "step": 8448 + }, + { + "epoch": 0.9278497693828245, + "grad_norm": 1.9972015619277954, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.6889939904212952, + "num_tokens": 215028322.0, + "step": 8449 + }, + { + "epoch": 0.9279595870854381, + "grad_norm": 2.230940580368042, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7267166972160339, + "num_tokens": 215052213.0, + "step": 8450 + }, + { + "epoch": 0.9280694047880519, + "grad_norm": 2.2546627521514893, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.6860374212265015, + "num_tokens": 215076764.0, + "step": 8451 + }, + { + "epoch": 0.9281792224906655, + "grad_norm": 2.0861520767211914, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.7034257650375366, + "num_tokens": 215105054.0, + "step": 8452 + }, + { + "epoch": 0.9282890401932792, + "grad_norm": 2.122520923614502, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7086584568023682, + "num_tokens": 215130041.0, + "step": 8453 + }, + { + "epoch": 0.9283988578958928, + "grad_norm": 2.3215131759643555, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7109569907188416, + "num_tokens": 215153055.0, + "step": 8454 + }, + { + "epoch": 0.9285086755985065, + "grad_norm": 2.0667471885681152, + "learning_rate": 1e-06, + "loss": 0.9888, + "mean_token_accuracy": 0.6999475359916687, + "num_tokens": 215179759.0, + "step": 8455 + }, + { + "epoch": 0.9286184933011201, + "grad_norm": 1.7370949983596802, + "learning_rate": 1e-06, + "loss": 1.017, + "mean_token_accuracy": 0.6912802457809448, + "num_tokens": 215217694.0, + "step": 8456 + }, + { + "epoch": 0.9287283110037338, + "grad_norm": 2.06923508644104, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.713904619216919, + "num_tokens": 215245805.0, + "step": 8457 + }, + { + "epoch": 0.9288381287063474, + "grad_norm": 2.4639463424682617, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7170169353485107, + "num_tokens": 215267484.0, + "step": 8458 + }, + { + "epoch": 0.9289479464089612, + "grad_norm": 1.952245831489563, + "learning_rate": 1e-06, + "loss": 1.0597, + "mean_token_accuracy": 0.6774423718452454, + "num_tokens": 215298155.0, + "step": 8459 + }, + { + "epoch": 0.9290577641115748, + "grad_norm": 2.2412121295928955, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7191999554634094, + "num_tokens": 215322337.0, + "step": 8460 + }, + { + "epoch": 0.9291675818141885, + "grad_norm": 1.998602271080017, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7217149138450623, + "num_tokens": 215350699.0, + "step": 8461 + }, + { + "epoch": 0.9292773995168021, + "grad_norm": 2.067716598510742, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7213284969329834, + "num_tokens": 215379466.0, + "step": 8462 + }, + { + "epoch": 0.9293872172194158, + "grad_norm": 2.370857000350952, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7232173085212708, + "num_tokens": 215400831.0, + "step": 8463 + }, + { + "epoch": 0.9294970349220294, + "grad_norm": 2.567772388458252, + "learning_rate": 1e-06, + "loss": 0.972, + "mean_token_accuracy": 0.7011883854866028, + "num_tokens": 215420583.0, + "step": 8464 + }, + { + "epoch": 0.929606852624643, + "grad_norm": 1.8265862464904785, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7259063720703125, + "num_tokens": 215454946.0, + "step": 8465 + }, + { + "epoch": 0.9297166703272568, + "grad_norm": 2.062743663787842, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7092779874801636, + "num_tokens": 215484902.0, + "step": 8466 + }, + { + "epoch": 0.9298264880298704, + "grad_norm": 2.172389268875122, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7160772085189819, + "num_tokens": 215509238.0, + "step": 8467 + }, + { + "epoch": 0.9299363057324841, + "grad_norm": 2.30200457572937, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.701351523399353, + "num_tokens": 215531512.0, + "step": 8468 + }, + { + "epoch": 0.9300461234350977, + "grad_norm": 2.2168242931365967, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.721016526222229, + "num_tokens": 215554726.0, + "step": 8469 + }, + { + "epoch": 0.9301559411377114, + "grad_norm": 2.210390567779541, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7038792371749878, + "num_tokens": 215580912.0, + "step": 8470 + }, + { + "epoch": 0.930265758840325, + "grad_norm": 2.0080432891845703, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7162162661552429, + "num_tokens": 215608702.0, + "step": 8471 + }, + { + "epoch": 0.9303755765429387, + "grad_norm": 2.294377326965332, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7052360773086548, + "num_tokens": 215633182.0, + "step": 8472 + }, + { + "epoch": 0.9304853942455524, + "grad_norm": 2.153637647628784, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.6943680644035339, + "num_tokens": 215659809.0, + "step": 8473 + }, + { + "epoch": 0.9305952119481661, + "grad_norm": 2.5497183799743652, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7261598706245422, + "num_tokens": 215677904.0, + "step": 8474 + }, + { + "epoch": 0.9307050296507797, + "grad_norm": 2.157149076461792, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7095850706100464, + "num_tokens": 215703181.0, + "step": 8475 + }, + { + "epoch": 0.9308148473533934, + "grad_norm": 2.59002947807312, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7208747267723083, + "num_tokens": 215721049.0, + "step": 8476 + }, + { + "epoch": 0.930924665056007, + "grad_norm": 2.605663299560547, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7086786031723022, + "num_tokens": 215739641.0, + "step": 8477 + }, + { + "epoch": 0.9310344827586207, + "grad_norm": 2.665609359741211, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7180235385894775, + "num_tokens": 215757496.0, + "step": 8478 + }, + { + "epoch": 0.9311443004612343, + "grad_norm": 2.1580870151519775, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7083758115768433, + "num_tokens": 215785119.0, + "step": 8479 + }, + { + "epoch": 0.9312541181638481, + "grad_norm": 2.6379010677337646, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7184733152389526, + "num_tokens": 215804551.0, + "step": 8480 + }, + { + "epoch": 0.9313639358664617, + "grad_norm": 2.210143566131592, + "learning_rate": 1e-06, + "loss": 1.0153, + "mean_token_accuracy": 0.6947972774505615, + "num_tokens": 215831234.0, + "step": 8481 + }, + { + "epoch": 0.9314737535690754, + "grad_norm": 2.29889178276062, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7023266553878784, + "num_tokens": 215858239.0, + "step": 8482 + }, + { + "epoch": 0.931583571271689, + "grad_norm": 2.4037747383117676, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6980671286582947, + "num_tokens": 215879911.0, + "step": 8483 + }, + { + "epoch": 0.9316933889743026, + "grad_norm": 2.0027356147766113, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7116358876228333, + "num_tokens": 215907661.0, + "step": 8484 + }, + { + "epoch": 0.9318032066769163, + "grad_norm": 2.059326410293579, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7121269702911377, + "num_tokens": 215934534.0, + "step": 8485 + }, + { + "epoch": 0.9319130243795299, + "grad_norm": 2.1349382400512695, + "learning_rate": 1e-06, + "loss": 1.0327, + "mean_token_accuracy": 0.6819204092025757, + "num_tokens": 215961921.0, + "step": 8486 + }, + { + "epoch": 0.9320228420821436, + "grad_norm": 2.11604905128479, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.6986596584320068, + "num_tokens": 215989043.0, + "step": 8487 + }, + { + "epoch": 0.9321326597847573, + "grad_norm": 2.1549434661865234, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6990720629692078, + "num_tokens": 216014683.0, + "step": 8488 + }, + { + "epoch": 0.932242477487371, + "grad_norm": 2.284700632095337, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7043783068656921, + "num_tokens": 216037606.0, + "step": 8489 + }, + { + "epoch": 0.9323522951899846, + "grad_norm": 2.414332866668701, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6962085366249084, + "num_tokens": 216059106.0, + "step": 8490 + }, + { + "epoch": 0.9324621128925983, + "grad_norm": 2.091474771499634, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7099905014038086, + "num_tokens": 216083397.0, + "step": 8491 + }, + { + "epoch": 0.9325719305952119, + "grad_norm": 2.327191114425659, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7021185755729675, + "num_tokens": 216106527.0, + "step": 8492 + }, + { + "epoch": 0.9326817482978256, + "grad_norm": 2.0478439331054688, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7114935517311096, + "num_tokens": 216134013.0, + "step": 8493 + }, + { + "epoch": 0.9327915660004392, + "grad_norm": 2.1050360202789307, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.6964359283447266, + "num_tokens": 216161084.0, + "step": 8494 + }, + { + "epoch": 0.932901383703053, + "grad_norm": 2.003694534301758, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7306013703346252, + "num_tokens": 216188961.0, + "step": 8495 + }, + { + "epoch": 0.9330112014056666, + "grad_norm": 2.2814290523529053, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7007743120193481, + "num_tokens": 216212631.0, + "step": 8496 + }, + { + "epoch": 0.9331210191082803, + "grad_norm": 2.1667943000793457, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.719704270362854, + "num_tokens": 216237688.0, + "step": 8497 + }, + { + "epoch": 0.9332308368108939, + "grad_norm": 2.3395018577575684, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7282745838165283, + "num_tokens": 216260042.0, + "step": 8498 + }, + { + "epoch": 0.9333406545135076, + "grad_norm": 2.359609365463257, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.723387598991394, + "num_tokens": 216282587.0, + "step": 8499 + }, + { + "epoch": 0.9334504722161212, + "grad_norm": 2.255469560623169, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7038863301277161, + "num_tokens": 216307805.0, + "step": 8500 + }, + { + "epoch": 0.9335602899187349, + "grad_norm": 1.9475775957107544, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.6976748704910278, + "num_tokens": 216338207.0, + "step": 8501 + }, + { + "epoch": 0.9336701076213486, + "grad_norm": 2.2426199913024902, + "learning_rate": 1e-06, + "loss": 0.9965, + "mean_token_accuracy": 0.7018284201622009, + "num_tokens": 216361365.0, + "step": 8502 + }, + { + "epoch": 0.9337799253239623, + "grad_norm": 2.085514783859253, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7185742259025574, + "num_tokens": 216387898.0, + "step": 8503 + }, + { + "epoch": 0.9338897430265759, + "grad_norm": 2.2149925231933594, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.7035619020462036, + "num_tokens": 216414354.0, + "step": 8504 + }, + { + "epoch": 0.9339995607291895, + "grad_norm": 2.0582005977630615, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.705197811126709, + "num_tokens": 216440709.0, + "step": 8505 + }, + { + "epoch": 0.9341093784318032, + "grad_norm": 2.393397808074951, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7066250443458557, + "num_tokens": 216463150.0, + "step": 8506 + }, + { + "epoch": 0.9342191961344168, + "grad_norm": 2.140368938446045, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7116435170173645, + "num_tokens": 216489485.0, + "step": 8507 + }, + { + "epoch": 0.9343290138370305, + "grad_norm": 2.646711587905884, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7106015086174011, + "num_tokens": 216508917.0, + "step": 8508 + }, + { + "epoch": 0.9344388315396442, + "grad_norm": 2.3762543201446533, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7239558696746826, + "num_tokens": 216531373.0, + "step": 8509 + }, + { + "epoch": 0.9345486492422579, + "grad_norm": 2.497901439666748, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7140420079231262, + "num_tokens": 216550942.0, + "step": 8510 + }, + { + "epoch": 0.9346584669448715, + "grad_norm": 2.238780975341797, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7153372764587402, + "num_tokens": 216573475.0, + "step": 8511 + }, + { + "epoch": 0.9347682846474852, + "grad_norm": 2.2667548656463623, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.70924973487854, + "num_tokens": 216597134.0, + "step": 8512 + }, + { + "epoch": 0.9348781023500988, + "grad_norm": 2.1746647357940674, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.6978887915611267, + "num_tokens": 216621688.0, + "step": 8513 + }, + { + "epoch": 0.9349879200527125, + "grad_norm": 2.089219808578491, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.703995406627655, + "num_tokens": 216650276.0, + "step": 8514 + }, + { + "epoch": 0.9350977377553261, + "grad_norm": 2.2457592487335205, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7255430221557617, + "num_tokens": 216670771.0, + "step": 8515 + }, + { + "epoch": 0.9352075554579398, + "grad_norm": 1.9028406143188477, + "learning_rate": 1e-06, + "loss": 1.0356, + "mean_token_accuracy": 0.6883934736251831, + "num_tokens": 216702870.0, + "step": 8516 + }, + { + "epoch": 0.9353173731605535, + "grad_norm": 2.129270315170288, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7039154767990112, + "num_tokens": 216728635.0, + "step": 8517 + }, + { + "epoch": 0.9354271908631672, + "grad_norm": 2.3277788162231445, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.712317943572998, + "num_tokens": 216752156.0, + "step": 8518 + }, + { + "epoch": 0.9355370085657808, + "grad_norm": 2.325385093688965, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7146167159080505, + "num_tokens": 216776548.0, + "step": 8519 + }, + { + "epoch": 0.9356468262683945, + "grad_norm": 2.209122657775879, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7053922414779663, + "num_tokens": 216801404.0, + "step": 8520 + }, + { + "epoch": 0.9357566439710081, + "grad_norm": 2.3790602684020996, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7158204317092896, + "num_tokens": 216821935.0, + "step": 8521 + }, + { + "epoch": 0.9358664616736218, + "grad_norm": 2.243387460708618, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7138625383377075, + "num_tokens": 216846529.0, + "step": 8522 + }, + { + "epoch": 0.9359762793762354, + "grad_norm": 2.0466132164001465, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7024326920509338, + "num_tokens": 216874457.0, + "step": 8523 + }, + { + "epoch": 0.9360860970788492, + "grad_norm": 2.562268018722534, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7238770723342896, + "num_tokens": 216891587.0, + "step": 8524 + }, + { + "epoch": 0.9361959147814628, + "grad_norm": 1.9076108932495117, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.721778154373169, + "num_tokens": 216921948.0, + "step": 8525 + }, + { + "epoch": 0.9363057324840764, + "grad_norm": 2.085066318511963, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7089117169380188, + "num_tokens": 216948311.0, + "step": 8526 + }, + { + "epoch": 0.9364155501866901, + "grad_norm": 2.184128522872925, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7260488271713257, + "num_tokens": 216970902.0, + "step": 8527 + }, + { + "epoch": 0.9365253678893037, + "grad_norm": 2.2400882244110107, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7365589141845703, + "num_tokens": 216993488.0, + "step": 8528 + }, + { + "epoch": 0.9366351855919174, + "grad_norm": 2.2581326961517334, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7367987632751465, + "num_tokens": 217014276.0, + "step": 8529 + }, + { + "epoch": 0.936745003294531, + "grad_norm": 2.4332892894744873, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7195636034011841, + "num_tokens": 217035342.0, + "step": 8530 + }, + { + "epoch": 0.9368548209971448, + "grad_norm": 2.1196184158325195, + "learning_rate": 1e-06, + "loss": 1.0226, + "mean_token_accuracy": 0.689916729927063, + "num_tokens": 217063929.0, + "step": 8531 + }, + { + "epoch": 0.9369646386997584, + "grad_norm": 2.0857362747192383, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7007855176925659, + "num_tokens": 217091978.0, + "step": 8532 + }, + { + "epoch": 0.9370744564023721, + "grad_norm": 2.5838186740875244, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7110437154769897, + "num_tokens": 217110643.0, + "step": 8533 + }, + { + "epoch": 0.9371842741049857, + "grad_norm": 2.297586679458618, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.7100303173065186, + "num_tokens": 217133517.0, + "step": 8534 + }, + { + "epoch": 0.9372940918075994, + "grad_norm": 2.1853880882263184, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7064449787139893, + "num_tokens": 217159493.0, + "step": 8535 + }, + { + "epoch": 0.937403909510213, + "grad_norm": 1.9466235637664795, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.720055341720581, + "num_tokens": 217189696.0, + "step": 8536 + }, + { + "epoch": 0.9375137272128267, + "grad_norm": 2.0057013034820557, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.6945536136627197, + "num_tokens": 217218448.0, + "step": 8537 + }, + { + "epoch": 0.9376235449154404, + "grad_norm": 2.0415446758270264, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7033693194389343, + "num_tokens": 217247084.0, + "step": 8538 + }, + { + "epoch": 0.9377333626180541, + "grad_norm": 2.1378841400146484, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.7005441188812256, + "num_tokens": 217273995.0, + "step": 8539 + }, + { + "epoch": 0.9378431803206677, + "grad_norm": 2.094090223312378, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7165615558624268, + "num_tokens": 217300349.0, + "step": 8540 + }, + { + "epoch": 0.9379529980232814, + "grad_norm": 2.22339129447937, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7131459712982178, + "num_tokens": 217324279.0, + "step": 8541 + }, + { + "epoch": 0.938062815725895, + "grad_norm": 2.416212320327759, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7149035334587097, + "num_tokens": 217347218.0, + "step": 8542 + }, + { + "epoch": 0.9381726334285087, + "grad_norm": 2.3341290950775146, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7125700116157532, + "num_tokens": 217371209.0, + "step": 8543 + }, + { + "epoch": 0.9382824511311223, + "grad_norm": 2.0482683181762695, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.7003151178359985, + "num_tokens": 217400484.0, + "step": 8544 + }, + { + "epoch": 0.9383922688337359, + "grad_norm": 2.095566511154175, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7113412618637085, + "num_tokens": 217426176.0, + "step": 8545 + }, + { + "epoch": 0.9385020865363497, + "grad_norm": 2.09909987449646, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7082065343856812, + "num_tokens": 217452733.0, + "step": 8546 + }, + { + "epoch": 0.9386119042389633, + "grad_norm": 1.898341178894043, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.692997932434082, + "num_tokens": 217483986.0, + "step": 8547 + }, + { + "epoch": 0.938721721941577, + "grad_norm": 2.3049163818359375, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7043960690498352, + "num_tokens": 217508893.0, + "step": 8548 + }, + { + "epoch": 0.9388315396441906, + "grad_norm": 2.3045825958251953, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.711523711681366, + "num_tokens": 217531850.0, + "step": 8549 + }, + { + "epoch": 0.9389413573468043, + "grad_norm": 2.1568150520324707, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.6961780190467834, + "num_tokens": 217558667.0, + "step": 8550 + }, + { + "epoch": 0.9390511750494179, + "grad_norm": 2.1052656173706055, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7340872287750244, + "num_tokens": 217584720.0, + "step": 8551 + }, + { + "epoch": 0.9391609927520316, + "grad_norm": 2.3621747493743896, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7258053421974182, + "num_tokens": 217607403.0, + "step": 8552 + }, + { + "epoch": 0.9392708104546453, + "grad_norm": 2.636638641357422, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7007228136062622, + "num_tokens": 217627860.0, + "step": 8553 + }, + { + "epoch": 0.939380628157259, + "grad_norm": 2.7233755588531494, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.6979866027832031, + "num_tokens": 217645853.0, + "step": 8554 + }, + { + "epoch": 0.9394904458598726, + "grad_norm": 2.2070491313934326, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7084714770317078, + "num_tokens": 217671637.0, + "step": 8555 + }, + { + "epoch": 0.9396002635624863, + "grad_norm": 2.1859028339385986, + "learning_rate": 1e-06, + "loss": 1.0405, + "mean_token_accuracy": 0.6923685073852539, + "num_tokens": 217697336.0, + "step": 8556 + }, + { + "epoch": 0.9397100812650999, + "grad_norm": 2.0931859016418457, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7108029127120972, + "num_tokens": 217725925.0, + "step": 8557 + }, + { + "epoch": 0.9398198989677136, + "grad_norm": 2.153259515762329, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6907274127006531, + "num_tokens": 217751318.0, + "step": 8558 + }, + { + "epoch": 0.9399297166703272, + "grad_norm": 2.565326690673828, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7251673936843872, + "num_tokens": 217770366.0, + "step": 8559 + }, + { + "epoch": 0.940039534372941, + "grad_norm": 2.239546775817871, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.7025073766708374, + "num_tokens": 217794352.0, + "step": 8560 + }, + { + "epoch": 0.9401493520755546, + "grad_norm": 2.1535961627960205, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7176672220230103, + "num_tokens": 217819985.0, + "step": 8561 + }, + { + "epoch": 0.9402591697781683, + "grad_norm": 2.039769411087036, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6969804763793945, + "num_tokens": 217849863.0, + "step": 8562 + }, + { + "epoch": 0.9403689874807819, + "grad_norm": 2.38114595413208, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7194740176200867, + "num_tokens": 217871526.0, + "step": 8563 + }, + { + "epoch": 0.9404788051833955, + "grad_norm": 2.016793966293335, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7087876200675964, + "num_tokens": 217899125.0, + "step": 8564 + }, + { + "epoch": 0.9405886228860092, + "grad_norm": 2.0242738723754883, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7105282545089722, + "num_tokens": 217929162.0, + "step": 8565 + }, + { + "epoch": 0.9406984405886228, + "grad_norm": 2.301602840423584, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7249847650527954, + "num_tokens": 217951112.0, + "step": 8566 + }, + { + "epoch": 0.9408082582912366, + "grad_norm": 2.230692148208618, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7239798307418823, + "num_tokens": 217974350.0, + "step": 8567 + }, + { + "epoch": 0.9409180759938502, + "grad_norm": 1.9848533868789673, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7079415321350098, + "num_tokens": 218004236.0, + "step": 8568 + }, + { + "epoch": 0.9410278936964639, + "grad_norm": 2.1018409729003906, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7041902542114258, + "num_tokens": 218032460.0, + "step": 8569 + }, + { + "epoch": 0.9411377113990775, + "grad_norm": 2.3962533473968506, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7232571840286255, + "num_tokens": 218053613.0, + "step": 8570 + }, + { + "epoch": 0.9412475291016912, + "grad_norm": 2.1124186515808105, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7254981398582458, + "num_tokens": 218080397.0, + "step": 8571 + }, + { + "epoch": 0.9413573468043048, + "grad_norm": 2.260591506958008, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6946811676025391, + "num_tokens": 218105799.0, + "step": 8572 + }, + { + "epoch": 0.9414671645069185, + "grad_norm": 2.2408690452575684, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7340977191925049, + "num_tokens": 218127545.0, + "step": 8573 + }, + { + "epoch": 0.9415769822095322, + "grad_norm": 2.0744292736053467, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.714449405670166, + "num_tokens": 218157096.0, + "step": 8574 + }, + { + "epoch": 0.9416867999121459, + "grad_norm": 2.0886993408203125, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.6993275880813599, + "num_tokens": 218184994.0, + "step": 8575 + }, + { + "epoch": 0.9417966176147595, + "grad_norm": 2.086862087249756, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7087118029594421, + "num_tokens": 218214019.0, + "step": 8576 + }, + { + "epoch": 0.9419064353173732, + "grad_norm": 2.087965965270996, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.704833984375, + "num_tokens": 218240960.0, + "step": 8577 + }, + { + "epoch": 0.9420162530199868, + "grad_norm": 2.509127140045166, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7183994054794312, + "num_tokens": 218260846.0, + "step": 8578 + }, + { + "epoch": 0.9421260707226005, + "grad_norm": 2.175724744796753, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7243540287017822, + "num_tokens": 218287476.0, + "step": 8579 + }, + { + "epoch": 0.9422358884252141, + "grad_norm": 2.3811869621276855, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7141786813735962, + "num_tokens": 218309883.0, + "step": 8580 + }, + { + "epoch": 0.9423457061278278, + "grad_norm": 2.145047187805176, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.711517333984375, + "num_tokens": 218335249.0, + "step": 8581 + }, + { + "epoch": 0.9424555238304415, + "grad_norm": 2.1631951332092285, + "learning_rate": 1e-06, + "loss": 1.0537, + "mean_token_accuracy": 0.6850000619888306, + "num_tokens": 218361778.0, + "step": 8582 + }, + { + "epoch": 0.9425653415330552, + "grad_norm": 2.625807523727417, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7158644795417786, + "num_tokens": 218379755.0, + "step": 8583 + }, + { + "epoch": 0.9426751592356688, + "grad_norm": 2.050245523452759, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.690833330154419, + "num_tokens": 218408521.0, + "step": 8584 + }, + { + "epoch": 0.9427849769382824, + "grad_norm": 2.371793508529663, + "learning_rate": 1e-06, + "loss": 0.7909, + "mean_token_accuracy": 0.7509536743164062, + "num_tokens": 218429220.0, + "step": 8585 + }, + { + "epoch": 0.9428947946408961, + "grad_norm": 1.9474952220916748, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.7002609968185425, + "num_tokens": 218461731.0, + "step": 8586 + }, + { + "epoch": 0.9430046123435097, + "grad_norm": 2.142352342605591, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7071729302406311, + "num_tokens": 218486309.0, + "step": 8587 + }, + { + "epoch": 0.9431144300461234, + "grad_norm": 2.3849539756774902, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7194539308547974, + "num_tokens": 218508769.0, + "step": 8588 + }, + { + "epoch": 0.9432242477487371, + "grad_norm": 2.4850947856903076, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.737816572189331, + "num_tokens": 218527518.0, + "step": 8589 + }, + { + "epoch": 0.9433340654513508, + "grad_norm": 2.583009958267212, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7185477018356323, + "num_tokens": 218549528.0, + "step": 8590 + }, + { + "epoch": 0.9434438831539644, + "grad_norm": 2.2328360080718994, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7023407220840454, + "num_tokens": 218573241.0, + "step": 8591 + }, + { + "epoch": 0.9435537008565781, + "grad_norm": 2.209568977355957, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7044244408607483, + "num_tokens": 218596469.0, + "step": 8592 + }, + { + "epoch": 0.9436635185591917, + "grad_norm": 2.1885809898376465, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7316220998764038, + "num_tokens": 218619931.0, + "step": 8593 + }, + { + "epoch": 0.9437733362618054, + "grad_norm": 1.8466472625732422, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6960228681564331, + "num_tokens": 218653664.0, + "step": 8594 + }, + { + "epoch": 0.943883153964419, + "grad_norm": 2.0388872623443604, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.691161036491394, + "num_tokens": 218681060.0, + "step": 8595 + }, + { + "epoch": 0.9439929716670328, + "grad_norm": 2.221576452255249, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7154014706611633, + "num_tokens": 218705599.0, + "step": 8596 + }, + { + "epoch": 0.9441027893696464, + "grad_norm": 1.7952370643615723, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7230346202850342, + "num_tokens": 218736922.0, + "step": 8597 + }, + { + "epoch": 0.9442126070722601, + "grad_norm": 2.233208179473877, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7312873005867004, + "num_tokens": 218762763.0, + "step": 8598 + }, + { + "epoch": 0.9443224247748737, + "grad_norm": 2.2628304958343506, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7025726437568665, + "num_tokens": 218785809.0, + "step": 8599 + }, + { + "epoch": 0.9444322424774874, + "grad_norm": 2.209151029586792, + "learning_rate": 1e-06, + "loss": 1.0118, + "mean_token_accuracy": 0.6961736679077148, + "num_tokens": 218812751.0, + "step": 8600 + }, + { + "epoch": 0.944542060180101, + "grad_norm": 2.2969698905944824, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.7004091739654541, + "num_tokens": 218835467.0, + "step": 8601 + }, + { + "epoch": 0.9446518778827147, + "grad_norm": 2.08490252494812, + "learning_rate": 1e-06, + "loss": 0.9975, + "mean_token_accuracy": 0.707004725933075, + "num_tokens": 218863457.0, + "step": 8602 + }, + { + "epoch": 0.9447616955853284, + "grad_norm": 2.2727394104003906, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7168014049530029, + "num_tokens": 218887545.0, + "step": 8603 + }, + { + "epoch": 0.944871513287942, + "grad_norm": 2.1974620819091797, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7226828336715698, + "num_tokens": 218911535.0, + "step": 8604 + }, + { + "epoch": 0.9449813309905557, + "grad_norm": 2.075732707977295, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.6987322568893433, + "num_tokens": 218938138.0, + "step": 8605 + }, + { + "epoch": 0.9450911486931693, + "grad_norm": 2.119920253753662, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6933101415634155, + "num_tokens": 218966361.0, + "step": 8606 + }, + { + "epoch": 0.945200966395783, + "grad_norm": 2.178229331970215, + "learning_rate": 1e-06, + "loss": 1.0501, + "mean_token_accuracy": 0.6823903322219849, + "num_tokens": 218994193.0, + "step": 8607 + }, + { + "epoch": 0.9453107840983966, + "grad_norm": 1.783835768699646, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.701029896736145, + "num_tokens": 219027520.0, + "step": 8608 + }, + { + "epoch": 0.9454206018010103, + "grad_norm": 2.1413724422454834, + "learning_rate": 1e-06, + "loss": 1.0292, + "mean_token_accuracy": 0.6827317476272583, + "num_tokens": 219055375.0, + "step": 8609 + }, + { + "epoch": 0.9455304195036239, + "grad_norm": 2.1647636890411377, + "learning_rate": 1e-06, + "loss": 1.0393, + "mean_token_accuracy": 0.6827452182769775, + "num_tokens": 219081733.0, + "step": 8610 + }, + { + "epoch": 0.9456402372062377, + "grad_norm": 1.878906488418579, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7112119197845459, + "num_tokens": 219112289.0, + "step": 8611 + }, + { + "epoch": 0.9457500549088513, + "grad_norm": 2.0344974994659424, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7306972742080688, + "num_tokens": 219136714.0, + "step": 8612 + }, + { + "epoch": 0.945859872611465, + "grad_norm": 2.2331595420837402, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7030813694000244, + "num_tokens": 219160232.0, + "step": 8613 + }, + { + "epoch": 0.9459696903140786, + "grad_norm": 2.8328022956848145, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7138703465461731, + "num_tokens": 219176616.0, + "step": 8614 + }, + { + "epoch": 0.9460795080166923, + "grad_norm": 2.178790330886841, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.6889590620994568, + "num_tokens": 219200269.0, + "step": 8615 + }, + { + "epoch": 0.9461893257193059, + "grad_norm": 2.2265827655792236, + "learning_rate": 1e-06, + "loss": 1.0207, + "mean_token_accuracy": 0.6950702667236328, + "num_tokens": 219225231.0, + "step": 8616 + }, + { + "epoch": 0.9462991434219196, + "grad_norm": 2.382113218307495, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7202547192573547, + "num_tokens": 219245292.0, + "step": 8617 + }, + { + "epoch": 0.9464089611245333, + "grad_norm": 2.042447328567505, + "learning_rate": 1e-06, + "loss": 0.9899, + "mean_token_accuracy": 0.6990160346031189, + "num_tokens": 219272657.0, + "step": 8618 + }, + { + "epoch": 0.946518778827147, + "grad_norm": 2.193436861038208, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7228057384490967, + "num_tokens": 219296753.0, + "step": 8619 + }, + { + "epoch": 0.9466285965297606, + "grad_norm": 2.047380208969116, + "learning_rate": 1e-06, + "loss": 1.0315, + "mean_token_accuracy": 0.6827735900878906, + "num_tokens": 219325206.0, + "step": 8620 + }, + { + "epoch": 0.9467384142323743, + "grad_norm": 1.823483943939209, + "learning_rate": 1e-06, + "loss": 1.0586, + "mean_token_accuracy": 0.6854532957077026, + "num_tokens": 219362696.0, + "step": 8621 + }, + { + "epoch": 0.9468482319349879, + "grad_norm": 2.584803342819214, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7020944356918335, + "num_tokens": 219383712.0, + "step": 8622 + }, + { + "epoch": 0.9469580496376016, + "grad_norm": 2.170386791229248, + "learning_rate": 1e-06, + "loss": 1.0337, + "mean_token_accuracy": 0.6884865760803223, + "num_tokens": 219411746.0, + "step": 8623 + }, + { + "epoch": 0.9470678673402152, + "grad_norm": 2.136021375656128, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7138473987579346, + "num_tokens": 219438035.0, + "step": 8624 + }, + { + "epoch": 0.947177685042829, + "grad_norm": 2.1937482357025146, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7186646461486816, + "num_tokens": 219463744.0, + "step": 8625 + }, + { + "epoch": 0.9472875027454426, + "grad_norm": 2.1129722595214844, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7028003931045532, + "num_tokens": 219491707.0, + "step": 8626 + }, + { + "epoch": 0.9473973204480562, + "grad_norm": 2.273226022720337, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7333182692527771, + "num_tokens": 219514072.0, + "step": 8627 + }, + { + "epoch": 0.9475071381506699, + "grad_norm": 2.2533159255981445, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7194597125053406, + "num_tokens": 219538153.0, + "step": 8628 + }, + { + "epoch": 0.9476169558532835, + "grad_norm": 2.2160191535949707, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7140548229217529, + "num_tokens": 219564478.0, + "step": 8629 + }, + { + "epoch": 0.9477267735558972, + "grad_norm": 2.1934566497802734, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7072715759277344, + "num_tokens": 219590739.0, + "step": 8630 + }, + { + "epoch": 0.9478365912585108, + "grad_norm": 2.159893035888672, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6975565552711487, + "num_tokens": 219616906.0, + "step": 8631 + }, + { + "epoch": 0.9479464089611246, + "grad_norm": 2.5972487926483154, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.711753785610199, + "num_tokens": 219636381.0, + "step": 8632 + }, + { + "epoch": 0.9480562266637382, + "grad_norm": 2.593499183654785, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7260867953300476, + "num_tokens": 219656158.0, + "step": 8633 + }, + { + "epoch": 0.9481660443663519, + "grad_norm": 2.0630435943603516, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7170703411102295, + "num_tokens": 219681973.0, + "step": 8634 + }, + { + "epoch": 0.9482758620689655, + "grad_norm": 2.2285304069519043, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7131927609443665, + "num_tokens": 219706038.0, + "step": 8635 + }, + { + "epoch": 0.9483856797715792, + "grad_norm": 2.381032705307007, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7119336128234863, + "num_tokens": 219727602.0, + "step": 8636 + }, + { + "epoch": 0.9484954974741928, + "grad_norm": 2.3504581451416016, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7345609664916992, + "num_tokens": 219749040.0, + "step": 8637 + }, + { + "epoch": 0.9486053151768065, + "grad_norm": 2.2115604877471924, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7038859128952026, + "num_tokens": 219774947.0, + "step": 8638 + }, + { + "epoch": 0.9487151328794201, + "grad_norm": 2.1568715572357178, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7279064059257507, + "num_tokens": 219798732.0, + "step": 8639 + }, + { + "epoch": 0.9488249505820339, + "grad_norm": 2.6641042232513428, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7423557043075562, + "num_tokens": 219817121.0, + "step": 8640 + }, + { + "epoch": 0.9489347682846475, + "grad_norm": 3.5453877449035645, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.6992936730384827, + "num_tokens": 219838050.0, + "step": 8641 + }, + { + "epoch": 0.9490445859872612, + "grad_norm": 2.3674354553222656, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.708338737487793, + "num_tokens": 219858931.0, + "step": 8642 + }, + { + "epoch": 0.9491544036898748, + "grad_norm": 2.0755999088287354, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7218976020812988, + "num_tokens": 219884436.0, + "step": 8643 + }, + { + "epoch": 0.9492642213924884, + "grad_norm": 2.0826544761657715, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.706400990486145, + "num_tokens": 219914136.0, + "step": 8644 + }, + { + "epoch": 0.9493740390951021, + "grad_norm": 2.127422571182251, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.6985988616943359, + "num_tokens": 219941118.0, + "step": 8645 + }, + { + "epoch": 0.9494838567977157, + "grad_norm": 2.2274582386016846, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7180262804031372, + "num_tokens": 219963421.0, + "step": 8646 + }, + { + "epoch": 0.9495936745003295, + "grad_norm": 2.180670976638794, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7045371532440186, + "num_tokens": 219988784.0, + "step": 8647 + }, + { + "epoch": 0.9497034922029431, + "grad_norm": 2.3892323970794678, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7068995237350464, + "num_tokens": 220010531.0, + "step": 8648 + }, + { + "epoch": 0.9498133099055568, + "grad_norm": 2.0185959339141846, + "learning_rate": 1e-06, + "loss": 1.0269, + "mean_token_accuracy": 0.6875077486038208, + "num_tokens": 220038544.0, + "step": 8649 + }, + { + "epoch": 0.9499231276081704, + "grad_norm": 1.9864214658737183, + "learning_rate": 1e-06, + "loss": 1.0129, + "mean_token_accuracy": 0.7032003402709961, + "num_tokens": 220068122.0, + "step": 8650 + }, + { + "epoch": 0.9500329453107841, + "grad_norm": 2.3003368377685547, + "learning_rate": 1e-06, + "loss": 1.0059, + "mean_token_accuracy": 0.6975165605545044, + "num_tokens": 220092870.0, + "step": 8651 + }, + { + "epoch": 0.9501427630133977, + "grad_norm": 2.302480936050415, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7163807153701782, + "num_tokens": 220113246.0, + "step": 8652 + }, + { + "epoch": 0.9502525807160114, + "grad_norm": 2.260746955871582, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7097709774971008, + "num_tokens": 220135988.0, + "step": 8653 + }, + { + "epoch": 0.9503623984186251, + "grad_norm": 1.8440593481063843, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.6990936398506165, + "num_tokens": 220168549.0, + "step": 8654 + }, + { + "epoch": 0.9504722161212388, + "grad_norm": 2.086418628692627, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7077701687812805, + "num_tokens": 220196201.0, + "step": 8655 + }, + { + "epoch": 0.9505820338238524, + "grad_norm": 2.2206451892852783, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.6985822916030884, + "num_tokens": 220221562.0, + "step": 8656 + }, + { + "epoch": 0.9506918515264661, + "grad_norm": 1.9258794784545898, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6979743838310242, + "num_tokens": 220253174.0, + "step": 8657 + }, + { + "epoch": 0.9508016692290797, + "grad_norm": 2.1264073848724365, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.6992970108985901, + "num_tokens": 220279054.0, + "step": 8658 + }, + { + "epoch": 0.9509114869316934, + "grad_norm": 2.296755313873291, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7299630045890808, + "num_tokens": 220301118.0, + "step": 8659 + }, + { + "epoch": 0.951021304634307, + "grad_norm": 2.285100221633911, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7203514575958252, + "num_tokens": 220323510.0, + "step": 8660 + }, + { + "epoch": 0.9511311223369208, + "grad_norm": 2.1079905033111572, + "learning_rate": 1e-06, + "loss": 1.0247, + "mean_token_accuracy": 0.6959859728813171, + "num_tokens": 220349336.0, + "step": 8661 + }, + { + "epoch": 0.9512409400395344, + "grad_norm": 2.175933361053467, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7104347944259644, + "num_tokens": 220371878.0, + "step": 8662 + }, + { + "epoch": 0.9513507577421481, + "grad_norm": 2.168980598449707, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7049506902694702, + "num_tokens": 220396769.0, + "step": 8663 + }, + { + "epoch": 0.9514605754447617, + "grad_norm": 2.5468859672546387, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7297024130821228, + "num_tokens": 220414800.0, + "step": 8664 + }, + { + "epoch": 0.9515703931473753, + "grad_norm": 2.179255247116089, + "learning_rate": 1e-06, + "loss": 1.034, + "mean_token_accuracy": 0.6886094212532043, + "num_tokens": 220440995.0, + "step": 8665 + }, + { + "epoch": 0.951680210849989, + "grad_norm": 2.090235710144043, + "learning_rate": 1e-06, + "loss": 1.0545, + "mean_token_accuracy": 0.683881402015686, + "num_tokens": 220469076.0, + "step": 8666 + }, + { + "epoch": 0.9517900285526026, + "grad_norm": 2.4062247276306152, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7062653303146362, + "num_tokens": 220493602.0, + "step": 8667 + }, + { + "epoch": 0.9518998462552163, + "grad_norm": 2.1883544921875, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7117094397544861, + "num_tokens": 220518635.0, + "step": 8668 + }, + { + "epoch": 0.95200966395783, + "grad_norm": 2.3058197498321533, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7103273868560791, + "num_tokens": 220541617.0, + "step": 8669 + }, + { + "epoch": 0.9521194816604437, + "grad_norm": 2.110194444656372, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7091248035430908, + "num_tokens": 220569454.0, + "step": 8670 + }, + { + "epoch": 0.9522292993630573, + "grad_norm": 2.0461220741271973, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7146416902542114, + "num_tokens": 220598464.0, + "step": 8671 + }, + { + "epoch": 0.952339117065671, + "grad_norm": 2.261009931564331, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.7079833149909973, + "num_tokens": 220623900.0, + "step": 8672 + }, + { + "epoch": 0.9524489347682846, + "grad_norm": 2.190723180770874, + "learning_rate": 1e-06, + "loss": 1.0168, + "mean_token_accuracy": 0.6991630792617798, + "num_tokens": 220651341.0, + "step": 8673 + }, + { + "epoch": 0.9525587524708983, + "grad_norm": 2.2853314876556396, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7060608267784119, + "num_tokens": 220672794.0, + "step": 8674 + }, + { + "epoch": 0.9526685701735119, + "grad_norm": 2.061786413192749, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7273268103599548, + "num_tokens": 220696883.0, + "step": 8675 + }, + { + "epoch": 0.9527783878761257, + "grad_norm": 1.8926148414611816, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.6980181932449341, + "num_tokens": 220729532.0, + "step": 8676 + }, + { + "epoch": 0.9528882055787393, + "grad_norm": 2.3998208045959473, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7262144088745117, + "num_tokens": 220750761.0, + "step": 8677 + }, + { + "epoch": 0.952998023281353, + "grad_norm": 2.4471771717071533, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7261340022087097, + "num_tokens": 220771457.0, + "step": 8678 + }, + { + "epoch": 0.9531078409839666, + "grad_norm": 1.8156254291534424, + "learning_rate": 1e-06, + "loss": 1.0326, + "mean_token_accuracy": 0.6923693418502808, + "num_tokens": 220806267.0, + "step": 8679 + }, + { + "epoch": 0.9532176586865803, + "grad_norm": 2.167388439178467, + "learning_rate": 1e-06, + "loss": 1.0838, + "mean_token_accuracy": 0.6705356240272522, + "num_tokens": 220835333.0, + "step": 8680 + }, + { + "epoch": 0.9533274763891939, + "grad_norm": 1.9187732934951782, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7068462371826172, + "num_tokens": 220867025.0, + "step": 8681 + }, + { + "epoch": 0.9534372940918076, + "grad_norm": 2.253509759902954, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7397345304489136, + "num_tokens": 220892324.0, + "step": 8682 + }, + { + "epoch": 0.9535471117944213, + "grad_norm": 2.055828332901001, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.6894246339797974, + "num_tokens": 220920297.0, + "step": 8683 + }, + { + "epoch": 0.953656929497035, + "grad_norm": 2.1516573429107666, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.701418399810791, + "num_tokens": 220946909.0, + "step": 8684 + }, + { + "epoch": 0.9537667471996486, + "grad_norm": 2.048874855041504, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.7046066522598267, + "num_tokens": 220974138.0, + "step": 8685 + }, + { + "epoch": 0.9538765649022622, + "grad_norm": 2.089770555496216, + "learning_rate": 1e-06, + "loss": 1.0139, + "mean_token_accuracy": 0.6968467235565186, + "num_tokens": 221002661.0, + "step": 8686 + }, + { + "epoch": 0.9539863826048759, + "grad_norm": 2.1283719539642334, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7117098569869995, + "num_tokens": 221028921.0, + "step": 8687 + }, + { + "epoch": 0.9540962003074895, + "grad_norm": 2.1775898933410645, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7016181945800781, + "num_tokens": 221054161.0, + "step": 8688 + }, + { + "epoch": 0.9542060180101032, + "grad_norm": 2.0862925052642822, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.6944372653961182, + "num_tokens": 221082164.0, + "step": 8689 + }, + { + "epoch": 0.9543158357127169, + "grad_norm": 2.5995707511901855, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7100538015365601, + "num_tokens": 221101677.0, + "step": 8690 + }, + { + "epoch": 0.9544256534153306, + "grad_norm": 2.2583343982696533, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.697307825088501, + "num_tokens": 221125406.0, + "step": 8691 + }, + { + "epoch": 0.9545354711179442, + "grad_norm": 2.666541576385498, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7207130789756775, + "num_tokens": 221143385.0, + "step": 8692 + }, + { + "epoch": 0.9546452888205579, + "grad_norm": 2.302224636077881, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7326737642288208, + "num_tokens": 221164836.0, + "step": 8693 + }, + { + "epoch": 0.9547551065231715, + "grad_norm": 2.503875970840454, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7097702026367188, + "num_tokens": 221186255.0, + "step": 8694 + }, + { + "epoch": 0.9548649242257852, + "grad_norm": 2.28546404838562, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7120364904403687, + "num_tokens": 221210968.0, + "step": 8695 + }, + { + "epoch": 0.9549747419283988, + "grad_norm": 2.3112006187438965, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7179802060127258, + "num_tokens": 221232624.0, + "step": 8696 + }, + { + "epoch": 0.9550845596310125, + "grad_norm": 2.432612657546997, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7214081883430481, + "num_tokens": 221254492.0, + "step": 8697 + }, + { + "epoch": 0.9551943773336262, + "grad_norm": 2.3894364833831787, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7267466187477112, + "num_tokens": 221276922.0, + "step": 8698 + }, + { + "epoch": 0.9553041950362399, + "grad_norm": 2.1699399948120117, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6909955739974976, + "num_tokens": 221304290.0, + "step": 8699 + }, + { + "epoch": 0.9554140127388535, + "grad_norm": 2.3468430042266846, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7301917672157288, + "num_tokens": 221327389.0, + "step": 8700 + }, + { + "epoch": 0.9555238304414672, + "grad_norm": 2.2883994579315186, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7099087834358215, + "num_tokens": 221348963.0, + "step": 8701 + }, + { + "epoch": 0.9556336481440808, + "grad_norm": 2.3433499336242676, + "learning_rate": 1e-06, + "loss": 1.0768, + "mean_token_accuracy": 0.6844273209571838, + "num_tokens": 221373549.0, + "step": 8702 + }, + { + "epoch": 0.9557434658466945, + "grad_norm": 2.0809359550476074, + "learning_rate": 1e-06, + "loss": 1.0227, + "mean_token_accuracy": 0.6843752861022949, + "num_tokens": 221400728.0, + "step": 8703 + }, + { + "epoch": 0.9558532835493081, + "grad_norm": 2.159954309463501, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.6877858638763428, + "num_tokens": 221428415.0, + "step": 8704 + }, + { + "epoch": 0.9559631012519219, + "grad_norm": 2.134467601776123, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7192827463150024, + "num_tokens": 221456063.0, + "step": 8705 + }, + { + "epoch": 0.9560729189545355, + "grad_norm": 2.0442235469818115, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7164432406425476, + "num_tokens": 221482614.0, + "step": 8706 + }, + { + "epoch": 0.9561827366571491, + "grad_norm": 1.9868783950805664, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7237691879272461, + "num_tokens": 221510973.0, + "step": 8707 + }, + { + "epoch": 0.9562925543597628, + "grad_norm": 2.205627679824829, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7191134691238403, + "num_tokens": 221534542.0, + "step": 8708 + }, + { + "epoch": 0.9564023720623764, + "grad_norm": 1.9588996171951294, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7150612473487854, + "num_tokens": 221564549.0, + "step": 8709 + }, + { + "epoch": 0.9565121897649901, + "grad_norm": 2.1865196228027344, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7170330286026001, + "num_tokens": 221589396.0, + "step": 8710 + }, + { + "epoch": 0.9566220074676037, + "grad_norm": 1.902254581451416, + "learning_rate": 1e-06, + "loss": 1.0901, + "mean_token_accuracy": 0.6732584834098816, + "num_tokens": 221624101.0, + "step": 8711 + }, + { + "epoch": 0.9567318251702175, + "grad_norm": 2.1849381923675537, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7467929124832153, + "num_tokens": 221646146.0, + "step": 8712 + }, + { + "epoch": 0.9568416428728311, + "grad_norm": 2.3858702182769775, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7139009237289429, + "num_tokens": 221668000.0, + "step": 8713 + }, + { + "epoch": 0.9569514605754448, + "grad_norm": 2.194964647293091, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.6981246471405029, + "num_tokens": 221694229.0, + "step": 8714 + }, + { + "epoch": 0.9570612782780584, + "grad_norm": 2.280522584915161, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7255043983459473, + "num_tokens": 221717288.0, + "step": 8715 + }, + { + "epoch": 0.9571710959806721, + "grad_norm": 2.009545087814331, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7414235472679138, + "num_tokens": 221744612.0, + "step": 8716 + }, + { + "epoch": 0.9572809136832857, + "grad_norm": 2.1890482902526855, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7218358516693115, + "num_tokens": 221769668.0, + "step": 8717 + }, + { + "epoch": 0.9573907313858994, + "grad_norm": 2.470024347305298, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7099596261978149, + "num_tokens": 221792160.0, + "step": 8718 + }, + { + "epoch": 0.9575005490885131, + "grad_norm": 2.2481539249420166, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6937165260314941, + "num_tokens": 221816581.0, + "step": 8719 + }, + { + "epoch": 0.9576103667911268, + "grad_norm": 1.9954562187194824, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7103410363197327, + "num_tokens": 221844014.0, + "step": 8720 + }, + { + "epoch": 0.9577201844937404, + "grad_norm": 1.9805752038955688, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7104415893554688, + "num_tokens": 221871298.0, + "step": 8721 + }, + { + "epoch": 0.9578300021963541, + "grad_norm": 2.151195764541626, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.724729061126709, + "num_tokens": 221894254.0, + "step": 8722 + }, + { + "epoch": 0.9579398198989677, + "grad_norm": 2.1989965438842773, + "learning_rate": 1e-06, + "loss": 1.031, + "mean_token_accuracy": 0.7014293074607849, + "num_tokens": 221920755.0, + "step": 8723 + }, + { + "epoch": 0.9580496376015813, + "grad_norm": 1.9204014539718628, + "learning_rate": 1e-06, + "loss": 1.0481, + "mean_token_accuracy": 0.6821269392967224, + "num_tokens": 221954459.0, + "step": 8724 + }, + { + "epoch": 0.958159455304195, + "grad_norm": 2.2628209590911865, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7249637246131897, + "num_tokens": 221976792.0, + "step": 8725 + }, + { + "epoch": 0.9582692730068088, + "grad_norm": 2.243112325668335, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6898469924926758, + "num_tokens": 222004367.0, + "step": 8726 + }, + { + "epoch": 0.9583790907094224, + "grad_norm": 2.075631618499756, + "learning_rate": 1e-06, + "loss": 1.0309, + "mean_token_accuracy": 0.6848414540290833, + "num_tokens": 222032495.0, + "step": 8727 + }, + { + "epoch": 0.958488908412036, + "grad_norm": 2.1198527812957764, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7239545583724976, + "num_tokens": 222056754.0, + "step": 8728 + }, + { + "epoch": 0.9585987261146497, + "grad_norm": 2.1116838455200195, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7206698060035706, + "num_tokens": 222083817.0, + "step": 8729 + }, + { + "epoch": 0.9587085438172633, + "grad_norm": 2.145845890045166, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7063146829605103, + "num_tokens": 222111826.0, + "step": 8730 + }, + { + "epoch": 0.958818361519877, + "grad_norm": 2.113856554031372, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7027633786201477, + "num_tokens": 222137924.0, + "step": 8731 + }, + { + "epoch": 0.9589281792224906, + "grad_norm": 2.304241895675659, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7065724730491638, + "num_tokens": 222162746.0, + "step": 8732 + }, + { + "epoch": 0.9590379969251043, + "grad_norm": 2.05806827545166, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7151404619216919, + "num_tokens": 222191367.0, + "step": 8733 + }, + { + "epoch": 0.959147814627718, + "grad_norm": 2.078136682510376, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7102891206741333, + "num_tokens": 222217187.0, + "step": 8734 + }, + { + "epoch": 0.9592576323303317, + "grad_norm": 2.2404980659484863, + "learning_rate": 1e-06, + "loss": 1.0077, + "mean_token_accuracy": 0.6983652114868164, + "num_tokens": 222240683.0, + "step": 8735 + }, + { + "epoch": 0.9593674500329453, + "grad_norm": 1.883312463760376, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.6942439079284668, + "num_tokens": 222276124.0, + "step": 8736 + }, + { + "epoch": 0.959477267735559, + "grad_norm": 1.9365195035934448, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.6985621452331543, + "num_tokens": 222308011.0, + "step": 8737 + }, + { + "epoch": 0.9595870854381726, + "grad_norm": 1.9283357858657837, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7280917167663574, + "num_tokens": 222337780.0, + "step": 8738 + }, + { + "epoch": 0.9596969031407863, + "grad_norm": 1.8898818492889404, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6859546899795532, + "num_tokens": 222369427.0, + "step": 8739 + }, + { + "epoch": 0.9598067208433999, + "grad_norm": 2.398571014404297, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7188182473182678, + "num_tokens": 222393162.0, + "step": 8740 + }, + { + "epoch": 0.9599165385460137, + "grad_norm": 2.475029468536377, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7212896943092346, + "num_tokens": 222413995.0, + "step": 8741 + }, + { + "epoch": 0.9600263562486273, + "grad_norm": 2.219972848892212, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7060761451721191, + "num_tokens": 222438110.0, + "step": 8742 + }, + { + "epoch": 0.960136173951241, + "grad_norm": 2.2538294792175293, + "learning_rate": 1e-06, + "loss": 1.0002, + "mean_token_accuracy": 0.6990511417388916, + "num_tokens": 222464762.0, + "step": 8743 + }, + { + "epoch": 0.9602459916538546, + "grad_norm": 2.2597243785858154, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7266802191734314, + "num_tokens": 222486631.0, + "step": 8744 + }, + { + "epoch": 0.9603558093564682, + "grad_norm": 2.2052786350250244, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7321124076843262, + "num_tokens": 222511054.0, + "step": 8745 + }, + { + "epoch": 0.9604656270590819, + "grad_norm": 2.4212067127227783, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7271575927734375, + "num_tokens": 222529820.0, + "step": 8746 + }, + { + "epoch": 0.9605754447616955, + "grad_norm": 2.3052899837493896, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.717071533203125, + "num_tokens": 222552909.0, + "step": 8747 + }, + { + "epoch": 0.9606852624643093, + "grad_norm": 2.2049386501312256, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7022868990898132, + "num_tokens": 222578993.0, + "step": 8748 + }, + { + "epoch": 0.9607950801669229, + "grad_norm": 2.179201126098633, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7033560276031494, + "num_tokens": 222605211.0, + "step": 8749 + }, + { + "epoch": 0.9609048978695366, + "grad_norm": 2.302032470703125, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.707710862159729, + "num_tokens": 222627860.0, + "step": 8750 + }, + { + "epoch": 0.9610147155721502, + "grad_norm": 2.2857961654663086, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7188799381256104, + "num_tokens": 222649838.0, + "step": 8751 + }, + { + "epoch": 0.9611245332747639, + "grad_norm": 2.336798667907715, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6967275738716125, + "num_tokens": 222673115.0, + "step": 8752 + }, + { + "epoch": 0.9612343509773775, + "grad_norm": 2.310443639755249, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7251519560813904, + "num_tokens": 222695357.0, + "step": 8753 + }, + { + "epoch": 0.9613441686799912, + "grad_norm": 2.396615743637085, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7162410616874695, + "num_tokens": 222716879.0, + "step": 8754 + }, + { + "epoch": 0.9614539863826049, + "grad_norm": 2.220533847808838, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7120985984802246, + "num_tokens": 222740346.0, + "step": 8755 + }, + { + "epoch": 0.9615638040852186, + "grad_norm": 1.9752212762832642, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7496169805526733, + "num_tokens": 222766228.0, + "step": 8756 + }, + { + "epoch": 0.9616736217878322, + "grad_norm": 2.0021636486053467, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7095773816108704, + "num_tokens": 222796451.0, + "step": 8757 + }, + { + "epoch": 0.9617834394904459, + "grad_norm": 2.10097599029541, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.6960741877555847, + "num_tokens": 222820888.0, + "step": 8758 + }, + { + "epoch": 0.9618932571930595, + "grad_norm": 2.1918671131134033, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.6895875930786133, + "num_tokens": 222846873.0, + "step": 8759 + }, + { + "epoch": 0.9620030748956732, + "grad_norm": 1.9200241565704346, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7233861088752747, + "num_tokens": 222874862.0, + "step": 8760 + }, + { + "epoch": 0.9621128925982868, + "grad_norm": 2.2673113346099854, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7124677896499634, + "num_tokens": 222898435.0, + "step": 8761 + }, + { + "epoch": 0.9622227103009005, + "grad_norm": 1.9088281393051147, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7044743299484253, + "num_tokens": 222931134.0, + "step": 8762 + }, + { + "epoch": 0.9623325280035142, + "grad_norm": 2.1373450756073, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6905001997947693, + "num_tokens": 222956890.0, + "step": 8763 + }, + { + "epoch": 0.9624423457061279, + "grad_norm": 2.230159282684326, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.705476701259613, + "num_tokens": 222981110.0, + "step": 8764 + }, + { + "epoch": 0.9625521634087415, + "grad_norm": 2.1766538619995117, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7314313650131226, + "num_tokens": 223005515.0, + "step": 8765 + }, + { + "epoch": 0.9626619811113551, + "grad_norm": 2.350083351135254, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.6963691711425781, + "num_tokens": 223028699.0, + "step": 8766 + }, + { + "epoch": 0.9627717988139688, + "grad_norm": 2.4055187702178955, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7239716053009033, + "num_tokens": 223049934.0, + "step": 8767 + }, + { + "epoch": 0.9628816165165824, + "grad_norm": 2.1190102100372314, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7025945782661438, + "num_tokens": 223076502.0, + "step": 8768 + }, + { + "epoch": 0.9629914342191961, + "grad_norm": 1.9178344011306763, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7102110385894775, + "num_tokens": 223108767.0, + "step": 8769 + }, + { + "epoch": 0.9631012519218098, + "grad_norm": 2.0833280086517334, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7175253629684448, + "num_tokens": 223135716.0, + "step": 8770 + }, + { + "epoch": 0.9632110696244235, + "grad_norm": 2.2909300327301025, + "learning_rate": 1e-06, + "loss": 1.0626, + "mean_token_accuracy": 0.6834582686424255, + "num_tokens": 223161563.0, + "step": 8771 + }, + { + "epoch": 0.9633208873270371, + "grad_norm": 2.0495076179504395, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7093280553817749, + "num_tokens": 223190819.0, + "step": 8772 + }, + { + "epoch": 0.9634307050296508, + "grad_norm": 2.3671276569366455, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7103737592697144, + "num_tokens": 223213457.0, + "step": 8773 + }, + { + "epoch": 0.9635405227322644, + "grad_norm": 2.0612306594848633, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7053802013397217, + "num_tokens": 223240646.0, + "step": 8774 + }, + { + "epoch": 0.9636503404348781, + "grad_norm": 2.052037000656128, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7017402052879333, + "num_tokens": 223267969.0, + "step": 8775 + }, + { + "epoch": 0.9637601581374917, + "grad_norm": 2.193058967590332, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7161138653755188, + "num_tokens": 223290772.0, + "step": 8776 + }, + { + "epoch": 0.9638699758401055, + "grad_norm": 2.364208698272705, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.715880274772644, + "num_tokens": 223313053.0, + "step": 8777 + }, + { + "epoch": 0.9639797935427191, + "grad_norm": 2.3836746215820312, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.7059671878814697, + "num_tokens": 223336870.0, + "step": 8778 + }, + { + "epoch": 0.9640896112453328, + "grad_norm": 2.367914915084839, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7181868553161621, + "num_tokens": 223358167.0, + "step": 8779 + }, + { + "epoch": 0.9641994289479464, + "grad_norm": 1.9410388469696045, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6881985664367676, + "num_tokens": 223390442.0, + "step": 8780 + }, + { + "epoch": 0.9643092466505601, + "grad_norm": 1.8909538984298706, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7113874554634094, + "num_tokens": 223421434.0, + "step": 8781 + }, + { + "epoch": 0.9644190643531737, + "grad_norm": 2.4285237789154053, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7059645652770996, + "num_tokens": 223443634.0, + "step": 8782 + }, + { + "epoch": 0.9645288820557874, + "grad_norm": 2.5833797454833984, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.7008750438690186, + "num_tokens": 223464433.0, + "step": 8783 + }, + { + "epoch": 0.9646386997584011, + "grad_norm": 2.431854248046875, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7155845165252686, + "num_tokens": 223485853.0, + "step": 8784 + }, + { + "epoch": 0.9647485174610148, + "grad_norm": 2.2178475856781006, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.701318621635437, + "num_tokens": 223510401.0, + "step": 8785 + }, + { + "epoch": 0.9648583351636284, + "grad_norm": 2.1379833221435547, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7028422355651855, + "num_tokens": 223538145.0, + "step": 8786 + }, + { + "epoch": 0.964968152866242, + "grad_norm": 2.3123533725738525, + "learning_rate": 1e-06, + "loss": 0.9898, + "mean_token_accuracy": 0.7054797410964966, + "num_tokens": 223561792.0, + "step": 8787 + }, + { + "epoch": 0.9650779705688557, + "grad_norm": 2.2781577110290527, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7143280506134033, + "num_tokens": 223583706.0, + "step": 8788 + }, + { + "epoch": 0.9651877882714693, + "grad_norm": 2.1636366844177246, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7138646841049194, + "num_tokens": 223608415.0, + "step": 8789 + }, + { + "epoch": 0.965297605974083, + "grad_norm": 2.2737274169921875, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7111883163452148, + "num_tokens": 223632364.0, + "step": 8790 + }, + { + "epoch": 0.9654074236766966, + "grad_norm": 2.413759708404541, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7075322866439819, + "num_tokens": 223654010.0, + "step": 8791 + }, + { + "epoch": 0.9655172413793104, + "grad_norm": 2.554490566253662, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7481178045272827, + "num_tokens": 223673155.0, + "step": 8792 + }, + { + "epoch": 0.965627059081924, + "grad_norm": 2.355294942855835, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7085915803909302, + "num_tokens": 223695970.0, + "step": 8793 + }, + { + "epoch": 0.9657368767845377, + "grad_norm": 2.080465793609619, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7133764028549194, + "num_tokens": 223724845.0, + "step": 8794 + }, + { + "epoch": 0.9658466944871513, + "grad_norm": 2.067199468612671, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.710777223110199, + "num_tokens": 223752647.0, + "step": 8795 + }, + { + "epoch": 0.965956512189765, + "grad_norm": 2.1956050395965576, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.7030972242355347, + "num_tokens": 223779483.0, + "step": 8796 + }, + { + "epoch": 0.9660663298923786, + "grad_norm": 2.3277580738067627, + "learning_rate": 1e-06, + "loss": 0.864, + "mean_token_accuracy": 0.7336509227752686, + "num_tokens": 223800621.0, + "step": 8797 + }, + { + "epoch": 0.9661761475949923, + "grad_norm": 2.4799857139587402, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7264155149459839, + "num_tokens": 223819424.0, + "step": 8798 + }, + { + "epoch": 0.966285965297606, + "grad_norm": 2.1129019260406494, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7056599855422974, + "num_tokens": 223846620.0, + "step": 8799 + }, + { + "epoch": 0.9663957830002197, + "grad_norm": 2.0793564319610596, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7317136526107788, + "num_tokens": 223870980.0, + "step": 8800 + }, + { + "epoch": 0.9665056007028333, + "grad_norm": 2.403550148010254, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7051589488983154, + "num_tokens": 223893316.0, + "step": 8801 + }, + { + "epoch": 0.966615418405447, + "grad_norm": 2.1693661212921143, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7111678123474121, + "num_tokens": 223917627.0, + "step": 8802 + }, + { + "epoch": 0.9667252361080606, + "grad_norm": 2.3718578815460205, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.725891649723053, + "num_tokens": 223939106.0, + "step": 8803 + }, + { + "epoch": 0.9668350538106742, + "grad_norm": 2.5124263763427734, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7209564447402954, + "num_tokens": 223958345.0, + "step": 8804 + }, + { + "epoch": 0.9669448715132879, + "grad_norm": 2.258044481277466, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7307643890380859, + "num_tokens": 223980238.0, + "step": 8805 + }, + { + "epoch": 0.9670546892159017, + "grad_norm": 2.181682825088501, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7332566976547241, + "num_tokens": 224003325.0, + "step": 8806 + }, + { + "epoch": 0.9671645069185153, + "grad_norm": 2.0944488048553467, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7061656713485718, + "num_tokens": 224030264.0, + "step": 8807 + }, + { + "epoch": 0.967274324621129, + "grad_norm": 2.069423198699951, + "learning_rate": 1e-06, + "loss": 1.0191, + "mean_token_accuracy": 0.689203679561615, + "num_tokens": 224060560.0, + "step": 8808 + }, + { + "epoch": 0.9673841423237426, + "grad_norm": 2.28519606590271, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7116467952728271, + "num_tokens": 224084023.0, + "step": 8809 + }, + { + "epoch": 0.9674939600263562, + "grad_norm": 2.177628755569458, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7160778045654297, + "num_tokens": 224107472.0, + "step": 8810 + }, + { + "epoch": 0.9676037777289699, + "grad_norm": 1.9915980100631714, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7298190593719482, + "num_tokens": 224134750.0, + "step": 8811 + }, + { + "epoch": 0.9677135954315835, + "grad_norm": 2.5021817684173584, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7154725193977356, + "num_tokens": 224154415.0, + "step": 8812 + }, + { + "epoch": 0.9678234131341973, + "grad_norm": 2.10994553565979, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.716017484664917, + "num_tokens": 224182596.0, + "step": 8813 + }, + { + "epoch": 0.9679332308368109, + "grad_norm": 2.1452512741088867, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7283690571784973, + "num_tokens": 224209122.0, + "step": 8814 + }, + { + "epoch": 0.9680430485394246, + "grad_norm": 2.285569906234741, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7089979648590088, + "num_tokens": 224234061.0, + "step": 8815 + }, + { + "epoch": 0.9681528662420382, + "grad_norm": 2.250000476837158, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7224356532096863, + "num_tokens": 224256647.0, + "step": 8816 + }, + { + "epoch": 0.9682626839446519, + "grad_norm": 2.2759010791778564, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7058011293411255, + "num_tokens": 224279279.0, + "step": 8817 + }, + { + "epoch": 0.9683725016472655, + "grad_norm": 2.104367971420288, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.72173011302948, + "num_tokens": 224306675.0, + "step": 8818 + }, + { + "epoch": 0.9684823193498792, + "grad_norm": 2.327204704284668, + "learning_rate": 1e-06, + "loss": 1.0555, + "mean_token_accuracy": 0.6901112794876099, + "num_tokens": 224331256.0, + "step": 8819 + }, + { + "epoch": 0.9685921370524928, + "grad_norm": 2.001579523086548, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.717312753200531, + "num_tokens": 224358683.0, + "step": 8820 + }, + { + "epoch": 0.9687019547551066, + "grad_norm": 2.268767833709717, + "learning_rate": 1e-06, + "loss": 1.0419, + "mean_token_accuracy": 0.6858161091804504, + "num_tokens": 224381619.0, + "step": 8821 + }, + { + "epoch": 0.9688117724577202, + "grad_norm": 2.158055305480957, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7169934511184692, + "num_tokens": 224405512.0, + "step": 8822 + }, + { + "epoch": 0.9689215901603339, + "grad_norm": 2.2545247077941895, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7317632436752319, + "num_tokens": 224428278.0, + "step": 8823 + }, + { + "epoch": 0.9690314078629475, + "grad_norm": 2.495185136795044, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7203889489173889, + "num_tokens": 224449121.0, + "step": 8824 + }, + { + "epoch": 0.9691412255655611, + "grad_norm": 2.231471061706543, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6940488219261169, + "num_tokens": 224474483.0, + "step": 8825 + }, + { + "epoch": 0.9692510432681748, + "grad_norm": 2.1166300773620605, + "learning_rate": 1e-06, + "loss": 1.029, + "mean_token_accuracy": 0.6843181848526001, + "num_tokens": 224502162.0, + "step": 8826 + }, + { + "epoch": 0.9693608609707884, + "grad_norm": 2.2081236839294434, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7152736186981201, + "num_tokens": 224526616.0, + "step": 8827 + }, + { + "epoch": 0.9694706786734022, + "grad_norm": 2.101064920425415, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7431171536445618, + "num_tokens": 224551041.0, + "step": 8828 + }, + { + "epoch": 0.9695804963760158, + "grad_norm": 2.0234997272491455, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7258065342903137, + "num_tokens": 224579651.0, + "step": 8829 + }, + { + "epoch": 0.9696903140786295, + "grad_norm": 2.1017279624938965, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7113138437271118, + "num_tokens": 224606433.0, + "step": 8830 + }, + { + "epoch": 0.9698001317812431, + "grad_norm": 2.263519287109375, + "learning_rate": 1e-06, + "loss": 1.0043, + "mean_token_accuracy": 0.6927818059921265, + "num_tokens": 224632357.0, + "step": 8831 + }, + { + "epoch": 0.9699099494838568, + "grad_norm": 2.0629119873046875, + "learning_rate": 1e-06, + "loss": 1.0237, + "mean_token_accuracy": 0.6944724321365356, + "num_tokens": 224663633.0, + "step": 8832 + }, + { + "epoch": 0.9700197671864704, + "grad_norm": 2.516026735305786, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7202563285827637, + "num_tokens": 224684043.0, + "step": 8833 + }, + { + "epoch": 0.9701295848890841, + "grad_norm": 2.1381051540374756, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7167701721191406, + "num_tokens": 224708496.0, + "step": 8834 + }, + { + "epoch": 0.9702394025916978, + "grad_norm": 2.3154783248901367, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7330405712127686, + "num_tokens": 224730874.0, + "step": 8835 + }, + { + "epoch": 0.9703492202943115, + "grad_norm": 2.0654332637786865, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.7038889527320862, + "num_tokens": 224759603.0, + "step": 8836 + }, + { + "epoch": 0.9704590379969251, + "grad_norm": 1.9229812622070312, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.7047709226608276, + "num_tokens": 224790674.0, + "step": 8837 + }, + { + "epoch": 0.9705688556995388, + "grad_norm": 2.0992064476013184, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7252706289291382, + "num_tokens": 224816863.0, + "step": 8838 + }, + { + "epoch": 0.9706786734021524, + "grad_norm": 2.3129959106445312, + "learning_rate": 1e-06, + "loss": 1.0232, + "mean_token_accuracy": 0.6909113526344299, + "num_tokens": 224840943.0, + "step": 8839 + }, + { + "epoch": 0.9707884911047661, + "grad_norm": 2.064265489578247, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7094411253929138, + "num_tokens": 224868251.0, + "step": 8840 + }, + { + "epoch": 0.9708983088073797, + "grad_norm": 2.1872150897979736, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7112137079238892, + "num_tokens": 224893835.0, + "step": 8841 + }, + { + "epoch": 0.9710081265099935, + "grad_norm": 2.1526572704315186, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.703156054019928, + "num_tokens": 224918808.0, + "step": 8842 + }, + { + "epoch": 0.9711179442126071, + "grad_norm": 1.8944491147994995, + "learning_rate": 1e-06, + "loss": 1.0863, + "mean_token_accuracy": 0.6857928037643433, + "num_tokens": 224952609.0, + "step": 8843 + }, + { + "epoch": 0.9712277619152208, + "grad_norm": 1.8684252500534058, + "learning_rate": 1e-06, + "loss": 1.0397, + "mean_token_accuracy": 0.6919321417808533, + "num_tokens": 224985955.0, + "step": 8844 + }, + { + "epoch": 0.9713375796178344, + "grad_norm": 2.304638624191284, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7120555639266968, + "num_tokens": 225008387.0, + "step": 8845 + }, + { + "epoch": 0.971447397320448, + "grad_norm": 2.2724099159240723, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.7032625079154968, + "num_tokens": 225032601.0, + "step": 8846 + }, + { + "epoch": 0.9715572150230617, + "grad_norm": 2.1161046028137207, + "learning_rate": 1e-06, + "loss": 1.0095, + "mean_token_accuracy": 0.69436115026474, + "num_tokens": 225059772.0, + "step": 8847 + }, + { + "epoch": 0.9716670327256753, + "grad_norm": 2.4066312313079834, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7140944600105286, + "num_tokens": 225081193.0, + "step": 8848 + }, + { + "epoch": 0.971776850428289, + "grad_norm": 1.9244766235351562, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7214010953903198, + "num_tokens": 225110388.0, + "step": 8849 + }, + { + "epoch": 0.9718866681309027, + "grad_norm": 2.475170612335205, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7276034951210022, + "num_tokens": 225129198.0, + "step": 8850 + }, + { + "epoch": 0.9719964858335164, + "grad_norm": 1.9743350744247437, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7043987512588501, + "num_tokens": 225161538.0, + "step": 8851 + }, + { + "epoch": 0.97210630353613, + "grad_norm": 2.035003900527954, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.6948847770690918, + "num_tokens": 225187094.0, + "step": 8852 + }, + { + "epoch": 0.9722161212387437, + "grad_norm": 1.7637670040130615, + "learning_rate": 1e-06, + "loss": 1.0853, + "mean_token_accuracy": 0.6801148056983948, + "num_tokens": 225223370.0, + "step": 8853 + }, + { + "epoch": 0.9723259389413573, + "grad_norm": 2.390624523162842, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7165429592132568, + "num_tokens": 225244217.0, + "step": 8854 + }, + { + "epoch": 0.972435756643971, + "grad_norm": 2.1822986602783203, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7415772080421448, + "num_tokens": 225269153.0, + "step": 8855 + }, + { + "epoch": 0.9725455743465846, + "grad_norm": 2.1462738513946533, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.735654354095459, + "num_tokens": 225293092.0, + "step": 8856 + }, + { + "epoch": 0.9726553920491984, + "grad_norm": 1.8304952383041382, + "learning_rate": 1e-06, + "loss": 1.055, + "mean_token_accuracy": 0.6890382766723633, + "num_tokens": 225327587.0, + "step": 8857 + }, + { + "epoch": 0.972765209751812, + "grad_norm": 2.2106950283050537, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.706731379032135, + "num_tokens": 225353755.0, + "step": 8858 + }, + { + "epoch": 0.9728750274544257, + "grad_norm": 2.2765464782714844, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.7017383575439453, + "num_tokens": 225378003.0, + "step": 8859 + }, + { + "epoch": 0.9729848451570393, + "grad_norm": 2.155284881591797, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7073051929473877, + "num_tokens": 225404528.0, + "step": 8860 + }, + { + "epoch": 0.973094662859653, + "grad_norm": 2.2193470001220703, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7180947661399841, + "num_tokens": 225427695.0, + "step": 8861 + }, + { + "epoch": 0.9732044805622666, + "grad_norm": 2.127739906311035, + "learning_rate": 1e-06, + "loss": 1.0639, + "mean_token_accuracy": 0.678627073764801, + "num_tokens": 225455720.0, + "step": 8862 + }, + { + "epoch": 0.9733142982648803, + "grad_norm": 2.4512884616851807, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7304279804229736, + "num_tokens": 225475721.0, + "step": 8863 + }, + { + "epoch": 0.973424115967494, + "grad_norm": 2.386272668838501, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7023522853851318, + "num_tokens": 225498818.0, + "step": 8864 + }, + { + "epoch": 0.9735339336701077, + "grad_norm": 2.176069736480713, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.6968183517456055, + "num_tokens": 225524276.0, + "step": 8865 + }, + { + "epoch": 0.9736437513727213, + "grad_norm": 2.0932390689849854, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7043272852897644, + "num_tokens": 225550091.0, + "step": 8866 + }, + { + "epoch": 0.973753569075335, + "grad_norm": 2.1206634044647217, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6953501105308533, + "num_tokens": 225578034.0, + "step": 8867 + }, + { + "epoch": 0.9738633867779486, + "grad_norm": 2.1759274005889893, + "learning_rate": 1e-06, + "loss": 1.0157, + "mean_token_accuracy": 0.6845718622207642, + "num_tokens": 225604173.0, + "step": 8868 + }, + { + "epoch": 0.9739732044805622, + "grad_norm": 2.2663984298706055, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7132068872451782, + "num_tokens": 225628512.0, + "step": 8869 + }, + { + "epoch": 0.9740830221831759, + "grad_norm": 2.350123167037964, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7070302367210388, + "num_tokens": 225649104.0, + "step": 8870 + }, + { + "epoch": 0.9741928398857896, + "grad_norm": 2.3266372680664062, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7078536748886108, + "num_tokens": 225672965.0, + "step": 8871 + }, + { + "epoch": 0.9743026575884033, + "grad_norm": 2.3059632778167725, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7344857454299927, + "num_tokens": 225695844.0, + "step": 8872 + }, + { + "epoch": 0.9744124752910169, + "grad_norm": 2.101257801055908, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6928746104240417, + "num_tokens": 225722675.0, + "step": 8873 + }, + { + "epoch": 0.9745222929936306, + "grad_norm": 2.2378549575805664, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7179838418960571, + "num_tokens": 225745691.0, + "step": 8874 + }, + { + "epoch": 0.9746321106962442, + "grad_norm": 2.36917781829834, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7334303259849548, + "num_tokens": 225766912.0, + "step": 8875 + }, + { + "epoch": 0.9747419283988579, + "grad_norm": 2.188340902328491, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6990007162094116, + "num_tokens": 225792546.0, + "step": 8876 + }, + { + "epoch": 0.9748517461014715, + "grad_norm": 2.042790174484253, + "learning_rate": 1e-06, + "loss": 1.0084, + "mean_token_accuracy": 0.6889692544937134, + "num_tokens": 225820494.0, + "step": 8877 + }, + { + "epoch": 0.9749615638040853, + "grad_norm": 2.412085771560669, + "learning_rate": 1e-06, + "loss": 1.0345, + "mean_token_accuracy": 0.6953921914100647, + "num_tokens": 225842144.0, + "step": 8878 + }, + { + "epoch": 0.9750713815066989, + "grad_norm": 2.5027825832366943, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.713062047958374, + "num_tokens": 225862335.0, + "step": 8879 + }, + { + "epoch": 0.9751811992093126, + "grad_norm": 2.3358778953552246, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7141938209533691, + "num_tokens": 225882836.0, + "step": 8880 + }, + { + "epoch": 0.9752910169119262, + "grad_norm": 2.0417187213897705, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7150541543960571, + "num_tokens": 225910753.0, + "step": 8881 + }, + { + "epoch": 0.9754008346145399, + "grad_norm": 2.3589279651641846, + "learning_rate": 1e-06, + "loss": 0.9671, + "mean_token_accuracy": 0.7024524211883545, + "num_tokens": 225934516.0, + "step": 8882 + }, + { + "epoch": 0.9755106523171535, + "grad_norm": 2.113741636276245, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.710202693939209, + "num_tokens": 225961502.0, + "step": 8883 + }, + { + "epoch": 0.9756204700197671, + "grad_norm": 2.508418560028076, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7053305506706238, + "num_tokens": 225981895.0, + "step": 8884 + }, + { + "epoch": 0.9757302877223808, + "grad_norm": 2.3122305870056152, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7174221277236938, + "num_tokens": 226006074.0, + "step": 8885 + }, + { + "epoch": 0.9758401054249946, + "grad_norm": 2.227166175842285, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6920993328094482, + "num_tokens": 226029535.0, + "step": 8886 + }, + { + "epoch": 0.9759499231276082, + "grad_norm": 1.915941596031189, + "learning_rate": 1e-06, + "loss": 1.0316, + "mean_token_accuracy": 0.6850128173828125, + "num_tokens": 226063462.0, + "step": 8887 + }, + { + "epoch": 0.9760597408302218, + "grad_norm": 1.9908419847488403, + "learning_rate": 1e-06, + "loss": 0.9756, + "mean_token_accuracy": 0.7154303789138794, + "num_tokens": 226094284.0, + "step": 8888 + }, + { + "epoch": 0.9761695585328355, + "grad_norm": 1.9666374921798706, + "learning_rate": 1e-06, + "loss": 1.0154, + "mean_token_accuracy": 0.6965872049331665, + "num_tokens": 226125153.0, + "step": 8889 + }, + { + "epoch": 0.9762793762354491, + "grad_norm": 2.0596811771392822, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6957810521125793, + "num_tokens": 226155399.0, + "step": 8890 + }, + { + "epoch": 0.9763891939380628, + "grad_norm": 2.238828659057617, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7275553345680237, + "num_tokens": 226178760.0, + "step": 8891 + }, + { + "epoch": 0.9764990116406764, + "grad_norm": 2.0211009979248047, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7170464992523193, + "num_tokens": 226206582.0, + "step": 8892 + }, + { + "epoch": 0.9766088293432902, + "grad_norm": 2.4149677753448486, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7279513478279114, + "num_tokens": 226229194.0, + "step": 8893 + }, + { + "epoch": 0.9767186470459038, + "grad_norm": 2.0881307125091553, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7154432535171509, + "num_tokens": 226255539.0, + "step": 8894 + }, + { + "epoch": 0.9768284647485175, + "grad_norm": 2.29166841506958, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7301964163780212, + "num_tokens": 226277902.0, + "step": 8895 + }, + { + "epoch": 0.9769382824511311, + "grad_norm": 2.134711265563965, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7173290252685547, + "num_tokens": 226304006.0, + "step": 8896 + }, + { + "epoch": 0.9770481001537448, + "grad_norm": 2.0504629611968994, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7255094051361084, + "num_tokens": 226332440.0, + "step": 8897 + }, + { + "epoch": 0.9771579178563584, + "grad_norm": 1.9987114667892456, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7012166976928711, + "num_tokens": 226360525.0, + "step": 8898 + }, + { + "epoch": 0.9772677355589721, + "grad_norm": 2.1510202884674072, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7011950016021729, + "num_tokens": 226387130.0, + "step": 8899 + }, + { + "epoch": 0.9773775532615858, + "grad_norm": 2.23758602142334, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.711609959602356, + "num_tokens": 226411143.0, + "step": 8900 + }, + { + "epoch": 0.9774873709641995, + "grad_norm": 2.292679786682129, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7070803642272949, + "num_tokens": 226434040.0, + "step": 8901 + }, + { + "epoch": 0.9775971886668131, + "grad_norm": 2.168632745742798, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7136191725730896, + "num_tokens": 226460311.0, + "step": 8902 + }, + { + "epoch": 0.9777070063694268, + "grad_norm": 2.269307851791382, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7035267353057861, + "num_tokens": 226484600.0, + "step": 8903 + }, + { + "epoch": 0.9778168240720404, + "grad_norm": 1.9184880256652832, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7184651494026184, + "num_tokens": 226515001.0, + "step": 8904 + }, + { + "epoch": 0.977926641774654, + "grad_norm": 2.4228827953338623, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7139058113098145, + "num_tokens": 226536815.0, + "step": 8905 + }, + { + "epoch": 0.9780364594772677, + "grad_norm": 2.2960541248321533, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.707030713558197, + "num_tokens": 226559914.0, + "step": 8906 + }, + { + "epoch": 0.9781462771798815, + "grad_norm": 1.9251335859298706, + "learning_rate": 1e-06, + "loss": 1.051, + "mean_token_accuracy": 0.6820623874664307, + "num_tokens": 226592342.0, + "step": 8907 + }, + { + "epoch": 0.9782560948824951, + "grad_norm": 2.8846230506896973, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.711989164352417, + "num_tokens": 226607856.0, + "step": 8908 + }, + { + "epoch": 0.9783659125851087, + "grad_norm": 2.000495195388794, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7244978547096252, + "num_tokens": 226636846.0, + "step": 8909 + }, + { + "epoch": 0.9784757302877224, + "grad_norm": 2.2279348373413086, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7224745750427246, + "num_tokens": 226660451.0, + "step": 8910 + }, + { + "epoch": 0.978585547990336, + "grad_norm": 1.920171856880188, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.694679856300354, + "num_tokens": 226694353.0, + "step": 8911 + }, + { + "epoch": 0.9786953656929497, + "grad_norm": 2.114151954650879, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6838342547416687, + "num_tokens": 226722289.0, + "step": 8912 + }, + { + "epoch": 0.9788051833955633, + "grad_norm": 2.1747076511383057, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7147480845451355, + "num_tokens": 226747829.0, + "step": 8913 + }, + { + "epoch": 0.978915001098177, + "grad_norm": 2.1712093353271484, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7337173223495483, + "num_tokens": 226771105.0, + "step": 8914 + }, + { + "epoch": 0.9790248188007907, + "grad_norm": 2.021725654602051, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.70401930809021, + "num_tokens": 226800211.0, + "step": 8915 + }, + { + "epoch": 0.9791346365034044, + "grad_norm": 2.231910467147827, + "learning_rate": 1e-06, + "loss": 1.0124, + "mean_token_accuracy": 0.688399076461792, + "num_tokens": 226824770.0, + "step": 8916 + }, + { + "epoch": 0.979244454206018, + "grad_norm": 2.21923565864563, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7004884481430054, + "num_tokens": 226850343.0, + "step": 8917 + }, + { + "epoch": 0.9793542719086317, + "grad_norm": 1.8396281003952026, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7329431176185608, + "num_tokens": 226882860.0, + "step": 8918 + }, + { + "epoch": 0.9794640896112453, + "grad_norm": 1.9207950830459595, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7006456851959229, + "num_tokens": 226915213.0, + "step": 8919 + }, + { + "epoch": 0.979573907313859, + "grad_norm": 2.067821502685547, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7184541821479797, + "num_tokens": 226941426.0, + "step": 8920 + }, + { + "epoch": 0.9796837250164726, + "grad_norm": 2.2099177837371826, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7022032141685486, + "num_tokens": 226967271.0, + "step": 8921 + }, + { + "epoch": 0.9797935427190864, + "grad_norm": 2.1519455909729004, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7020875215530396, + "num_tokens": 226993105.0, + "step": 8922 + }, + { + "epoch": 0.9799033604217, + "grad_norm": 2.081507921218872, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7003099322319031, + "num_tokens": 227021903.0, + "step": 8923 + }, + { + "epoch": 0.9800131781243137, + "grad_norm": 2.029611825942993, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7220560312271118, + "num_tokens": 227048613.0, + "step": 8924 + }, + { + "epoch": 0.9801229958269273, + "grad_norm": 2.297635316848755, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7261502742767334, + "num_tokens": 227070744.0, + "step": 8925 + }, + { + "epoch": 0.980232813529541, + "grad_norm": 1.9539769887924194, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.7111525535583496, + "num_tokens": 227097196.0, + "step": 8926 + }, + { + "epoch": 0.9803426312321546, + "grad_norm": 2.1170530319213867, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7102861404418945, + "num_tokens": 227121734.0, + "step": 8927 + }, + { + "epoch": 0.9804524489347682, + "grad_norm": 1.9696117639541626, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.721217691898346, + "num_tokens": 227151152.0, + "step": 8928 + }, + { + "epoch": 0.980562266637382, + "grad_norm": 2.0873117446899414, + "learning_rate": 1e-06, + "loss": 1.0284, + "mean_token_accuracy": 0.6916439533233643, + "num_tokens": 227178297.0, + "step": 8929 + }, + { + "epoch": 0.9806720843399956, + "grad_norm": 2.3298120498657227, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.698729395866394, + "num_tokens": 227200555.0, + "step": 8930 + }, + { + "epoch": 0.9807819020426093, + "grad_norm": 2.3336410522460938, + "learning_rate": 1e-06, + "loss": 1.0323, + "mean_token_accuracy": 0.6851297616958618, + "num_tokens": 227223978.0, + "step": 8931 + }, + { + "epoch": 0.9808917197452229, + "grad_norm": 2.224132537841797, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7051132321357727, + "num_tokens": 227246590.0, + "step": 8932 + }, + { + "epoch": 0.9810015374478366, + "grad_norm": 2.24432110786438, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.7006607055664062, + "num_tokens": 227271417.0, + "step": 8933 + }, + { + "epoch": 0.9811113551504502, + "grad_norm": 2.2206852436065674, + "learning_rate": 1e-06, + "loss": 1.0253, + "mean_token_accuracy": 0.6951544880867004, + "num_tokens": 227297768.0, + "step": 8934 + }, + { + "epoch": 0.9812211728530639, + "grad_norm": 2.217881441116333, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7080861330032349, + "num_tokens": 227323273.0, + "step": 8935 + }, + { + "epoch": 0.9813309905556776, + "grad_norm": 2.2712059020996094, + "learning_rate": 1e-06, + "loss": 0.9934, + "mean_token_accuracy": 0.6961748600006104, + "num_tokens": 227347924.0, + "step": 8936 + }, + { + "epoch": 0.9814408082582913, + "grad_norm": 2.4681687355041504, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7160866856575012, + "num_tokens": 227368270.0, + "step": 8937 + }, + { + "epoch": 0.9815506259609049, + "grad_norm": 2.4995102882385254, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.730118989944458, + "num_tokens": 227388356.0, + "step": 8938 + }, + { + "epoch": 0.9816604436635186, + "grad_norm": 2.2498114109039307, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6938393115997314, + "num_tokens": 227413227.0, + "step": 8939 + }, + { + "epoch": 0.9817702613661322, + "grad_norm": 2.556753158569336, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7255055904388428, + "num_tokens": 227431768.0, + "step": 8940 + }, + { + "epoch": 0.9818800790687459, + "grad_norm": 2.1272449493408203, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7288011312484741, + "num_tokens": 227456057.0, + "step": 8941 + }, + { + "epoch": 0.9819898967713595, + "grad_norm": 2.312653064727783, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7317174077033997, + "num_tokens": 227479096.0, + "step": 8942 + }, + { + "epoch": 0.9820997144739732, + "grad_norm": 2.0291037559509277, + "learning_rate": 1e-06, + "loss": 1.0424, + "mean_token_accuracy": 0.6892691850662231, + "num_tokens": 227507806.0, + "step": 8943 + }, + { + "epoch": 0.9822095321765869, + "grad_norm": 2.6061830520629883, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7098367214202881, + "num_tokens": 227527616.0, + "step": 8944 + }, + { + "epoch": 0.9823193498792006, + "grad_norm": 2.1085400581359863, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7002021670341492, + "num_tokens": 227557242.0, + "step": 8945 + }, + { + "epoch": 0.9824291675818142, + "grad_norm": 2.3590431213378906, + "learning_rate": 1e-06, + "loss": 1.0259, + "mean_token_accuracy": 0.6939342021942139, + "num_tokens": 227581029.0, + "step": 8946 + }, + { + "epoch": 0.9825389852844278, + "grad_norm": 2.485894203186035, + "learning_rate": 1e-06, + "loss": 0.8149, + "mean_token_accuracy": 0.7399704456329346, + "num_tokens": 227599621.0, + "step": 8947 + }, + { + "epoch": 0.9826488029870415, + "grad_norm": 2.077576160430908, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7244608402252197, + "num_tokens": 227626270.0, + "step": 8948 + }, + { + "epoch": 0.9827586206896551, + "grad_norm": 2.160431146621704, + "learning_rate": 1e-06, + "loss": 1.0271, + "mean_token_accuracy": 0.6870691180229187, + "num_tokens": 227652887.0, + "step": 8949 + }, + { + "epoch": 0.9828684383922688, + "grad_norm": 2.4123685359954834, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.6982071995735168, + "num_tokens": 227674404.0, + "step": 8950 + }, + { + "epoch": 0.9829782560948825, + "grad_norm": 1.9138962030410767, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6950664520263672, + "num_tokens": 227705719.0, + "step": 8951 + }, + { + "epoch": 0.9830880737974962, + "grad_norm": 2.2244715690612793, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7206791639328003, + "num_tokens": 227728417.0, + "step": 8952 + }, + { + "epoch": 0.9831978915001098, + "grad_norm": 2.457268238067627, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7094687819480896, + "num_tokens": 227747335.0, + "step": 8953 + }, + { + "epoch": 0.9833077092027235, + "grad_norm": 2.0310957431793213, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.716657280921936, + "num_tokens": 227774613.0, + "step": 8954 + }, + { + "epoch": 0.9834175269053371, + "grad_norm": 2.564762830734253, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7024784684181213, + "num_tokens": 227794608.0, + "step": 8955 + }, + { + "epoch": 0.9835273446079508, + "grad_norm": 2.156170129776001, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7163873314857483, + "num_tokens": 227818811.0, + "step": 8956 + }, + { + "epoch": 0.9836371623105644, + "grad_norm": 2.2590787410736084, + "learning_rate": 1e-06, + "loss": 1.0224, + "mean_token_accuracy": 0.693213939666748, + "num_tokens": 227844742.0, + "step": 8957 + }, + { + "epoch": 0.9837469800131782, + "grad_norm": 2.2008087635040283, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7145445346832275, + "num_tokens": 227869843.0, + "step": 8958 + }, + { + "epoch": 0.9838567977157918, + "grad_norm": 2.379905939102173, + "learning_rate": 1e-06, + "loss": 0.9937, + "mean_token_accuracy": 0.6990193128585815, + "num_tokens": 227894034.0, + "step": 8959 + }, + { + "epoch": 0.9839666154184055, + "grad_norm": 2.0918986797332764, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7368643283843994, + "num_tokens": 227918509.0, + "step": 8960 + }, + { + "epoch": 0.9840764331210191, + "grad_norm": 2.115610361099243, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7138172388076782, + "num_tokens": 227944722.0, + "step": 8961 + }, + { + "epoch": 0.9841862508236328, + "grad_norm": 2.1468284130096436, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7338037490844727, + "num_tokens": 227969627.0, + "step": 8962 + }, + { + "epoch": 0.9842960685262464, + "grad_norm": 2.647606611251831, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7145540714263916, + "num_tokens": 227990468.0, + "step": 8963 + }, + { + "epoch": 0.98440588622886, + "grad_norm": 1.88859224319458, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7068338990211487, + "num_tokens": 228023261.0, + "step": 8964 + }, + { + "epoch": 0.9845157039314738, + "grad_norm": 2.4163851737976074, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7055114507675171, + "num_tokens": 228044598.0, + "step": 8965 + }, + { + "epoch": 0.9846255216340875, + "grad_norm": 2.226184368133545, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.695780873298645, + "num_tokens": 228069032.0, + "step": 8966 + }, + { + "epoch": 0.9847353393367011, + "grad_norm": 1.8602381944656372, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.708487868309021, + "num_tokens": 228100823.0, + "step": 8967 + }, + { + "epoch": 0.9848451570393147, + "grad_norm": 1.9899080991744995, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7008137702941895, + "num_tokens": 228130564.0, + "step": 8968 + }, + { + "epoch": 0.9849549747419284, + "grad_norm": 1.9560203552246094, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7102440595626831, + "num_tokens": 228162874.0, + "step": 8969 + }, + { + "epoch": 0.985064792444542, + "grad_norm": 2.266939401626587, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7136805057525635, + "num_tokens": 228186646.0, + "step": 8970 + }, + { + "epoch": 0.9851746101471557, + "grad_norm": 2.1206214427948, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.729246973991394, + "num_tokens": 228213791.0, + "step": 8971 + }, + { + "epoch": 0.9852844278497693, + "grad_norm": 2.3146159648895264, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6947377324104309, + "num_tokens": 228237158.0, + "step": 8972 + }, + { + "epoch": 0.9853942455523831, + "grad_norm": 2.0378477573394775, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7168999910354614, + "num_tokens": 228266256.0, + "step": 8973 + }, + { + "epoch": 0.9855040632549967, + "grad_norm": 2.059596061706543, + "learning_rate": 1e-06, + "loss": 1.0512, + "mean_token_accuracy": 0.6804027557373047, + "num_tokens": 228294891.0, + "step": 8974 + }, + { + "epoch": 0.9856138809576104, + "grad_norm": 2.1881802082061768, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.6942735314369202, + "num_tokens": 228321819.0, + "step": 8975 + }, + { + "epoch": 0.985723698660224, + "grad_norm": 2.223163366317749, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7289040684700012, + "num_tokens": 228345560.0, + "step": 8976 + }, + { + "epoch": 0.9858335163628377, + "grad_norm": 2.455909252166748, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7145918607711792, + "num_tokens": 228364984.0, + "step": 8977 + }, + { + "epoch": 0.9859433340654513, + "grad_norm": 2.5572526454925537, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6947234869003296, + "num_tokens": 228384603.0, + "step": 8978 + }, + { + "epoch": 0.986053151768065, + "grad_norm": 2.04107403755188, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6921356916427612, + "num_tokens": 228415740.0, + "step": 8979 + }, + { + "epoch": 0.9861629694706787, + "grad_norm": 2.589975595474243, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7170842885971069, + "num_tokens": 228434128.0, + "step": 8980 + }, + { + "epoch": 0.9862727871732924, + "grad_norm": 2.2662246227264404, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7200808525085449, + "num_tokens": 228456088.0, + "step": 8981 + }, + { + "epoch": 0.986382604875906, + "grad_norm": 2.536099910736084, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7157994508743286, + "num_tokens": 228475958.0, + "step": 8982 + }, + { + "epoch": 0.9864924225785197, + "grad_norm": 1.982764720916748, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7227140665054321, + "num_tokens": 228506241.0, + "step": 8983 + }, + { + "epoch": 0.9866022402811333, + "grad_norm": 2.284571647644043, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7086074352264404, + "num_tokens": 228531089.0, + "step": 8984 + }, + { + "epoch": 0.986712057983747, + "grad_norm": 2.3966245651245117, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7152168154716492, + "num_tokens": 228553350.0, + "step": 8985 + }, + { + "epoch": 0.9868218756863606, + "grad_norm": 2.375701427459717, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7090528607368469, + "num_tokens": 228575857.0, + "step": 8986 + }, + { + "epoch": 0.9869316933889744, + "grad_norm": 2.4364430904388428, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7084842324256897, + "num_tokens": 228597693.0, + "step": 8987 + }, + { + "epoch": 0.987041511091588, + "grad_norm": 2.2222979068756104, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.727406919002533, + "num_tokens": 228622641.0, + "step": 8988 + }, + { + "epoch": 0.9871513287942016, + "grad_norm": 2.4109654426574707, + "learning_rate": 1e-06, + "loss": 1.0148, + "mean_token_accuracy": 0.6861585378646851, + "num_tokens": 228646205.0, + "step": 8989 + }, + { + "epoch": 0.9872611464968153, + "grad_norm": 2.7324013710021973, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7146170735359192, + "num_tokens": 228664662.0, + "step": 8990 + }, + { + "epoch": 0.9873709641994289, + "grad_norm": 2.6006360054016113, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7217555046081543, + "num_tokens": 228683918.0, + "step": 8991 + }, + { + "epoch": 0.9874807819020426, + "grad_norm": 2.334244966506958, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7303996086120605, + "num_tokens": 228705475.0, + "step": 8992 + }, + { + "epoch": 0.9875905996046562, + "grad_norm": 2.109811544418335, + "learning_rate": 1e-06, + "loss": 1.0527, + "mean_token_accuracy": 0.6813270449638367, + "num_tokens": 228734011.0, + "step": 8993 + }, + { + "epoch": 0.98770041730727, + "grad_norm": 1.9319392442703247, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7192521095275879, + "num_tokens": 228763518.0, + "step": 8994 + }, + { + "epoch": 0.9878102350098836, + "grad_norm": 2.3720333576202393, + "learning_rate": 1e-06, + "loss": 0.7826, + "mean_token_accuracy": 0.7484180331230164, + "num_tokens": 228783371.0, + "step": 8995 + }, + { + "epoch": 0.9879200527124973, + "grad_norm": 2.271348476409912, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7110288143157959, + "num_tokens": 228808613.0, + "step": 8996 + }, + { + "epoch": 0.9880298704151109, + "grad_norm": 2.2252390384674072, + "learning_rate": 1e-06, + "loss": 1.0637, + "mean_token_accuracy": 0.6811009049415588, + "num_tokens": 228837151.0, + "step": 8997 + }, + { + "epoch": 0.9881396881177246, + "grad_norm": 2.169740676879883, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7067671418190002, + "num_tokens": 228862438.0, + "step": 8998 + }, + { + "epoch": 0.9882495058203382, + "grad_norm": 2.5775928497314453, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7437517642974854, + "num_tokens": 228882082.0, + "step": 8999 + }, + { + "epoch": 0.9883593235229519, + "grad_norm": 2.171987295150757, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7012777328491211, + "num_tokens": 228909099.0, + "step": 9000 + }, + { + "epoch": 0.9884691412255655, + "grad_norm": 2.0759117603302, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7130215167999268, + "num_tokens": 228935036.0, + "step": 9001 + }, + { + "epoch": 0.9885789589281793, + "grad_norm": 2.305793046951294, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7030044198036194, + "num_tokens": 228959070.0, + "step": 9002 + }, + { + "epoch": 0.9886887766307929, + "grad_norm": 1.9381932020187378, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7021839618682861, + "num_tokens": 228990012.0, + "step": 9003 + }, + { + "epoch": 0.9887985943334066, + "grad_norm": 2.275261640548706, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7374107837677002, + "num_tokens": 229011727.0, + "step": 9004 + }, + { + "epoch": 0.9889084120360202, + "grad_norm": 2.0031161308288574, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7386074066162109, + "num_tokens": 229039594.0, + "step": 9005 + }, + { + "epoch": 0.9890182297386338, + "grad_norm": 2.346311330795288, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.6939976215362549, + "num_tokens": 229060562.0, + "step": 9006 + }, + { + "epoch": 0.9891280474412475, + "grad_norm": 2.360477924346924, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.6998967528343201, + "num_tokens": 229084337.0, + "step": 9007 + }, + { + "epoch": 0.9892378651438611, + "grad_norm": 2.4262988567352295, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6929994225502014, + "num_tokens": 229108620.0, + "step": 9008 + }, + { + "epoch": 0.9893476828464749, + "grad_norm": 2.164950370788574, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6992651224136353, + "num_tokens": 229133833.0, + "step": 9009 + }, + { + "epoch": 0.9894575005490885, + "grad_norm": 2.5629515647888184, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7319568395614624, + "num_tokens": 229151885.0, + "step": 9010 + }, + { + "epoch": 0.9895673182517022, + "grad_norm": 2.1681947708129883, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.7007296085357666, + "num_tokens": 229178246.0, + "step": 9011 + }, + { + "epoch": 0.9896771359543158, + "grad_norm": 2.240907907485962, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7150061726570129, + "num_tokens": 229203094.0, + "step": 9012 + }, + { + "epoch": 0.9897869536569295, + "grad_norm": 2.1969451904296875, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7245775461196899, + "num_tokens": 229229620.0, + "step": 9013 + }, + { + "epoch": 0.9898967713595431, + "grad_norm": 2.2346906661987305, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7044631838798523, + "num_tokens": 229253045.0, + "step": 9014 + }, + { + "epoch": 0.9900065890621568, + "grad_norm": 2.125720500946045, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7411046624183655, + "num_tokens": 229277875.0, + "step": 9015 + }, + { + "epoch": 0.9901164067647705, + "grad_norm": 2.23456072807312, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7042539119720459, + "num_tokens": 229302191.0, + "step": 9016 + }, + { + "epoch": 0.9902262244673842, + "grad_norm": 1.8959797620773315, + "learning_rate": 1e-06, + "loss": 1.0779, + "mean_token_accuracy": 0.6729730367660522, + "num_tokens": 229336418.0, + "step": 9017 + }, + { + "epoch": 0.9903360421699978, + "grad_norm": 2.123167037963867, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7210472822189331, + "num_tokens": 229363199.0, + "step": 9018 + }, + { + "epoch": 0.9904458598726115, + "grad_norm": 2.5378799438476562, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.71443110704422, + "num_tokens": 229382143.0, + "step": 9019 + }, + { + "epoch": 0.9905556775752251, + "grad_norm": 2.1912474632263184, + "learning_rate": 1e-06, + "loss": 1.0387, + "mean_token_accuracy": 0.6825328469276428, + "num_tokens": 229409235.0, + "step": 9020 + }, + { + "epoch": 0.9906654952778388, + "grad_norm": 2.238013982772827, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7123311758041382, + "num_tokens": 229434106.0, + "step": 9021 + }, + { + "epoch": 0.9907753129804524, + "grad_norm": 2.1769039630889893, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7248457670211792, + "num_tokens": 229459834.0, + "step": 9022 + }, + { + "epoch": 0.9908851306830662, + "grad_norm": 2.0531558990478516, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7183657288551331, + "num_tokens": 229486237.0, + "step": 9023 + }, + { + "epoch": 0.9909949483856798, + "grad_norm": 2.383728265762329, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.70705246925354, + "num_tokens": 229509230.0, + "step": 9024 + }, + { + "epoch": 0.9911047660882935, + "grad_norm": 1.9684406518936157, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7063666582107544, + "num_tokens": 229538412.0, + "step": 9025 + }, + { + "epoch": 0.9912145837909071, + "grad_norm": 2.353003740310669, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7125171422958374, + "num_tokens": 229559688.0, + "step": 9026 + }, + { + "epoch": 0.9913244014935207, + "grad_norm": 2.6023988723754883, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7189359664916992, + "num_tokens": 229578078.0, + "step": 9027 + }, + { + "epoch": 0.9914342191961344, + "grad_norm": 2.175489902496338, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.710146427154541, + "num_tokens": 229604607.0, + "step": 9028 + }, + { + "epoch": 0.991544036898748, + "grad_norm": 2.0541813373565674, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7049621343612671, + "num_tokens": 229632871.0, + "step": 9029 + }, + { + "epoch": 0.9916538546013618, + "grad_norm": 2.279611349105835, + "learning_rate": 1e-06, + "loss": 1.0345, + "mean_token_accuracy": 0.691483736038208, + "num_tokens": 229664580.0, + "step": 9030 + }, + { + "epoch": 0.9917636723039754, + "grad_norm": 2.565180778503418, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7200732231140137, + "num_tokens": 229684310.0, + "step": 9031 + }, + { + "epoch": 0.9918734900065891, + "grad_norm": 2.2588613033294678, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7325417995452881, + "num_tokens": 229705779.0, + "step": 9032 + }, + { + "epoch": 0.9919833077092027, + "grad_norm": 2.033456563949585, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7346980571746826, + "num_tokens": 229730686.0, + "step": 9033 + }, + { + "epoch": 0.9920931254118164, + "grad_norm": 2.0728561878204346, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7301881909370422, + "num_tokens": 229759963.0, + "step": 9034 + }, + { + "epoch": 0.99220294311443, + "grad_norm": 2.2667722702026367, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.713247537612915, + "num_tokens": 229784011.0, + "step": 9035 + }, + { + "epoch": 0.9923127608170437, + "grad_norm": 2.307469129562378, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.7099428772926331, + "num_tokens": 229806638.0, + "step": 9036 + }, + { + "epoch": 0.9924225785196573, + "grad_norm": 2.2835285663604736, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7240631580352783, + "num_tokens": 229830362.0, + "step": 9037 + }, + { + "epoch": 0.9925323962222711, + "grad_norm": 2.208451986312866, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7273903489112854, + "num_tokens": 229854333.0, + "step": 9038 + }, + { + "epoch": 0.9926422139248847, + "grad_norm": 2.240359306335449, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7249451279640198, + "num_tokens": 229877280.0, + "step": 9039 + }, + { + "epoch": 0.9927520316274984, + "grad_norm": 2.0349373817443848, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7162126302719116, + "num_tokens": 229904814.0, + "step": 9040 + }, + { + "epoch": 0.992861849330112, + "grad_norm": 2.1243860721588135, + "learning_rate": 1e-06, + "loss": 1.0231, + "mean_token_accuracy": 0.7060078382492065, + "num_tokens": 229932525.0, + "step": 9041 + }, + { + "epoch": 0.9929716670327257, + "grad_norm": 2.316425323486328, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7012083530426025, + "num_tokens": 229955503.0, + "step": 9042 + }, + { + "epoch": 0.9930814847353393, + "grad_norm": 1.7815078496932983, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.692150890827179, + "num_tokens": 229991950.0, + "step": 9043 + }, + { + "epoch": 0.993191302437953, + "grad_norm": 2.2698745727539062, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7373848557472229, + "num_tokens": 230015081.0, + "step": 9044 + }, + { + "epoch": 0.9933011201405667, + "grad_norm": 2.0358004570007324, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.6868491768836975, + "num_tokens": 230043031.0, + "step": 9045 + }, + { + "epoch": 0.9934109378431804, + "grad_norm": 2.007512331008911, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7269237637519836, + "num_tokens": 230069665.0, + "step": 9046 + }, + { + "epoch": 0.993520755545794, + "grad_norm": 2.6411941051483154, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.7375996708869934, + "num_tokens": 230089107.0, + "step": 9047 + }, + { + "epoch": 0.9936305732484076, + "grad_norm": 2.2288784980773926, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7149898409843445, + "num_tokens": 230114944.0, + "step": 9048 + }, + { + "epoch": 0.9937403909510213, + "grad_norm": 2.2542731761932373, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7231506109237671, + "num_tokens": 230139369.0, + "step": 9049 + }, + { + "epoch": 0.9938502086536349, + "grad_norm": 2.0430142879486084, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7296714186668396, + "num_tokens": 230168219.0, + "step": 9050 + }, + { + "epoch": 0.9939600263562486, + "grad_norm": 2.1176598072052, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7046658992767334, + "num_tokens": 230192617.0, + "step": 9051 + }, + { + "epoch": 0.9940698440588623, + "grad_norm": 2.2362422943115234, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7274242639541626, + "num_tokens": 230216458.0, + "step": 9052 + }, + { + "epoch": 0.994179661761476, + "grad_norm": 2.191161632537842, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7057567238807678, + "num_tokens": 230245257.0, + "step": 9053 + }, + { + "epoch": 0.9942894794640896, + "grad_norm": 2.3133130073547363, + "learning_rate": 1e-06, + "loss": 0.9922, + "mean_token_accuracy": 0.7044801115989685, + "num_tokens": 230267706.0, + "step": 9054 + }, + { + "epoch": 0.9943992971667033, + "grad_norm": 2.3605940341949463, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.712239146232605, + "num_tokens": 230290076.0, + "step": 9055 + }, + { + "epoch": 0.9945091148693169, + "grad_norm": 2.284337282180786, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.706872820854187, + "num_tokens": 230313921.0, + "step": 9056 + }, + { + "epoch": 0.9946189325719306, + "grad_norm": 2.4716005325317383, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.6994336843490601, + "num_tokens": 230334135.0, + "step": 9057 + }, + { + "epoch": 0.9947287502745442, + "grad_norm": 2.2858619689941406, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7015156745910645, + "num_tokens": 230360801.0, + "step": 9058 + }, + { + "epoch": 0.994838567977158, + "grad_norm": 2.28132963180542, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7086358070373535, + "num_tokens": 230384079.0, + "step": 9059 + }, + { + "epoch": 0.9949483856797716, + "grad_norm": 2.2518882751464844, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7087566256523132, + "num_tokens": 230409054.0, + "step": 9060 + }, + { + "epoch": 0.9950582033823853, + "grad_norm": 2.0586304664611816, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7346472144126892, + "num_tokens": 230433880.0, + "step": 9061 + }, + { + "epoch": 0.9951680210849989, + "grad_norm": 1.8610063791275024, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7213279008865356, + "num_tokens": 230465918.0, + "step": 9062 + }, + { + "epoch": 0.9952778387876126, + "grad_norm": 2.5569446086883545, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.7023087739944458, + "num_tokens": 230485386.0, + "step": 9063 + }, + { + "epoch": 0.9953876564902262, + "grad_norm": 2.0992605686187744, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7114778757095337, + "num_tokens": 230512040.0, + "step": 9064 + }, + { + "epoch": 0.9954974741928398, + "grad_norm": 2.257781505584717, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7144078612327576, + "num_tokens": 230534575.0, + "step": 9065 + }, + { + "epoch": 0.9956072918954535, + "grad_norm": 2.0663015842437744, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6947077512741089, + "num_tokens": 230561926.0, + "step": 9066 + }, + { + "epoch": 0.9957171095980673, + "grad_norm": 2.4512057304382324, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7322922348976135, + "num_tokens": 230581011.0, + "step": 9067 + }, + { + "epoch": 0.9958269273006809, + "grad_norm": 1.93175208568573, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6934741735458374, + "num_tokens": 230612544.0, + "step": 9068 + }, + { + "epoch": 0.9959367450032945, + "grad_norm": 2.17195725440979, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7265377044677734, + "num_tokens": 230636331.0, + "step": 9069 + }, + { + "epoch": 0.9960465627059082, + "grad_norm": 1.9846256971359253, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7157660126686096, + "num_tokens": 230664659.0, + "step": 9070 + }, + { + "epoch": 0.9961563804085218, + "grad_norm": 2.2422595024108887, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.706224799156189, + "num_tokens": 230689101.0, + "step": 9071 + }, + { + "epoch": 0.9962661981111355, + "grad_norm": 2.278707981109619, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7162278294563293, + "num_tokens": 230711806.0, + "step": 9072 + }, + { + "epoch": 0.9963760158137491, + "grad_norm": 2.3725550174713135, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7199062705039978, + "num_tokens": 230733572.0, + "step": 9073 + }, + { + "epoch": 0.9964858335163629, + "grad_norm": 2.3241970539093018, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.701995849609375, + "num_tokens": 230757843.0, + "step": 9074 + }, + { + "epoch": 0.9965956512189765, + "grad_norm": 2.3301444053649902, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7118211388587952, + "num_tokens": 230782793.0, + "step": 9075 + }, + { + "epoch": 0.9967054689215902, + "grad_norm": 2.6891183853149414, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7114008665084839, + "num_tokens": 230801896.0, + "step": 9076 + }, + { + "epoch": 0.9968152866242038, + "grad_norm": 2.3158435821533203, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7222973108291626, + "num_tokens": 230823942.0, + "step": 9077 + }, + { + "epoch": 0.9969251043268175, + "grad_norm": 2.143066883087158, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7105119228363037, + "num_tokens": 230848224.0, + "step": 9078 + }, + { + "epoch": 0.9970349220294311, + "grad_norm": 2.133824586868286, + "learning_rate": 1e-06, + "loss": 0.9914, + "mean_token_accuracy": 0.7003618478775024, + "num_tokens": 230875947.0, + "step": 9079 + }, + { + "epoch": 0.9971447397320448, + "grad_norm": 2.3253660202026367, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7016254663467407, + "num_tokens": 230899613.0, + "step": 9080 + }, + { + "epoch": 0.9972545574346585, + "grad_norm": 2.23571515083313, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7191609740257263, + "num_tokens": 230926389.0, + "step": 9081 + }, + { + "epoch": 0.9973643751372722, + "grad_norm": 2.100173234939575, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.723669171333313, + "num_tokens": 230952705.0, + "step": 9082 + }, + { + "epoch": 0.9974741928398858, + "grad_norm": 2.069831132888794, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7114672064781189, + "num_tokens": 230979020.0, + "step": 9083 + }, + { + "epoch": 0.9975840105424995, + "grad_norm": 2.18953013420105, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7274930477142334, + "num_tokens": 231002606.0, + "step": 9084 + }, + { + "epoch": 0.9976938282451131, + "grad_norm": 2.087934732437134, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.701547384262085, + "num_tokens": 231030007.0, + "step": 9085 + }, + { + "epoch": 0.9978036459477267, + "grad_norm": 2.162559747695923, + "learning_rate": 1e-06, + "loss": 0.9828, + "mean_token_accuracy": 0.696245551109314, + "num_tokens": 231054719.0, + "step": 9086 + }, + { + "epoch": 0.9979134636503404, + "grad_norm": 2.403998374938965, + "learning_rate": 1e-06, + "loss": 0.7807, + "mean_token_accuracy": 0.7507118582725525, + "num_tokens": 231074043.0, + "step": 9087 + }, + { + "epoch": 0.9980232813529542, + "grad_norm": 2.5331125259399414, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.731133759021759, + "num_tokens": 231093161.0, + "step": 9088 + }, + { + "epoch": 0.9981330990555678, + "grad_norm": 2.192042112350464, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7046657800674438, + "num_tokens": 231116992.0, + "step": 9089 + }, + { + "epoch": 0.9982429167581814, + "grad_norm": 2.271958351135254, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7235733270645142, + "num_tokens": 231138817.0, + "step": 9090 + }, + { + "epoch": 0.9983527344607951, + "grad_norm": 2.1474292278289795, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7059675455093384, + "num_tokens": 231165257.0, + "step": 9091 + }, + { + "epoch": 0.9984625521634087, + "grad_norm": 2.456617832183838, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7087166905403137, + "num_tokens": 231185828.0, + "step": 9092 + }, + { + "epoch": 0.9985723698660224, + "grad_norm": 2.435281276702881, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.733871340751648, + "num_tokens": 231207762.0, + "step": 9093 + }, + { + "epoch": 0.998682187568636, + "grad_norm": 2.1894354820251465, + "learning_rate": 1e-06, + "loss": 1.0436, + "mean_token_accuracy": 0.6846349239349365, + "num_tokens": 231234317.0, + "step": 9094 + }, + { + "epoch": 0.9987920052712497, + "grad_norm": 2.3885884284973145, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7055116891860962, + "num_tokens": 231256586.0, + "step": 9095 + }, + { + "epoch": 0.9989018229738634, + "grad_norm": 2.193105459213257, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7075966596603394, + "num_tokens": 231278974.0, + "step": 9096 + }, + { + "epoch": 0.9990116406764771, + "grad_norm": 1.7785755395889282, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.6985788345336914, + "num_tokens": 231314192.0, + "step": 9097 + }, + { + "epoch": 0.9991214583790907, + "grad_norm": 2.1912479400634766, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6951613426208496, + "num_tokens": 231340511.0, + "step": 9098 + }, + { + "epoch": 0.9992312760817044, + "grad_norm": 2.1435940265655518, + "learning_rate": 1e-06, + "loss": 1.0433, + "mean_token_accuracy": 0.6892215013504028, + "num_tokens": 231366951.0, + "step": 9099 + }, + { + "epoch": 0.999341093784318, + "grad_norm": 2.18058705329895, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7200348377227783, + "num_tokens": 231393970.0, + "step": 9100 + }, + { + "epoch": 0.9994509114869317, + "grad_norm": 2.0506749153137207, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.6833289265632629, + "num_tokens": 231423544.0, + "step": 9101 + }, + { + "epoch": 0.9995607291895453, + "grad_norm": 2.3761038780212402, + "learning_rate": 1e-06, + "loss": 0.956, + "mean_token_accuracy": 0.7021403908729553, + "num_tokens": 231445269.0, + "step": 9102 + }, + { + "epoch": 0.9996705468921591, + "grad_norm": 2.127634048461914, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.711166262626648, + "num_tokens": 231472003.0, + "step": 9103 + }, + { + "epoch": 0.9997803645947727, + "grad_norm": 1.9161664247512817, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7072316408157349, + "num_tokens": 231500290.0, + "step": 9104 + }, + { + "epoch": 0.9998901822973864, + "grad_norm": 2.2307705879211426, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.728050947189331, + "num_tokens": 231522886.0, + "step": 9105 + }, + { + "epoch": 1.0, + "grad_norm": 2.122331380844116, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7078819870948792, + "num_tokens": 231549164.0, + "step": 9106 + }, + { + "epoch": 1.0001098177026138, + "grad_norm": 2.189965009689331, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7058147192001343, + "num_tokens": 231572957.0, + "step": 9107 + }, + { + "epoch": 1.0002196354052273, + "grad_norm": 2.270974636077881, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7023723125457764, + "num_tokens": 231595814.0, + "step": 9108 + }, + { + "epoch": 1.000329453107841, + "grad_norm": 2.153322696685791, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7277923226356506, + "num_tokens": 231621776.0, + "step": 9109 + }, + { + "epoch": 1.0004392708104546, + "grad_norm": 2.206050157546997, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7229541540145874, + "num_tokens": 231645369.0, + "step": 9110 + }, + { + "epoch": 1.0005490885130683, + "grad_norm": 2.4591503143310547, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7185583114624023, + "num_tokens": 231665020.0, + "step": 9111 + }, + { + "epoch": 1.0006589062156819, + "grad_norm": 2.124528408050537, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7133387327194214, + "num_tokens": 231690669.0, + "step": 9112 + }, + { + "epoch": 1.0007687239182956, + "grad_norm": 2.16139817237854, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7216045260429382, + "num_tokens": 231716829.0, + "step": 9113 + }, + { + "epoch": 1.0008785416209094, + "grad_norm": 1.9932918548583984, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7145596742630005, + "num_tokens": 231745272.0, + "step": 9114 + }, + { + "epoch": 1.000988359323523, + "grad_norm": 2.1049885749816895, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7258930802345276, + "num_tokens": 231771564.0, + "step": 9115 + }, + { + "epoch": 1.0010981770261367, + "grad_norm": 1.7162595987319946, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7089670300483704, + "num_tokens": 231812385.0, + "step": 9116 + }, + { + "epoch": 1.0012079947287502, + "grad_norm": 2.167619228363037, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.6983758807182312, + "num_tokens": 231839483.0, + "step": 9117 + }, + { + "epoch": 1.001317812431364, + "grad_norm": 2.1357991695404053, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.732420802116394, + "num_tokens": 231864948.0, + "step": 9118 + }, + { + "epoch": 1.0014276301339775, + "grad_norm": 2.052466630935669, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7217971682548523, + "num_tokens": 231895096.0, + "step": 9119 + }, + { + "epoch": 1.0015374478365913, + "grad_norm": 2.2601370811462402, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7302312850952148, + "num_tokens": 231918934.0, + "step": 9120 + }, + { + "epoch": 1.001647265539205, + "grad_norm": 2.033663511276245, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7351440787315369, + "num_tokens": 231947743.0, + "step": 9121 + }, + { + "epoch": 1.0017570832418186, + "grad_norm": 2.127211570739746, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7233990430831909, + "num_tokens": 231973916.0, + "step": 9122 + }, + { + "epoch": 1.0018669009444323, + "grad_norm": 2.3890419006347656, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.702174723148346, + "num_tokens": 231997583.0, + "step": 9123 + }, + { + "epoch": 1.0019767186470458, + "grad_norm": 2.103654146194458, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7182689905166626, + "num_tokens": 232024225.0, + "step": 9124 + }, + { + "epoch": 1.0020865363496596, + "grad_norm": 1.9434386491775513, + "learning_rate": 1e-06, + "loss": 1.0834, + "mean_token_accuracy": 0.6705371141433716, + "num_tokens": 232055731.0, + "step": 9125 + }, + { + "epoch": 1.0021963540522731, + "grad_norm": 2.4945619106292725, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7300201058387756, + "num_tokens": 232076475.0, + "step": 9126 + }, + { + "epoch": 1.002306171754887, + "grad_norm": 2.255650043487549, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7187156677246094, + "num_tokens": 232100262.0, + "step": 9127 + }, + { + "epoch": 1.0024159894575007, + "grad_norm": 2.3160362243652344, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.6918387413024902, + "num_tokens": 232124105.0, + "step": 9128 + }, + { + "epoch": 1.0025258071601142, + "grad_norm": 2.471465826034546, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.730187714099884, + "num_tokens": 232145486.0, + "step": 9129 + }, + { + "epoch": 1.002635624862728, + "grad_norm": 2.2488162517547607, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7402147054672241, + "num_tokens": 232168083.0, + "step": 9130 + }, + { + "epoch": 1.0027454425653415, + "grad_norm": 2.102454662322998, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7123675346374512, + "num_tokens": 232194734.0, + "step": 9131 + }, + { + "epoch": 1.0028552602679552, + "grad_norm": 2.081815481185913, + "learning_rate": 1e-06, + "loss": 1.05, + "mean_token_accuracy": 0.677247166633606, + "num_tokens": 232225196.0, + "step": 9132 + }, + { + "epoch": 1.0029650779705688, + "grad_norm": 2.187969923019409, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7306212782859802, + "num_tokens": 232250155.0, + "step": 9133 + }, + { + "epoch": 1.0030748956731825, + "grad_norm": 2.1098711490631104, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.729324221611023, + "num_tokens": 232275006.0, + "step": 9134 + }, + { + "epoch": 1.0031847133757963, + "grad_norm": 2.256868600845337, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.716581404209137, + "num_tokens": 232301306.0, + "step": 9135 + }, + { + "epoch": 1.0032945310784098, + "grad_norm": 2.0617380142211914, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.6989589333534241, + "num_tokens": 232329675.0, + "step": 9136 + }, + { + "epoch": 1.0034043487810236, + "grad_norm": 2.2560319900512695, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7380098104476929, + "num_tokens": 232352862.0, + "step": 9137 + }, + { + "epoch": 1.0035141664836371, + "grad_norm": 2.281360387802124, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7242727875709534, + "num_tokens": 232375994.0, + "step": 9138 + }, + { + "epoch": 1.0036239841862509, + "grad_norm": 2.1780991554260254, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7133394479751587, + "num_tokens": 232402180.0, + "step": 9139 + }, + { + "epoch": 1.0037338018888644, + "grad_norm": 2.45015549659729, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7197791337966919, + "num_tokens": 232424778.0, + "step": 9140 + }, + { + "epoch": 1.0038436195914782, + "grad_norm": 2.213486909866333, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7077406644821167, + "num_tokens": 232450001.0, + "step": 9141 + }, + { + "epoch": 1.0039534372940917, + "grad_norm": 2.0245442390441895, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6901437044143677, + "num_tokens": 232482488.0, + "step": 9142 + }, + { + "epoch": 1.0040632549967055, + "grad_norm": 2.007997751235962, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7104970216751099, + "num_tokens": 232512471.0, + "step": 9143 + }, + { + "epoch": 1.0041730726993192, + "grad_norm": 2.796666145324707, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.7307254672050476, + "num_tokens": 232530370.0, + "step": 9144 + }, + { + "epoch": 1.0042828904019327, + "grad_norm": 1.9524571895599365, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7154901623725891, + "num_tokens": 232561976.0, + "step": 9145 + }, + { + "epoch": 1.0043927081045465, + "grad_norm": 2.2173383235931396, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7129805088043213, + "num_tokens": 232588677.0, + "step": 9146 + }, + { + "epoch": 1.00450252580716, + "grad_norm": 1.8453824520111084, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6886623501777649, + "num_tokens": 232623712.0, + "step": 9147 + }, + { + "epoch": 1.0046123435097738, + "grad_norm": 2.25754451751709, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7274702191352844, + "num_tokens": 232649228.0, + "step": 9148 + }, + { + "epoch": 1.0047221612123873, + "grad_norm": 2.1835808753967285, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7172627449035645, + "num_tokens": 232673896.0, + "step": 9149 + }, + { + "epoch": 1.004831978915001, + "grad_norm": 2.8384134769439697, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.73017817735672, + "num_tokens": 232690631.0, + "step": 9150 + }, + { + "epoch": 1.0049417966176148, + "grad_norm": 2.3037376403808594, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.7028911709785461, + "num_tokens": 232716355.0, + "step": 9151 + }, + { + "epoch": 1.0050516143202284, + "grad_norm": 2.033426523208618, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.7000505328178406, + "num_tokens": 232747130.0, + "step": 9152 + }, + { + "epoch": 1.0051614320228421, + "grad_norm": 2.3999125957489014, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7293736934661865, + "num_tokens": 232768486.0, + "step": 9153 + }, + { + "epoch": 1.0052712497254557, + "grad_norm": 2.042834997177124, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7096294164657593, + "num_tokens": 232795548.0, + "step": 9154 + }, + { + "epoch": 1.0053810674280694, + "grad_norm": 2.249441385269165, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7220823168754578, + "num_tokens": 232822334.0, + "step": 9155 + }, + { + "epoch": 1.005490885130683, + "grad_norm": 2.0440309047698975, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7263397574424744, + "num_tokens": 232852012.0, + "step": 9156 + }, + { + "epoch": 1.0056007028332967, + "grad_norm": 2.221055030822754, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7292526364326477, + "num_tokens": 232877612.0, + "step": 9157 + }, + { + "epoch": 1.0057105205359105, + "grad_norm": 2.4920432567596436, + "learning_rate": 1e-06, + "loss": 0.9674, + "mean_token_accuracy": 0.7017711997032166, + "num_tokens": 232898654.0, + "step": 9158 + }, + { + "epoch": 1.005820338238524, + "grad_norm": 2.037625312805176, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.7002333402633667, + "num_tokens": 232929738.0, + "step": 9159 + }, + { + "epoch": 1.0059301559411378, + "grad_norm": 2.3626554012298584, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7277389764785767, + "num_tokens": 232951170.0, + "step": 9160 + }, + { + "epoch": 1.0060399736437513, + "grad_norm": 2.1314563751220703, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7199820280075073, + "num_tokens": 232979858.0, + "step": 9161 + }, + { + "epoch": 1.006149791346365, + "grad_norm": 2.6624062061309814, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.714017391204834, + "num_tokens": 232998025.0, + "step": 9162 + }, + { + "epoch": 1.0062596090489786, + "grad_norm": 2.6836066246032715, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7163247466087341, + "num_tokens": 233017501.0, + "step": 9163 + }, + { + "epoch": 1.0063694267515924, + "grad_norm": 2.3261935710906982, + "learning_rate": 1e-06, + "loss": 0.9208, + "mean_token_accuracy": 0.7094048261642456, + "num_tokens": 233040035.0, + "step": 9164 + }, + { + "epoch": 1.0064792444542061, + "grad_norm": 2.3610806465148926, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7093596458435059, + "num_tokens": 233064240.0, + "step": 9165 + }, + { + "epoch": 1.0065890621568196, + "grad_norm": 2.2371537685394287, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7178133130073547, + "num_tokens": 233090191.0, + "step": 9166 + }, + { + "epoch": 1.0066988798594334, + "grad_norm": 2.4929659366607666, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.729444146156311, + "num_tokens": 233110599.0, + "step": 9167 + }, + { + "epoch": 1.006808697562047, + "grad_norm": 2.3343214988708496, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7314053773880005, + "num_tokens": 233132998.0, + "step": 9168 + }, + { + "epoch": 1.0069185152646607, + "grad_norm": 2.0541703701019287, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7177683115005493, + "num_tokens": 233161001.0, + "step": 9169 + }, + { + "epoch": 1.0070283329672742, + "grad_norm": 2.382305383682251, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7336268424987793, + "num_tokens": 233181613.0, + "step": 9170 + }, + { + "epoch": 1.007138150669888, + "grad_norm": 2.2619376182556152, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7243620157241821, + "num_tokens": 233207232.0, + "step": 9171 + }, + { + "epoch": 1.0072479683725017, + "grad_norm": 2.3003227710723877, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7135013341903687, + "num_tokens": 233230281.0, + "step": 9172 + }, + { + "epoch": 1.0073577860751153, + "grad_norm": 1.9443511962890625, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7067064046859741, + "num_tokens": 233263850.0, + "step": 9173 + }, + { + "epoch": 1.007467603777729, + "grad_norm": 2.041621446609497, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7126645445823669, + "num_tokens": 233293487.0, + "step": 9174 + }, + { + "epoch": 1.0075774214803426, + "grad_norm": 2.3385164737701416, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7153943181037903, + "num_tokens": 233316340.0, + "step": 9175 + }, + { + "epoch": 1.0076872391829563, + "grad_norm": 2.2123334407806396, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7215032577514648, + "num_tokens": 233339450.0, + "step": 9176 + }, + { + "epoch": 1.0077970568855699, + "grad_norm": 2.542280912399292, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7416586875915527, + "num_tokens": 233360967.0, + "step": 9177 + }, + { + "epoch": 1.0079068745881836, + "grad_norm": 2.2613062858581543, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7234405875205994, + "num_tokens": 233384001.0, + "step": 9178 + }, + { + "epoch": 1.0080166922907974, + "grad_norm": 1.9063262939453125, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.7047296762466431, + "num_tokens": 233415302.0, + "step": 9179 + }, + { + "epoch": 1.008126509993411, + "grad_norm": 2.2160396575927734, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7358136177062988, + "num_tokens": 233440705.0, + "step": 9180 + }, + { + "epoch": 1.0082363276960247, + "grad_norm": 2.1365647315979004, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7378783822059631, + "num_tokens": 233466013.0, + "step": 9181 + }, + { + "epoch": 1.0083461453986382, + "grad_norm": 2.117999315261841, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7360761165618896, + "num_tokens": 233490801.0, + "step": 9182 + }, + { + "epoch": 1.008455963101252, + "grad_norm": 2.3579888343811035, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7324466109275818, + "num_tokens": 233513422.0, + "step": 9183 + }, + { + "epoch": 1.0085657808038655, + "grad_norm": 2.492715358734131, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.710486650466919, + "num_tokens": 233534451.0, + "step": 9184 + }, + { + "epoch": 1.0086755985064793, + "grad_norm": 2.2741003036499023, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7059540748596191, + "num_tokens": 233560779.0, + "step": 9185 + }, + { + "epoch": 1.008785416209093, + "grad_norm": 2.115985870361328, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7107283473014832, + "num_tokens": 233589264.0, + "step": 9186 + }, + { + "epoch": 1.0088952339117065, + "grad_norm": 2.25693678855896, + "learning_rate": 1e-06, + "loss": 1.0064, + "mean_token_accuracy": 0.6982553005218506, + "num_tokens": 233616792.0, + "step": 9187 + }, + { + "epoch": 1.0090050516143203, + "grad_norm": 2.2568886280059814, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.692589282989502, + "num_tokens": 233643640.0, + "step": 9188 + }, + { + "epoch": 1.0091148693169338, + "grad_norm": 2.177016496658325, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.721657395362854, + "num_tokens": 233668566.0, + "step": 9189 + }, + { + "epoch": 1.0092246870195476, + "grad_norm": 2.784186363220215, + "learning_rate": 1e-06, + "loss": 0.7526, + "mean_token_accuracy": 0.7542338371276855, + "num_tokens": 233685001.0, + "step": 9190 + }, + { + "epoch": 1.0093345047221611, + "grad_norm": 2.248701572418213, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7219561338424683, + "num_tokens": 233711216.0, + "step": 9191 + }, + { + "epoch": 1.0094443224247749, + "grad_norm": 2.2596590518951416, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.711885929107666, + "num_tokens": 233735779.0, + "step": 9192 + }, + { + "epoch": 1.0095541401273886, + "grad_norm": 2.553244113922119, + "learning_rate": 1e-06, + "loss": 0.7935, + "mean_token_accuracy": 0.7416959404945374, + "num_tokens": 233753646.0, + "step": 9193 + }, + { + "epoch": 1.0096639578300022, + "grad_norm": 2.401292085647583, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7155125141143799, + "num_tokens": 233775108.0, + "step": 9194 + }, + { + "epoch": 1.009773775532616, + "grad_norm": 1.9264169931411743, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7140711545944214, + "num_tokens": 233807653.0, + "step": 9195 + }, + { + "epoch": 1.0098835932352295, + "grad_norm": 2.484632968902588, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7295962572097778, + "num_tokens": 233829158.0, + "step": 9196 + }, + { + "epoch": 1.0099934109378432, + "grad_norm": 2.342716932296753, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7113566398620605, + "num_tokens": 233851768.0, + "step": 9197 + }, + { + "epoch": 1.0101032286404568, + "grad_norm": 2.2305381298065186, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7193847894668579, + "num_tokens": 233878894.0, + "step": 9198 + }, + { + "epoch": 1.0102130463430705, + "grad_norm": 2.0299954414367676, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.6990699768066406, + "num_tokens": 233908129.0, + "step": 9199 + }, + { + "epoch": 1.010322864045684, + "grad_norm": 2.180955648422241, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7168111801147461, + "num_tokens": 233934695.0, + "step": 9200 + }, + { + "epoch": 1.0104326817482978, + "grad_norm": 2.094820976257324, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6904848217964172, + "num_tokens": 233962080.0, + "step": 9201 + }, + { + "epoch": 1.0105424994509116, + "grad_norm": 2.168325185775757, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7215328812599182, + "num_tokens": 233988189.0, + "step": 9202 + }, + { + "epoch": 1.010652317153525, + "grad_norm": 1.949633002281189, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7167075276374817, + "num_tokens": 234021064.0, + "step": 9203 + }, + { + "epoch": 1.0107621348561389, + "grad_norm": 2.1986606121063232, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.7117481231689453, + "num_tokens": 234047764.0, + "step": 9204 + }, + { + "epoch": 1.0108719525587524, + "grad_norm": 2.072087049484253, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7413495779037476, + "num_tokens": 234074750.0, + "step": 9205 + }, + { + "epoch": 1.0109817702613662, + "grad_norm": 2.198012590408325, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7211732268333435, + "num_tokens": 234101748.0, + "step": 9206 + }, + { + "epoch": 1.0110915879639797, + "grad_norm": 2.2199866771698, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7126388549804688, + "num_tokens": 234128977.0, + "step": 9207 + }, + { + "epoch": 1.0112014056665934, + "grad_norm": 2.320268154144287, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.696947455406189, + "num_tokens": 234152901.0, + "step": 9208 + }, + { + "epoch": 1.0113112233692072, + "grad_norm": 1.9450039863586426, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7209118604660034, + "num_tokens": 234186561.0, + "step": 9209 + }, + { + "epoch": 1.0114210410718207, + "grad_norm": 2.0374512672424316, + "learning_rate": 1e-06, + "loss": 1.0155, + "mean_token_accuracy": 0.6900631785392761, + "num_tokens": 234222797.0, + "step": 9210 + }, + { + "epoch": 1.0115308587744345, + "grad_norm": 2.3277831077575684, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7349659204483032, + "num_tokens": 234247205.0, + "step": 9211 + }, + { + "epoch": 1.011640676477048, + "grad_norm": 2.2381491661071777, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7288002967834473, + "num_tokens": 234270740.0, + "step": 9212 + }, + { + "epoch": 1.0117504941796618, + "grad_norm": 2.411390542984009, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7351319789886475, + "num_tokens": 234292989.0, + "step": 9213 + }, + { + "epoch": 1.0118603118822753, + "grad_norm": 1.966897964477539, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7189444899559021, + "num_tokens": 234322450.0, + "step": 9214 + }, + { + "epoch": 1.011970129584889, + "grad_norm": 2.2449915409088135, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7154189348220825, + "num_tokens": 234347081.0, + "step": 9215 + }, + { + "epoch": 1.0120799472875028, + "grad_norm": 2.2745025157928467, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7205591201782227, + "num_tokens": 234372403.0, + "step": 9216 + }, + { + "epoch": 1.0121897649901164, + "grad_norm": 2.3348166942596436, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7163442373275757, + "num_tokens": 234395849.0, + "step": 9217 + }, + { + "epoch": 1.0122995826927301, + "grad_norm": 2.5566413402557373, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7267705798149109, + "num_tokens": 234415003.0, + "step": 9218 + }, + { + "epoch": 1.0124094003953437, + "grad_norm": 2.093024253845215, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7181161046028137, + "num_tokens": 234445250.0, + "step": 9219 + }, + { + "epoch": 1.0125192180979574, + "grad_norm": 2.296159029006958, + "learning_rate": 1e-06, + "loss": 0.7273, + "mean_token_accuracy": 0.7698969841003418, + "num_tokens": 234467635.0, + "step": 9220 + }, + { + "epoch": 1.012629035800571, + "grad_norm": 2.756739377975464, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7147518396377563, + "num_tokens": 234486912.0, + "step": 9221 + }, + { + "epoch": 1.0127388535031847, + "grad_norm": 2.6791279315948486, + "learning_rate": 1e-06, + "loss": 0.8043, + "mean_token_accuracy": 0.7446106672286987, + "num_tokens": 234505532.0, + "step": 9222 + }, + { + "epoch": 1.0128486712057985, + "grad_norm": 2.5382492542266846, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.7047125697135925, + "num_tokens": 234528056.0, + "step": 9223 + }, + { + "epoch": 1.012958488908412, + "grad_norm": 2.0565247535705566, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7018088102340698, + "num_tokens": 234560171.0, + "step": 9224 + }, + { + "epoch": 1.0130683066110258, + "grad_norm": 2.463033437728882, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7371770143508911, + "num_tokens": 234582245.0, + "step": 9225 + }, + { + "epoch": 1.0131781243136393, + "grad_norm": 2.1348226070404053, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7202684879302979, + "num_tokens": 234611076.0, + "step": 9226 + }, + { + "epoch": 1.013287942016253, + "grad_norm": 2.4555253982543945, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7307546138763428, + "num_tokens": 234631875.0, + "step": 9227 + }, + { + "epoch": 1.0133977597188666, + "grad_norm": 2.537385940551758, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7234404683113098, + "num_tokens": 234652784.0, + "step": 9228 + }, + { + "epoch": 1.0135075774214803, + "grad_norm": 2.316331148147583, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.743346631526947, + "num_tokens": 234675882.0, + "step": 9229 + }, + { + "epoch": 1.013617395124094, + "grad_norm": 2.093993902206421, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7215845584869385, + "num_tokens": 234703743.0, + "step": 9230 + }, + { + "epoch": 1.0137272128267076, + "grad_norm": 2.2829079627990723, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7251514196395874, + "num_tokens": 234727327.0, + "step": 9231 + }, + { + "epoch": 1.0138370305293214, + "grad_norm": 2.594811201095581, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7297222018241882, + "num_tokens": 234747543.0, + "step": 9232 + }, + { + "epoch": 1.013946848231935, + "grad_norm": 2.083493232727051, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7179954051971436, + "num_tokens": 234776638.0, + "step": 9233 + }, + { + "epoch": 1.0140566659345487, + "grad_norm": 2.1601359844207764, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7037311792373657, + "num_tokens": 234805203.0, + "step": 9234 + }, + { + "epoch": 1.0141664836371622, + "grad_norm": 2.3332345485687256, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7221812009811401, + "num_tokens": 234829100.0, + "step": 9235 + }, + { + "epoch": 1.014276301339776, + "grad_norm": 2.414506673812866, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7038625478744507, + "num_tokens": 234850194.0, + "step": 9236 + }, + { + "epoch": 1.0143861190423897, + "grad_norm": 2.326991081237793, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.6999147534370422, + "num_tokens": 234877477.0, + "step": 9237 + }, + { + "epoch": 1.0144959367450033, + "grad_norm": 2.5678555965423584, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7150119543075562, + "num_tokens": 234898301.0, + "step": 9238 + }, + { + "epoch": 1.014605754447617, + "grad_norm": 2.342324733734131, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7385501265525818, + "num_tokens": 234921154.0, + "step": 9239 + }, + { + "epoch": 1.0147155721502306, + "grad_norm": 2.0401551723480225, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7200666666030884, + "num_tokens": 234951071.0, + "step": 9240 + }, + { + "epoch": 1.0148253898528443, + "grad_norm": 2.1775779724121094, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.714571475982666, + "num_tokens": 234977257.0, + "step": 9241 + }, + { + "epoch": 1.0149352075554579, + "grad_norm": 2.49157452583313, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7250233888626099, + "num_tokens": 234998271.0, + "step": 9242 + }, + { + "epoch": 1.0150450252580716, + "grad_norm": 2.274843454360962, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7164016962051392, + "num_tokens": 235022309.0, + "step": 9243 + }, + { + "epoch": 1.0151548429606854, + "grad_norm": 2.440173625946045, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7345900535583496, + "num_tokens": 235043765.0, + "step": 9244 + }, + { + "epoch": 1.015264660663299, + "grad_norm": 1.9685769081115723, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6967592239379883, + "num_tokens": 235073944.0, + "step": 9245 + }, + { + "epoch": 1.0153744783659127, + "grad_norm": 2.09895920753479, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7178338170051575, + "num_tokens": 235105332.0, + "step": 9246 + }, + { + "epoch": 1.0154842960685262, + "grad_norm": 2.2384257316589355, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7238870859146118, + "num_tokens": 235131641.0, + "step": 9247 + }, + { + "epoch": 1.01559411377114, + "grad_norm": 2.0577139854431152, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7101703882217407, + "num_tokens": 235159158.0, + "step": 9248 + }, + { + "epoch": 1.0157039314737535, + "grad_norm": 2.1115972995758057, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.736968994140625, + "num_tokens": 235184893.0, + "step": 9249 + }, + { + "epoch": 1.0158137491763672, + "grad_norm": 2.2851686477661133, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7363159656524658, + "num_tokens": 235206750.0, + "step": 9250 + }, + { + "epoch": 1.015923566878981, + "grad_norm": 2.5055627822875977, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7113325595855713, + "num_tokens": 235228249.0, + "step": 9251 + }, + { + "epoch": 1.0160333845815945, + "grad_norm": 2.323730945587158, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.7103022336959839, + "num_tokens": 235252080.0, + "step": 9252 + }, + { + "epoch": 1.0161432022842083, + "grad_norm": 2.3291196823120117, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7468804717063904, + "num_tokens": 235274813.0, + "step": 9253 + }, + { + "epoch": 1.0162530199868218, + "grad_norm": 2.2414705753326416, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7328817248344421, + "num_tokens": 235299796.0, + "step": 9254 + }, + { + "epoch": 1.0163628376894356, + "grad_norm": 2.4259192943573, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7107415199279785, + "num_tokens": 235321316.0, + "step": 9255 + }, + { + "epoch": 1.0164726553920491, + "grad_norm": 2.2865233421325684, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7129462957382202, + "num_tokens": 235345413.0, + "step": 9256 + }, + { + "epoch": 1.0165824730946629, + "grad_norm": 2.4277377128601074, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7375401258468628, + "num_tokens": 235365646.0, + "step": 9257 + }, + { + "epoch": 1.0166922907972764, + "grad_norm": 2.295708179473877, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7027885317802429, + "num_tokens": 235388662.0, + "step": 9258 + }, + { + "epoch": 1.0168021084998902, + "grad_norm": 2.4284424781799316, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7314398288726807, + "num_tokens": 235411621.0, + "step": 9259 + }, + { + "epoch": 1.016911926202504, + "grad_norm": 1.9741164445877075, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.6949414014816284, + "num_tokens": 235441614.0, + "step": 9260 + }, + { + "epoch": 1.0170217439051175, + "grad_norm": 2.31634783744812, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7086068987846375, + "num_tokens": 235465734.0, + "step": 9261 + }, + { + "epoch": 1.0171315616077312, + "grad_norm": 2.1571483612060547, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7037765979766846, + "num_tokens": 235493743.0, + "step": 9262 + }, + { + "epoch": 1.0172413793103448, + "grad_norm": 2.379957437515259, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7343027591705322, + "num_tokens": 235514242.0, + "step": 9263 + }, + { + "epoch": 1.0173511970129585, + "grad_norm": 2.4745595455169678, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.7481118440628052, + "num_tokens": 235533939.0, + "step": 9264 + }, + { + "epoch": 1.017461014715572, + "grad_norm": 2.375206470489502, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7156751751899719, + "num_tokens": 235556288.0, + "step": 9265 + }, + { + "epoch": 1.0175708324181858, + "grad_norm": 2.2730586528778076, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7180765867233276, + "num_tokens": 235581944.0, + "step": 9266 + }, + { + "epoch": 1.0176806501207996, + "grad_norm": 2.121962070465088, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7290781736373901, + "num_tokens": 235610134.0, + "step": 9267 + }, + { + "epoch": 1.017790467823413, + "grad_norm": 2.3545243740081787, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.697908878326416, + "num_tokens": 235635469.0, + "step": 9268 + }, + { + "epoch": 1.0179002855260268, + "grad_norm": 2.641831636428833, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7247293591499329, + "num_tokens": 235653124.0, + "step": 9269 + }, + { + "epoch": 1.0180101032286404, + "grad_norm": 2.2324392795562744, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.711304247379303, + "num_tokens": 235678380.0, + "step": 9270 + }, + { + "epoch": 1.0181199209312541, + "grad_norm": 2.172727346420288, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.713387131690979, + "num_tokens": 235705221.0, + "step": 9271 + }, + { + "epoch": 1.0182297386338677, + "grad_norm": 1.935332179069519, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7178223133087158, + "num_tokens": 235736758.0, + "step": 9272 + }, + { + "epoch": 1.0183395563364814, + "grad_norm": 1.9719698429107666, + "learning_rate": 1e-06, + "loss": 0.8019, + "mean_token_accuracy": 0.7439553737640381, + "num_tokens": 235763986.0, + "step": 9273 + }, + { + "epoch": 1.0184493740390952, + "grad_norm": 2.480123519897461, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7257087230682373, + "num_tokens": 235784853.0, + "step": 9274 + }, + { + "epoch": 1.0185591917417087, + "grad_norm": 2.148306131362915, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6928228139877319, + "num_tokens": 235813388.0, + "step": 9275 + }, + { + "epoch": 1.0186690094443225, + "grad_norm": 2.3460235595703125, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7101181149482727, + "num_tokens": 235838098.0, + "step": 9276 + }, + { + "epoch": 1.018778827146936, + "grad_norm": 2.7291688919067383, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7312952876091003, + "num_tokens": 235855240.0, + "step": 9277 + }, + { + "epoch": 1.0188886448495498, + "grad_norm": 2.158435344696045, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7138704657554626, + "num_tokens": 235881673.0, + "step": 9278 + }, + { + "epoch": 1.0189984625521633, + "grad_norm": 2.0667994022369385, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7122535705566406, + "num_tokens": 235910063.0, + "step": 9279 + }, + { + "epoch": 1.019108280254777, + "grad_norm": 2.350922107696533, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.71888267993927, + "num_tokens": 235934949.0, + "step": 9280 + }, + { + "epoch": 1.0192180979573908, + "grad_norm": 2.5126395225524902, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7345893383026123, + "num_tokens": 235954571.0, + "step": 9281 + }, + { + "epoch": 1.0193279156600044, + "grad_norm": 2.5309898853302, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7230634689331055, + "num_tokens": 235976024.0, + "step": 9282 + }, + { + "epoch": 1.0194377333626181, + "grad_norm": 2.5675339698791504, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7226409316062927, + "num_tokens": 235996728.0, + "step": 9283 + }, + { + "epoch": 1.0195475510652316, + "grad_norm": 2.215782403945923, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7294763326644897, + "num_tokens": 236022236.0, + "step": 9284 + }, + { + "epoch": 1.0196573687678454, + "grad_norm": 2.320850372314453, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7118269205093384, + "num_tokens": 236046969.0, + "step": 9285 + }, + { + "epoch": 1.019767186470459, + "grad_norm": 2.0148885250091553, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7169303297996521, + "num_tokens": 236076294.0, + "step": 9286 + }, + { + "epoch": 1.0198770041730727, + "grad_norm": 2.229562520980835, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7199726700782776, + "num_tokens": 236100890.0, + "step": 9287 + }, + { + "epoch": 1.0199868218756865, + "grad_norm": 1.8289016485214233, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.700299859046936, + "num_tokens": 236137872.0, + "step": 9288 + }, + { + "epoch": 1.0200966395783, + "grad_norm": 2.3836236000061035, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.721630871295929, + "num_tokens": 236160571.0, + "step": 9289 + }, + { + "epoch": 1.0202064572809137, + "grad_norm": 2.2007577419281006, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7337239980697632, + "num_tokens": 236185958.0, + "step": 9290 + }, + { + "epoch": 1.0203162749835273, + "grad_norm": 2.2915847301483154, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7128167152404785, + "num_tokens": 236209682.0, + "step": 9291 + }, + { + "epoch": 1.020426092686141, + "grad_norm": 2.3749327659606934, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.7412078976631165, + "num_tokens": 236230575.0, + "step": 9292 + }, + { + "epoch": 1.0205359103887546, + "grad_norm": 1.9572230577468872, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7443932890892029, + "num_tokens": 236261494.0, + "step": 9293 + }, + { + "epoch": 1.0206457280913683, + "grad_norm": 2.3209996223449707, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7078108191490173, + "num_tokens": 236287973.0, + "step": 9294 + }, + { + "epoch": 1.020755545793982, + "grad_norm": 2.507477283477783, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7183207869529724, + "num_tokens": 236309040.0, + "step": 9295 + }, + { + "epoch": 1.0208653634965956, + "grad_norm": 2.473062038421631, + "learning_rate": 1e-06, + "loss": 0.787, + "mean_token_accuracy": 0.7463382482528687, + "num_tokens": 236328937.0, + "step": 9296 + }, + { + "epoch": 1.0209751811992094, + "grad_norm": 2.2934117317199707, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7107555866241455, + "num_tokens": 236354461.0, + "step": 9297 + }, + { + "epoch": 1.021084998901823, + "grad_norm": 2.137049913406372, + "learning_rate": 1e-06, + "loss": 0.9628, + "mean_token_accuracy": 0.6997730135917664, + "num_tokens": 236383374.0, + "step": 9298 + }, + { + "epoch": 1.0211948166044367, + "grad_norm": 2.1324548721313477, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7227996587753296, + "num_tokens": 236410372.0, + "step": 9299 + }, + { + "epoch": 1.0213046343070502, + "grad_norm": 2.4524917602539062, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7337583303451538, + "num_tokens": 236432810.0, + "step": 9300 + }, + { + "epoch": 1.021414452009664, + "grad_norm": 2.58087158203125, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7011773586273193, + "num_tokens": 236453545.0, + "step": 9301 + }, + { + "epoch": 1.0215242697122777, + "grad_norm": 2.186535358428955, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7259402275085449, + "num_tokens": 236479912.0, + "step": 9302 + }, + { + "epoch": 1.0216340874148913, + "grad_norm": 2.3041188716888428, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7306114435195923, + "num_tokens": 236502983.0, + "step": 9303 + }, + { + "epoch": 1.021743905117505, + "grad_norm": 2.5039825439453125, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7100769281387329, + "num_tokens": 236524322.0, + "step": 9304 + }, + { + "epoch": 1.0218537228201185, + "grad_norm": 1.9875901937484741, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7121230363845825, + "num_tokens": 236557547.0, + "step": 9305 + }, + { + "epoch": 1.0219635405227323, + "grad_norm": 2.3078951835632324, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7281109094619751, + "num_tokens": 236584526.0, + "step": 9306 + }, + { + "epoch": 1.0220733582253458, + "grad_norm": 2.0733232498168945, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7066106200218201, + "num_tokens": 236613539.0, + "step": 9307 + }, + { + "epoch": 1.0221831759279596, + "grad_norm": 2.168447256088257, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7188102602958679, + "num_tokens": 236640935.0, + "step": 9308 + }, + { + "epoch": 1.0222929936305734, + "grad_norm": 1.9842861890792847, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7235171794891357, + "num_tokens": 236669080.0, + "step": 9309 + }, + { + "epoch": 1.022402811333187, + "grad_norm": 2.53593373298645, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7071876525878906, + "num_tokens": 236692461.0, + "step": 9310 + }, + { + "epoch": 1.0225126290358006, + "grad_norm": 2.3183131217956543, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7169253826141357, + "num_tokens": 236719084.0, + "step": 9311 + }, + { + "epoch": 1.0226224467384142, + "grad_norm": 2.1054441928863525, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7057996988296509, + "num_tokens": 236748148.0, + "step": 9312 + }, + { + "epoch": 1.022732264441028, + "grad_norm": 2.109790086746216, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7177563905715942, + "num_tokens": 236773927.0, + "step": 9313 + }, + { + "epoch": 1.0228420821436415, + "grad_norm": 2.1699020862579346, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7295750379562378, + "num_tokens": 236799767.0, + "step": 9314 + }, + { + "epoch": 1.0229518998462552, + "grad_norm": 2.352745294570923, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7227444648742676, + "num_tokens": 236823708.0, + "step": 9315 + }, + { + "epoch": 1.023061717548869, + "grad_norm": 2.2579543590545654, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7465268969535828, + "num_tokens": 236846947.0, + "step": 9316 + }, + { + "epoch": 1.0231715352514825, + "grad_norm": 2.2766082286834717, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7113468647003174, + "num_tokens": 236874057.0, + "step": 9317 + }, + { + "epoch": 1.0232813529540963, + "grad_norm": 2.1816399097442627, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.718894898891449, + "num_tokens": 236899899.0, + "step": 9318 + }, + { + "epoch": 1.0233911706567098, + "grad_norm": 2.2428932189941406, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7169437408447266, + "num_tokens": 236925682.0, + "step": 9319 + }, + { + "epoch": 1.0235009883593236, + "grad_norm": 2.051922082901001, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7055978178977966, + "num_tokens": 236959427.0, + "step": 9320 + }, + { + "epoch": 1.023610806061937, + "grad_norm": 2.0006937980651855, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7125357389450073, + "num_tokens": 236989454.0, + "step": 9321 + }, + { + "epoch": 1.0237206237645509, + "grad_norm": 2.585242509841919, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7321824431419373, + "num_tokens": 237009373.0, + "step": 9322 + }, + { + "epoch": 1.0238304414671644, + "grad_norm": 2.292433261871338, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7183393836021423, + "num_tokens": 237033811.0, + "step": 9323 + }, + { + "epoch": 1.0239402591697782, + "grad_norm": 2.0438952445983887, + "learning_rate": 1e-06, + "loss": 0.9743, + "mean_token_accuracy": 0.7015949487686157, + "num_tokens": 237065736.0, + "step": 9324 + }, + { + "epoch": 1.024050076872392, + "grad_norm": 1.977481722831726, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7176799774169922, + "num_tokens": 237095408.0, + "step": 9325 + }, + { + "epoch": 1.0241598945750054, + "grad_norm": 2.238753080368042, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7244032621383667, + "num_tokens": 237119827.0, + "step": 9326 + }, + { + "epoch": 1.0242697122776192, + "grad_norm": 2.2086291313171387, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.702390193939209, + "num_tokens": 237144614.0, + "step": 9327 + }, + { + "epoch": 1.0243795299802327, + "grad_norm": 2.3909668922424316, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.7448825240135193, + "num_tokens": 237164823.0, + "step": 9328 + }, + { + "epoch": 1.0244893476828465, + "grad_norm": 2.0509393215179443, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7125711441040039, + "num_tokens": 237192847.0, + "step": 9329 + }, + { + "epoch": 1.02459916538546, + "grad_norm": 2.4052670001983643, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7260393500328064, + "num_tokens": 237213776.0, + "step": 9330 + }, + { + "epoch": 1.0247089830880738, + "grad_norm": 2.343688726425171, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7352122664451599, + "num_tokens": 237235751.0, + "step": 9331 + }, + { + "epoch": 1.0248188007906875, + "grad_norm": 1.9945091009140015, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7097871899604797, + "num_tokens": 237265926.0, + "step": 9332 + }, + { + "epoch": 1.024928618493301, + "grad_norm": 2.2471957206726074, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7468338012695312, + "num_tokens": 237289111.0, + "step": 9333 + }, + { + "epoch": 1.0250384361959148, + "grad_norm": 2.378821611404419, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7083806395530701, + "num_tokens": 237312241.0, + "step": 9334 + }, + { + "epoch": 1.0251482538985284, + "grad_norm": 1.9541844129562378, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7345638275146484, + "num_tokens": 237343233.0, + "step": 9335 + }, + { + "epoch": 1.0252580716011421, + "grad_norm": 2.450894355773926, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7206716537475586, + "num_tokens": 237363662.0, + "step": 9336 + }, + { + "epoch": 1.0253678893037557, + "grad_norm": 1.9656071662902832, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7129604816436768, + "num_tokens": 237397881.0, + "step": 9337 + }, + { + "epoch": 1.0254777070063694, + "grad_norm": 2.324540376663208, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7347620725631714, + "num_tokens": 237420968.0, + "step": 9338 + }, + { + "epoch": 1.0255875247089832, + "grad_norm": 2.3930420875549316, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7374582886695862, + "num_tokens": 237443137.0, + "step": 9339 + }, + { + "epoch": 1.0256973424115967, + "grad_norm": 1.9653136730194092, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.6981260776519775, + "num_tokens": 237476397.0, + "step": 9340 + }, + { + "epoch": 1.0258071601142105, + "grad_norm": 2.2355117797851562, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7228640913963318, + "num_tokens": 237502404.0, + "step": 9341 + }, + { + "epoch": 1.025916977816824, + "grad_norm": 2.1428864002227783, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7101054787635803, + "num_tokens": 237532033.0, + "step": 9342 + }, + { + "epoch": 1.0260267955194378, + "grad_norm": 2.188608169555664, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7379342913627625, + "num_tokens": 237556335.0, + "step": 9343 + }, + { + "epoch": 1.0261366132220513, + "grad_norm": 2.1362693309783936, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7241077423095703, + "num_tokens": 237583217.0, + "step": 9344 + }, + { + "epoch": 1.026246430924665, + "grad_norm": 1.9734315872192383, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7020705938339233, + "num_tokens": 237615859.0, + "step": 9345 + }, + { + "epoch": 1.0263562486272788, + "grad_norm": 2.344846725463867, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7257841229438782, + "num_tokens": 237640613.0, + "step": 9346 + }, + { + "epoch": 1.0264660663298923, + "grad_norm": 1.9856245517730713, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7215315103530884, + "num_tokens": 237669020.0, + "step": 9347 + }, + { + "epoch": 1.026575884032506, + "grad_norm": 2.5426251888275146, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.724153995513916, + "num_tokens": 237689323.0, + "step": 9348 + }, + { + "epoch": 1.0266857017351196, + "grad_norm": 2.121340036392212, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7018899917602539, + "num_tokens": 237717775.0, + "step": 9349 + }, + { + "epoch": 1.0267955194377334, + "grad_norm": 2.3614704608917236, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7039210796356201, + "num_tokens": 237743076.0, + "step": 9350 + }, + { + "epoch": 1.026905337140347, + "grad_norm": 2.272040605545044, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7395760416984558, + "num_tokens": 237766623.0, + "step": 9351 + }, + { + "epoch": 1.0270151548429607, + "grad_norm": 2.2122714519500732, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.711226224899292, + "num_tokens": 237794336.0, + "step": 9352 + }, + { + "epoch": 1.0271249725455744, + "grad_norm": 2.2618818283081055, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7409508228302002, + "num_tokens": 237818062.0, + "step": 9353 + }, + { + "epoch": 1.027234790248188, + "grad_norm": 1.9908164739608765, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7234201431274414, + "num_tokens": 237845678.0, + "step": 9354 + }, + { + "epoch": 1.0273446079508017, + "grad_norm": 2.2609755992889404, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7217442989349365, + "num_tokens": 237871637.0, + "step": 9355 + }, + { + "epoch": 1.0274544256534153, + "grad_norm": 2.4655263423919678, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7096735835075378, + "num_tokens": 237893598.0, + "step": 9356 + }, + { + "epoch": 1.027564243356029, + "grad_norm": 2.0347673892974854, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7074247598648071, + "num_tokens": 237922890.0, + "step": 9357 + }, + { + "epoch": 1.0276740610586426, + "grad_norm": 2.610076904296875, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7022863626480103, + "num_tokens": 237943848.0, + "step": 9358 + }, + { + "epoch": 1.0277838787612563, + "grad_norm": 2.179758310317993, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7089695334434509, + "num_tokens": 237968935.0, + "step": 9359 + }, + { + "epoch": 1.02789369646387, + "grad_norm": 2.1691203117370605, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7164453864097595, + "num_tokens": 237996616.0, + "step": 9360 + }, + { + "epoch": 1.0280035141664836, + "grad_norm": 2.036576509475708, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.7013744115829468, + "num_tokens": 238030910.0, + "step": 9361 + }, + { + "epoch": 1.0281133318690974, + "grad_norm": 2.127685070037842, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7264190912246704, + "num_tokens": 238058561.0, + "step": 9362 + }, + { + "epoch": 1.028223149571711, + "grad_norm": 2.1593332290649414, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.6875106692314148, + "num_tokens": 238086859.0, + "step": 9363 + }, + { + "epoch": 1.0283329672743247, + "grad_norm": 2.3228182792663574, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7371052503585815, + "num_tokens": 238109603.0, + "step": 9364 + }, + { + "epoch": 1.0284427849769382, + "grad_norm": 2.094977378845215, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7065399885177612, + "num_tokens": 238140596.0, + "step": 9365 + }, + { + "epoch": 1.028552602679552, + "grad_norm": 2.647294044494629, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7045482397079468, + "num_tokens": 238161749.0, + "step": 9366 + }, + { + "epoch": 1.0286624203821657, + "grad_norm": 2.514824151992798, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7361685037612915, + "num_tokens": 238182056.0, + "step": 9367 + }, + { + "epoch": 1.0287722380847792, + "grad_norm": 2.6498632431030273, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.737791895866394, + "num_tokens": 238206963.0, + "step": 9368 + }, + { + "epoch": 1.028882055787393, + "grad_norm": 2.3074982166290283, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7416560649871826, + "num_tokens": 238229476.0, + "step": 9369 + }, + { + "epoch": 1.0289918734900065, + "grad_norm": 2.337019443511963, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7295930981636047, + "num_tokens": 238256390.0, + "step": 9370 + }, + { + "epoch": 1.0291016911926203, + "grad_norm": 2.1753103733062744, + "learning_rate": 1e-06, + "loss": 1.0281, + "mean_token_accuracy": 0.6854286193847656, + "num_tokens": 238287275.0, + "step": 9371 + }, + { + "epoch": 1.0292115088952338, + "grad_norm": 2.125322103500366, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.719545304775238, + "num_tokens": 238316748.0, + "step": 9372 + }, + { + "epoch": 1.0293213265978476, + "grad_norm": 2.578833818435669, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7333642840385437, + "num_tokens": 238338025.0, + "step": 9373 + }, + { + "epoch": 1.0294311443004613, + "grad_norm": 2.275743246078491, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.719224214553833, + "num_tokens": 238364033.0, + "step": 9374 + }, + { + "epoch": 1.0295409620030749, + "grad_norm": 2.102341890335083, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.7049157619476318, + "num_tokens": 238393632.0, + "step": 9375 + }, + { + "epoch": 1.0296507797056886, + "grad_norm": 2.0517687797546387, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7072490453720093, + "num_tokens": 238421923.0, + "step": 9376 + }, + { + "epoch": 1.0297605974083022, + "grad_norm": 2.4255998134613037, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7149271368980408, + "num_tokens": 238445407.0, + "step": 9377 + }, + { + "epoch": 1.029870415110916, + "grad_norm": 2.159067153930664, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.7002270221710205, + "num_tokens": 238474522.0, + "step": 9378 + }, + { + "epoch": 1.0299802328135295, + "grad_norm": 2.5458457469940186, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7301130890846252, + "num_tokens": 238494928.0, + "step": 9379 + }, + { + "epoch": 1.0300900505161432, + "grad_norm": 2.05179500579834, + "learning_rate": 1e-06, + "loss": 1.0286, + "mean_token_accuracy": 0.6965972781181335, + "num_tokens": 238526594.0, + "step": 9380 + }, + { + "epoch": 1.030199868218757, + "grad_norm": 2.013173818588257, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7037706971168518, + "num_tokens": 238553977.0, + "step": 9381 + }, + { + "epoch": 1.0303096859213705, + "grad_norm": 2.3454272747039795, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7155373096466064, + "num_tokens": 238577158.0, + "step": 9382 + }, + { + "epoch": 1.0304195036239843, + "grad_norm": 2.121392011642456, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.6929100751876831, + "num_tokens": 238607096.0, + "step": 9383 + }, + { + "epoch": 1.0305293213265978, + "grad_norm": 2.1737899780273438, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7154675722122192, + "num_tokens": 238634207.0, + "step": 9384 + }, + { + "epoch": 1.0306391390292116, + "grad_norm": 2.293119192123413, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7095860242843628, + "num_tokens": 238658892.0, + "step": 9385 + }, + { + "epoch": 1.030748956731825, + "grad_norm": 2.300241470336914, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7091220021247864, + "num_tokens": 238684056.0, + "step": 9386 + }, + { + "epoch": 1.0308587744344389, + "grad_norm": 2.3028619289398193, + "learning_rate": 1e-06, + "loss": 1.0555, + "mean_token_accuracy": 0.6846387982368469, + "num_tokens": 238710419.0, + "step": 9387 + }, + { + "epoch": 1.0309685921370524, + "grad_norm": 2.3386688232421875, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7118793725967407, + "num_tokens": 238734072.0, + "step": 9388 + }, + { + "epoch": 1.0310784098396661, + "grad_norm": 2.3589940071105957, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7098755836486816, + "num_tokens": 238759285.0, + "step": 9389 + }, + { + "epoch": 1.03118822754228, + "grad_norm": 2.171124219894409, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.710267961025238, + "num_tokens": 238788362.0, + "step": 9390 + }, + { + "epoch": 1.0312980452448934, + "grad_norm": 2.2000105381011963, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.708829402923584, + "num_tokens": 238814630.0, + "step": 9391 + }, + { + "epoch": 1.0314078629475072, + "grad_norm": 1.9302117824554443, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7207255363464355, + "num_tokens": 238848293.0, + "step": 9392 + }, + { + "epoch": 1.0315176806501207, + "grad_norm": 2.2893168926239014, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7326429486274719, + "num_tokens": 238869819.0, + "step": 9393 + }, + { + "epoch": 1.0316274983527345, + "grad_norm": 2.424955129623413, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7110522389411926, + "num_tokens": 238892042.0, + "step": 9394 + }, + { + "epoch": 1.031737316055348, + "grad_norm": 2.5051703453063965, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7253910899162292, + "num_tokens": 238914240.0, + "step": 9395 + }, + { + "epoch": 1.0318471337579618, + "grad_norm": 4.1103315353393555, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7452943325042725, + "num_tokens": 238933632.0, + "step": 9396 + }, + { + "epoch": 1.0319569514605755, + "grad_norm": 2.203617572784424, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7062288522720337, + "num_tokens": 238961760.0, + "step": 9397 + }, + { + "epoch": 1.032066769163189, + "grad_norm": 2.4034719467163086, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7195637226104736, + "num_tokens": 238984511.0, + "step": 9398 + }, + { + "epoch": 1.0321765868658028, + "grad_norm": 2.4246137142181396, + "learning_rate": 1e-06, + "loss": 1.0162, + "mean_token_accuracy": 0.6902346611022949, + "num_tokens": 239009282.0, + "step": 9399 + }, + { + "epoch": 1.0322864045684164, + "grad_norm": 2.268653631210327, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7292011380195618, + "num_tokens": 239035614.0, + "step": 9400 + }, + { + "epoch": 1.0323962222710301, + "grad_norm": 2.585610866546631, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7111005783081055, + "num_tokens": 239058223.0, + "step": 9401 + }, + { + "epoch": 1.0325060399736437, + "grad_norm": 2.3007051944732666, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.720468282699585, + "num_tokens": 239082481.0, + "step": 9402 + }, + { + "epoch": 1.0326158576762574, + "grad_norm": 2.320080280303955, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7261762619018555, + "num_tokens": 239107523.0, + "step": 9403 + }, + { + "epoch": 1.0327256753788712, + "grad_norm": 2.012674331665039, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7074660062789917, + "num_tokens": 239137114.0, + "step": 9404 + }, + { + "epoch": 1.0328354930814847, + "grad_norm": 2.240250825881958, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7319974303245544, + "num_tokens": 239162670.0, + "step": 9405 + }, + { + "epoch": 1.0329453107840985, + "grad_norm": 2.3139779567718506, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7126555442810059, + "num_tokens": 239186452.0, + "step": 9406 + }, + { + "epoch": 1.033055128486712, + "grad_norm": 2.2171127796173096, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.711067795753479, + "num_tokens": 239211504.0, + "step": 9407 + }, + { + "epoch": 1.0331649461893258, + "grad_norm": 1.955195665359497, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7168900966644287, + "num_tokens": 239246630.0, + "step": 9408 + }, + { + "epoch": 1.0332747638919393, + "grad_norm": 2.307678461074829, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7123268246650696, + "num_tokens": 239271338.0, + "step": 9409 + }, + { + "epoch": 1.033384581594553, + "grad_norm": 2.210662364959717, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7080943584442139, + "num_tokens": 239297720.0, + "step": 9410 + }, + { + "epoch": 1.0334943992971668, + "grad_norm": 2.3129286766052246, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7222590446472168, + "num_tokens": 239321580.0, + "step": 9411 + }, + { + "epoch": 1.0336042169997803, + "grad_norm": 2.0394537448883057, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7131186723709106, + "num_tokens": 239349648.0, + "step": 9412 + }, + { + "epoch": 1.033714034702394, + "grad_norm": 2.3439760208129883, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7032294273376465, + "num_tokens": 239374755.0, + "step": 9413 + }, + { + "epoch": 1.0338238524050076, + "grad_norm": 2.3298933506011963, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7168376445770264, + "num_tokens": 239398711.0, + "step": 9414 + }, + { + "epoch": 1.0339336701076214, + "grad_norm": 2.609041929244995, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7202480435371399, + "num_tokens": 239417422.0, + "step": 9415 + }, + { + "epoch": 1.034043487810235, + "grad_norm": 2.3491568565368652, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7061094641685486, + "num_tokens": 239441063.0, + "step": 9416 + }, + { + "epoch": 1.0341533055128487, + "grad_norm": 2.5402257442474365, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7376521825790405, + "num_tokens": 239459778.0, + "step": 9417 + }, + { + "epoch": 1.0342631232154624, + "grad_norm": 2.3358139991760254, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7188014984130859, + "num_tokens": 239483160.0, + "step": 9418 + }, + { + "epoch": 1.034372940918076, + "grad_norm": 2.4635567665100098, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7214376926422119, + "num_tokens": 239506759.0, + "step": 9419 + }, + { + "epoch": 1.0344827586206897, + "grad_norm": 2.0629708766937256, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7297138571739197, + "num_tokens": 239534377.0, + "step": 9420 + }, + { + "epoch": 1.0345925763233033, + "grad_norm": 2.0180740356445312, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.6991056203842163, + "num_tokens": 239563977.0, + "step": 9421 + }, + { + "epoch": 1.034702394025917, + "grad_norm": 2.499443769454956, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7205637693405151, + "num_tokens": 239585352.0, + "step": 9422 + }, + { + "epoch": 1.0348122117285306, + "grad_norm": 2.464883804321289, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7119875550270081, + "num_tokens": 239607665.0, + "step": 9423 + }, + { + "epoch": 1.0349220294311443, + "grad_norm": 2.5609023571014404, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.7403687238693237, + "num_tokens": 239627404.0, + "step": 9424 + }, + { + "epoch": 1.035031847133758, + "grad_norm": 2.1405930519104004, + "learning_rate": 1e-06, + "loss": 0.9812, + "mean_token_accuracy": 0.6987744569778442, + "num_tokens": 239656032.0, + "step": 9425 + }, + { + "epoch": 1.0351416648363716, + "grad_norm": 2.1293106079101562, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7258809804916382, + "num_tokens": 239681745.0, + "step": 9426 + }, + { + "epoch": 1.0352514825389854, + "grad_norm": 2.0661916732788086, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6953558921813965, + "num_tokens": 239710314.0, + "step": 9427 + }, + { + "epoch": 1.035361300241599, + "grad_norm": 2.4745073318481445, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7190083265304565, + "num_tokens": 239732408.0, + "step": 9428 + }, + { + "epoch": 1.0354711179442126, + "grad_norm": 1.968603253364563, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7024763226509094, + "num_tokens": 239762816.0, + "step": 9429 + }, + { + "epoch": 1.0355809356468262, + "grad_norm": 2.3064253330230713, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7256791591644287, + "num_tokens": 239786805.0, + "step": 9430 + }, + { + "epoch": 1.03569075334944, + "grad_norm": 2.183202028274536, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7095715999603271, + "num_tokens": 239814224.0, + "step": 9431 + }, + { + "epoch": 1.0358005710520537, + "grad_norm": 2.075322151184082, + "learning_rate": 1e-06, + "loss": 0.9789, + "mean_token_accuracy": 0.6990896463394165, + "num_tokens": 239841749.0, + "step": 9432 + }, + { + "epoch": 1.0359103887546672, + "grad_norm": 2.1376454830169678, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6963522434234619, + "num_tokens": 239869758.0, + "step": 9433 + }, + { + "epoch": 1.036020206457281, + "grad_norm": 2.153561592102051, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7106889486312866, + "num_tokens": 239897558.0, + "step": 9434 + }, + { + "epoch": 1.0361300241598945, + "grad_norm": 2.049046754837036, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.708550214767456, + "num_tokens": 239931157.0, + "step": 9435 + }, + { + "epoch": 1.0362398418625083, + "grad_norm": 2.477782964706421, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7183996438980103, + "num_tokens": 239953870.0, + "step": 9436 + }, + { + "epoch": 1.0363496595651218, + "grad_norm": 2.1733155250549316, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.725071370601654, + "num_tokens": 239980860.0, + "step": 9437 + }, + { + "epoch": 1.0364594772677356, + "grad_norm": 2.351832628250122, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7531311511993408, + "num_tokens": 240002200.0, + "step": 9438 + }, + { + "epoch": 1.036569294970349, + "grad_norm": 2.21297287940979, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.715462327003479, + "num_tokens": 240030596.0, + "step": 9439 + }, + { + "epoch": 1.0366791126729629, + "grad_norm": 2.3714396953582764, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7229546308517456, + "num_tokens": 240052497.0, + "step": 9440 + }, + { + "epoch": 1.0367889303755766, + "grad_norm": 2.144033908843994, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7121767997741699, + "num_tokens": 240080528.0, + "step": 9441 + }, + { + "epoch": 1.0368987480781902, + "grad_norm": 2.1561195850372314, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.7403872013092041, + "num_tokens": 240104718.0, + "step": 9442 + }, + { + "epoch": 1.037008565780804, + "grad_norm": 2.0937976837158203, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7028270959854126, + "num_tokens": 240136134.0, + "step": 9443 + }, + { + "epoch": 1.0371183834834174, + "grad_norm": 1.8941806554794312, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.7033324837684631, + "num_tokens": 240171282.0, + "step": 9444 + }, + { + "epoch": 1.0372282011860312, + "grad_norm": 2.180914878845215, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7235474586486816, + "num_tokens": 240199010.0, + "step": 9445 + }, + { + "epoch": 1.037338018888645, + "grad_norm": 2.4778945446014404, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7146525382995605, + "num_tokens": 240221377.0, + "step": 9446 + }, + { + "epoch": 1.0374478365912585, + "grad_norm": 2.4860825538635254, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7325087785720825, + "num_tokens": 240242597.0, + "step": 9447 + }, + { + "epoch": 1.0375576542938723, + "grad_norm": 2.549713134765625, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7218502759933472, + "num_tokens": 240263166.0, + "step": 9448 + }, + { + "epoch": 1.0376674719964858, + "grad_norm": 2.3110358715057373, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7033697962760925, + "num_tokens": 240287420.0, + "step": 9449 + }, + { + "epoch": 1.0377772896990995, + "grad_norm": 2.2491002082824707, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7251320481300354, + "num_tokens": 240311391.0, + "step": 9450 + }, + { + "epoch": 1.037887107401713, + "grad_norm": 2.315991163253784, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7124109268188477, + "num_tokens": 240334555.0, + "step": 9451 + }, + { + "epoch": 1.0379969251043268, + "grad_norm": 2.3192620277404785, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7159286737442017, + "num_tokens": 240358265.0, + "step": 9452 + }, + { + "epoch": 1.0381067428069404, + "grad_norm": 2.1982290744781494, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7338037490844727, + "num_tokens": 240383927.0, + "step": 9453 + }, + { + "epoch": 1.0382165605095541, + "grad_norm": 2.2519938945770264, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7193925380706787, + "num_tokens": 240410073.0, + "step": 9454 + }, + { + "epoch": 1.0383263782121679, + "grad_norm": 2.5527541637420654, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7268498539924622, + "num_tokens": 240429247.0, + "step": 9455 + }, + { + "epoch": 1.0384361959147814, + "grad_norm": 2.434272527694702, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7349913716316223, + "num_tokens": 240450027.0, + "step": 9456 + }, + { + "epoch": 1.0385460136173952, + "grad_norm": 2.4319100379943848, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7041527628898621, + "num_tokens": 240471083.0, + "step": 9457 + }, + { + "epoch": 1.0386558313200087, + "grad_norm": 2.3426671028137207, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7239924669265747, + "num_tokens": 240494538.0, + "step": 9458 + }, + { + "epoch": 1.0387656490226225, + "grad_norm": 2.0315704345703125, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7184516787528992, + "num_tokens": 240525146.0, + "step": 9459 + }, + { + "epoch": 1.038875466725236, + "grad_norm": 2.549437999725342, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7079366445541382, + "num_tokens": 240545350.0, + "step": 9460 + }, + { + "epoch": 1.0389852844278498, + "grad_norm": 2.5052685737609863, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7390461564064026, + "num_tokens": 240565973.0, + "step": 9461 + }, + { + "epoch": 1.0390951021304635, + "grad_norm": 2.184661388397217, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.713309109210968, + "num_tokens": 240593351.0, + "step": 9462 + }, + { + "epoch": 1.039204919833077, + "grad_norm": 2.196275234222412, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7117899656295776, + "num_tokens": 240618329.0, + "step": 9463 + }, + { + "epoch": 1.0393147375356908, + "grad_norm": 2.7025060653686523, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7305633425712585, + "num_tokens": 240638993.0, + "step": 9464 + }, + { + "epoch": 1.0394245552383043, + "grad_norm": 2.342646598815918, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7233604192733765, + "num_tokens": 240664414.0, + "step": 9465 + }, + { + "epoch": 1.039534372940918, + "grad_norm": 2.1267080307006836, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7089498043060303, + "num_tokens": 240693506.0, + "step": 9466 + }, + { + "epoch": 1.0396441906435316, + "grad_norm": 2.2975759506225586, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.726096510887146, + "num_tokens": 240718135.0, + "step": 9467 + }, + { + "epoch": 1.0397540083461454, + "grad_norm": 2.2081992626190186, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7291550636291504, + "num_tokens": 240744763.0, + "step": 9468 + }, + { + "epoch": 1.0398638260487592, + "grad_norm": 2.137495517730713, + "learning_rate": 1e-06, + "loss": 1.0206, + "mean_token_accuracy": 0.6888756155967712, + "num_tokens": 240772355.0, + "step": 9469 + }, + { + "epoch": 1.0399736437513727, + "grad_norm": 2.02293062210083, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7096892595291138, + "num_tokens": 240803125.0, + "step": 9470 + }, + { + "epoch": 1.0400834614539864, + "grad_norm": 2.7193970680236816, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7270614504814148, + "num_tokens": 240822419.0, + "step": 9471 + }, + { + "epoch": 1.0401932791566, + "grad_norm": 2.075822353363037, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7041873931884766, + "num_tokens": 240852114.0, + "step": 9472 + }, + { + "epoch": 1.0403030968592137, + "grad_norm": 2.23256778717041, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7019566297531128, + "num_tokens": 240877748.0, + "step": 9473 + }, + { + "epoch": 1.0404129145618273, + "grad_norm": 2.334810495376587, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7339176535606384, + "num_tokens": 240900445.0, + "step": 9474 + }, + { + "epoch": 1.040522732264441, + "grad_norm": 2.383669137954712, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7266075015068054, + "num_tokens": 240923456.0, + "step": 9475 + }, + { + "epoch": 1.0406325499670548, + "grad_norm": 2.6856911182403564, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.707450270652771, + "num_tokens": 240942281.0, + "step": 9476 + }, + { + "epoch": 1.0407423676696683, + "grad_norm": 2.1218650341033936, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7107142806053162, + "num_tokens": 240969283.0, + "step": 9477 + }, + { + "epoch": 1.040852185372282, + "grad_norm": 1.9088705778121948, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7174866199493408, + "num_tokens": 241002366.0, + "step": 9478 + }, + { + "epoch": 1.0409620030748956, + "grad_norm": 2.226768970489502, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7039203643798828, + "num_tokens": 241028567.0, + "step": 9479 + }, + { + "epoch": 1.0410718207775094, + "grad_norm": 2.2299516201019287, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7054430246353149, + "num_tokens": 241054795.0, + "step": 9480 + }, + { + "epoch": 1.041181638480123, + "grad_norm": 1.998110055923462, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7271664142608643, + "num_tokens": 241085413.0, + "step": 9481 + }, + { + "epoch": 1.0412914561827367, + "grad_norm": 2.081756591796875, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7207132577896118, + "num_tokens": 241114531.0, + "step": 9482 + }, + { + "epoch": 1.0414012738853504, + "grad_norm": 2.249006509780884, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7131211757659912, + "num_tokens": 241140024.0, + "step": 9483 + }, + { + "epoch": 1.041511091587964, + "grad_norm": 2.2956128120422363, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.6999844312667847, + "num_tokens": 241164270.0, + "step": 9484 + }, + { + "epoch": 1.0416209092905777, + "grad_norm": 1.975087285041809, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7159959077835083, + "num_tokens": 241195802.0, + "step": 9485 + }, + { + "epoch": 1.0417307269931912, + "grad_norm": 2.4365901947021484, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7160623669624329, + "num_tokens": 241218760.0, + "step": 9486 + }, + { + "epoch": 1.041840544695805, + "grad_norm": 2.314495325088501, + "learning_rate": 1e-06, + "loss": 0.8141, + "mean_token_accuracy": 0.7406167387962341, + "num_tokens": 241240455.0, + "step": 9487 + }, + { + "epoch": 1.0419503623984185, + "grad_norm": 2.007639169692993, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.7035965323448181, + "num_tokens": 241270750.0, + "step": 9488 + }, + { + "epoch": 1.0420601801010323, + "grad_norm": 2.124221086502075, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7255632877349854, + "num_tokens": 241298121.0, + "step": 9489 + }, + { + "epoch": 1.042169997803646, + "grad_norm": 2.329846143722534, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7122988700866699, + "num_tokens": 241321053.0, + "step": 9490 + }, + { + "epoch": 1.0422798155062596, + "grad_norm": 2.302541494369507, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7163285613059998, + "num_tokens": 241346969.0, + "step": 9491 + }, + { + "epoch": 1.0423896332088733, + "grad_norm": 2.2750191688537598, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7219651341438293, + "num_tokens": 241370661.0, + "step": 9492 + }, + { + "epoch": 1.0424994509114869, + "grad_norm": 2.64072847366333, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7041836380958557, + "num_tokens": 241390901.0, + "step": 9493 + }, + { + "epoch": 1.0426092686141006, + "grad_norm": 2.4026522636413574, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7001375555992126, + "num_tokens": 241412131.0, + "step": 9494 + }, + { + "epoch": 1.0427190863167142, + "grad_norm": 1.8981677293777466, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7195095419883728, + "num_tokens": 241443916.0, + "step": 9495 + }, + { + "epoch": 1.042828904019328, + "grad_norm": 2.129067897796631, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7321935892105103, + "num_tokens": 241470331.0, + "step": 9496 + }, + { + "epoch": 1.0429387217219417, + "grad_norm": 2.421809196472168, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7359925508499146, + "num_tokens": 241492988.0, + "step": 9497 + }, + { + "epoch": 1.0430485394245552, + "grad_norm": 2.3966171741485596, + "learning_rate": 1e-06, + "loss": 0.8375, + "mean_token_accuracy": 0.7398422956466675, + "num_tokens": 241514560.0, + "step": 9498 + }, + { + "epoch": 1.043158357127169, + "grad_norm": 2.117915153503418, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7142846584320068, + "num_tokens": 241544217.0, + "step": 9499 + }, + { + "epoch": 1.0432681748297825, + "grad_norm": 2.2498152256011963, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.729122519493103, + "num_tokens": 241569713.0, + "step": 9500 + }, + { + "epoch": 1.0433779925323963, + "grad_norm": 2.24646258354187, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.705488920211792, + "num_tokens": 241593922.0, + "step": 9501 + }, + { + "epoch": 1.0434878102350098, + "grad_norm": 2.2543981075286865, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.72663414478302, + "num_tokens": 241617202.0, + "step": 9502 + }, + { + "epoch": 1.0435976279376236, + "grad_norm": 2.383479595184326, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.6996921300888062, + "num_tokens": 241640757.0, + "step": 9503 + }, + { + "epoch": 1.043707445640237, + "grad_norm": 2.227907180786133, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7328450679779053, + "num_tokens": 241665047.0, + "step": 9504 + }, + { + "epoch": 1.0438172633428509, + "grad_norm": 2.368872880935669, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7202728390693665, + "num_tokens": 241688019.0, + "step": 9505 + }, + { + "epoch": 1.0439270810454646, + "grad_norm": 2.479860305786133, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.722783625125885, + "num_tokens": 241710812.0, + "step": 9506 + }, + { + "epoch": 1.0440368987480781, + "grad_norm": 2.256195306777954, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7203499674797058, + "num_tokens": 241735783.0, + "step": 9507 + }, + { + "epoch": 1.044146716450692, + "grad_norm": 2.426227331161499, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7205289602279663, + "num_tokens": 241758677.0, + "step": 9508 + }, + { + "epoch": 1.0442565341533054, + "grad_norm": 2.0059938430786133, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7176464796066284, + "num_tokens": 241788797.0, + "step": 9509 + }, + { + "epoch": 1.0443663518559192, + "grad_norm": 2.379074811935425, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.737450361251831, + "num_tokens": 241811585.0, + "step": 9510 + }, + { + "epoch": 1.0444761695585327, + "grad_norm": 2.2182016372680664, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7329215407371521, + "num_tokens": 241835232.0, + "step": 9511 + }, + { + "epoch": 1.0445859872611465, + "grad_norm": 2.604862928390503, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7393929958343506, + "num_tokens": 241853984.0, + "step": 9512 + }, + { + "epoch": 1.0446958049637602, + "grad_norm": 2.223268508911133, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7279559373855591, + "num_tokens": 241879167.0, + "step": 9513 + }, + { + "epoch": 1.0448056226663738, + "grad_norm": 2.2753846645355225, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7276073098182678, + "num_tokens": 241902677.0, + "step": 9514 + }, + { + "epoch": 1.0449154403689875, + "grad_norm": 2.148920774459839, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.6899470686912537, + "num_tokens": 241929398.0, + "step": 9515 + }, + { + "epoch": 1.045025258071601, + "grad_norm": 2.185920476913452, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7367839217185974, + "num_tokens": 241954257.0, + "step": 9516 + }, + { + "epoch": 1.0451350757742148, + "grad_norm": 2.404634475708008, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7253114581108093, + "num_tokens": 241977506.0, + "step": 9517 + }, + { + "epoch": 1.0452448934768284, + "grad_norm": 2.323180913925171, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.6880146861076355, + "num_tokens": 242004776.0, + "step": 9518 + }, + { + "epoch": 1.0453547111794421, + "grad_norm": 2.409257173538208, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7021468877792358, + "num_tokens": 242028737.0, + "step": 9519 + }, + { + "epoch": 1.0454645288820559, + "grad_norm": 2.0777883529663086, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7035956978797913, + "num_tokens": 242058523.0, + "step": 9520 + }, + { + "epoch": 1.0455743465846694, + "grad_norm": 2.1678965091705322, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7027259469032288, + "num_tokens": 242088205.0, + "step": 9521 + }, + { + "epoch": 1.0456841642872832, + "grad_norm": 2.174017906188965, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7122766375541687, + "num_tokens": 242114766.0, + "step": 9522 + }, + { + "epoch": 1.0457939819898967, + "grad_norm": 2.1658732891082764, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.6968085765838623, + "num_tokens": 242143971.0, + "step": 9523 + }, + { + "epoch": 1.0459037996925105, + "grad_norm": 1.9473448991775513, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7371651530265808, + "num_tokens": 242174644.0, + "step": 9524 + }, + { + "epoch": 1.046013617395124, + "grad_norm": 2.5071005821228027, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7220560312271118, + "num_tokens": 242195797.0, + "step": 9525 + }, + { + "epoch": 1.0461234350977378, + "grad_norm": 2.126725673675537, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7304857969284058, + "num_tokens": 242223030.0, + "step": 9526 + }, + { + "epoch": 1.0462332528003515, + "grad_norm": 2.373821258544922, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7168728113174438, + "num_tokens": 242247202.0, + "step": 9527 + }, + { + "epoch": 1.046343070502965, + "grad_norm": 2.218991279602051, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7144352793693542, + "num_tokens": 242272058.0, + "step": 9528 + }, + { + "epoch": 1.0464528882055788, + "grad_norm": 2.0684151649475098, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7302100658416748, + "num_tokens": 242300584.0, + "step": 9529 + }, + { + "epoch": 1.0465627059081923, + "grad_norm": 2.092818260192871, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7135078310966492, + "num_tokens": 242329322.0, + "step": 9530 + }, + { + "epoch": 1.046672523610806, + "grad_norm": 2.0245227813720703, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.7175354957580566, + "num_tokens": 242359669.0, + "step": 9531 + }, + { + "epoch": 1.0467823413134196, + "grad_norm": 2.612985849380493, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7371389865875244, + "num_tokens": 242378607.0, + "step": 9532 + }, + { + "epoch": 1.0468921590160334, + "grad_norm": 2.4532313346862793, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7371875047683716, + "num_tokens": 242401168.0, + "step": 9533 + }, + { + "epoch": 1.0470019767186471, + "grad_norm": 2.188159942626953, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7450117468833923, + "num_tokens": 242425949.0, + "step": 9534 + }, + { + "epoch": 1.0471117944212607, + "grad_norm": 2.3950846195220947, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7317894697189331, + "num_tokens": 242448611.0, + "step": 9535 + }, + { + "epoch": 1.0472216121238744, + "grad_norm": 2.278317451477051, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7076050043106079, + "num_tokens": 242473666.0, + "step": 9536 + }, + { + "epoch": 1.047331429826488, + "grad_norm": 2.1280808448791504, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7178611755371094, + "num_tokens": 242500673.0, + "step": 9537 + }, + { + "epoch": 1.0474412475291017, + "grad_norm": 2.238711357116699, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7212439775466919, + "num_tokens": 242528418.0, + "step": 9538 + }, + { + "epoch": 1.0475510652317153, + "grad_norm": 2.194843053817749, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7102068662643433, + "num_tokens": 242555896.0, + "step": 9539 + }, + { + "epoch": 1.047660882934329, + "grad_norm": 2.249051809310913, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7273973822593689, + "num_tokens": 242582600.0, + "step": 9540 + }, + { + "epoch": 1.0477707006369428, + "grad_norm": 2.4248266220092773, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7251185774803162, + "num_tokens": 242603339.0, + "step": 9541 + }, + { + "epoch": 1.0478805183395563, + "grad_norm": 2.8245363235473633, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7158839702606201, + "num_tokens": 242620557.0, + "step": 9542 + }, + { + "epoch": 1.04799033604217, + "grad_norm": 2.087692975997925, + "learning_rate": 1e-06, + "loss": 1.0289, + "mean_token_accuracy": 0.6833863854408264, + "num_tokens": 242652719.0, + "step": 9543 + }, + { + "epoch": 1.0481001537447836, + "grad_norm": 2.210737943649292, + "learning_rate": 1e-06, + "loss": 1.0226, + "mean_token_accuracy": 0.6881843209266663, + "num_tokens": 242680274.0, + "step": 9544 + }, + { + "epoch": 1.0482099714473974, + "grad_norm": 2.4037885665893555, + "learning_rate": 1e-06, + "loss": 0.7868, + "mean_token_accuracy": 0.7457756996154785, + "num_tokens": 242701391.0, + "step": 9545 + }, + { + "epoch": 1.048319789150011, + "grad_norm": 2.270359516143799, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.6972675919532776, + "num_tokens": 242726754.0, + "step": 9546 + }, + { + "epoch": 1.0484296068526247, + "grad_norm": 2.1650562286376953, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7041476368904114, + "num_tokens": 242755432.0, + "step": 9547 + }, + { + "epoch": 1.0485394245552384, + "grad_norm": 2.197399616241455, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7017956972122192, + "num_tokens": 242781387.0, + "step": 9548 + }, + { + "epoch": 1.048649242257852, + "grad_norm": 2.094374895095825, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7376136779785156, + "num_tokens": 242808421.0, + "step": 9549 + }, + { + "epoch": 1.0487590599604657, + "grad_norm": 2.2880043983459473, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7120112776756287, + "num_tokens": 242831504.0, + "step": 9550 + }, + { + "epoch": 1.0488688776630792, + "grad_norm": 2.3510239124298096, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7147443294525146, + "num_tokens": 242854015.0, + "step": 9551 + }, + { + "epoch": 1.048978695365693, + "grad_norm": 2.2753796577453613, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7109453678131104, + "num_tokens": 242877734.0, + "step": 9552 + }, + { + "epoch": 1.0490885130683065, + "grad_norm": 2.519475221633911, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7389177083969116, + "num_tokens": 242898083.0, + "step": 9553 + }, + { + "epoch": 1.0491983307709203, + "grad_norm": 2.3319249153137207, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7428648471832275, + "num_tokens": 242921026.0, + "step": 9554 + }, + { + "epoch": 1.0493081484735338, + "grad_norm": 2.176815986633301, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.7014439105987549, + "num_tokens": 242948789.0, + "step": 9555 + }, + { + "epoch": 1.0494179661761476, + "grad_norm": 2.331113338470459, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.6989903450012207, + "num_tokens": 242972322.0, + "step": 9556 + }, + { + "epoch": 1.0495277838787613, + "grad_norm": 1.9533499479293823, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.714572548866272, + "num_tokens": 243004467.0, + "step": 9557 + }, + { + "epoch": 1.0496376015813749, + "grad_norm": 2.149073600769043, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7004738450050354, + "num_tokens": 243032296.0, + "step": 9558 + }, + { + "epoch": 1.0497474192839886, + "grad_norm": 2.2337582111358643, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7070277333259583, + "num_tokens": 243058607.0, + "step": 9559 + }, + { + "epoch": 1.0498572369866022, + "grad_norm": 2.505202293395996, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.717389702796936, + "num_tokens": 243079129.0, + "step": 9560 + }, + { + "epoch": 1.049967054689216, + "grad_norm": 2.4287047386169434, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7210482954978943, + "num_tokens": 243102297.0, + "step": 9561 + }, + { + "epoch": 1.0500768723918297, + "grad_norm": 1.855892300605774, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7103077173233032, + "num_tokens": 243136684.0, + "step": 9562 + }, + { + "epoch": 1.0501866900944432, + "grad_norm": 2.410928726196289, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7281978130340576, + "num_tokens": 243158262.0, + "step": 9563 + }, + { + "epoch": 1.050296507797057, + "grad_norm": 2.5531258583068848, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7252025008201599, + "num_tokens": 243178896.0, + "step": 9564 + }, + { + "epoch": 1.0504063254996705, + "grad_norm": 2.0279924869537354, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7191203832626343, + "num_tokens": 243211321.0, + "step": 9565 + }, + { + "epoch": 1.0505161432022843, + "grad_norm": 2.0041353702545166, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7007541656494141, + "num_tokens": 243243856.0, + "step": 9566 + }, + { + "epoch": 1.0506259609048978, + "grad_norm": 2.3283050060272217, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7151050567626953, + "num_tokens": 243267722.0, + "step": 9567 + }, + { + "epoch": 1.0507357786075116, + "grad_norm": 2.5012128353118896, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7330520749092102, + "num_tokens": 243289438.0, + "step": 9568 + }, + { + "epoch": 1.050845596310125, + "grad_norm": 2.3905997276306152, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7101103067398071, + "num_tokens": 243312567.0, + "step": 9569 + }, + { + "epoch": 1.0509554140127388, + "grad_norm": 2.569098949432373, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.746650218963623, + "num_tokens": 243331473.0, + "step": 9570 + }, + { + "epoch": 1.0510652317153526, + "grad_norm": 2.3838047981262207, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.695204496383667, + "num_tokens": 243356935.0, + "step": 9571 + }, + { + "epoch": 1.0511750494179661, + "grad_norm": 2.3614542484283447, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7175556421279907, + "num_tokens": 243380106.0, + "step": 9572 + }, + { + "epoch": 1.05128486712058, + "grad_norm": 2.141831636428833, + "learning_rate": 1e-06, + "loss": 0.7571, + "mean_token_accuracy": 0.7564957141876221, + "num_tokens": 243406786.0, + "step": 9573 + }, + { + "epoch": 1.0513946848231934, + "grad_norm": 2.1348493099212646, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7146459817886353, + "num_tokens": 243432873.0, + "step": 9574 + }, + { + "epoch": 1.0515045025258072, + "grad_norm": 2.412137746810913, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7251996994018555, + "num_tokens": 243455618.0, + "step": 9575 + }, + { + "epoch": 1.0516143202284207, + "grad_norm": 2.183950662612915, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7335418462753296, + "num_tokens": 243481326.0, + "step": 9576 + }, + { + "epoch": 1.0517241379310345, + "grad_norm": 2.1969573497772217, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7126618027687073, + "num_tokens": 243507164.0, + "step": 9577 + }, + { + "epoch": 1.0518339556336482, + "grad_norm": 2.249782085418701, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7189335227012634, + "num_tokens": 243532282.0, + "step": 9578 + }, + { + "epoch": 1.0519437733362618, + "grad_norm": 2.5392544269561768, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7394031286239624, + "num_tokens": 243551144.0, + "step": 9579 + }, + { + "epoch": 1.0520535910388755, + "grad_norm": 2.2415108680725098, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.716591477394104, + "num_tokens": 243576174.0, + "step": 9580 + }, + { + "epoch": 1.052163408741489, + "grad_norm": 2.3863961696624756, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7302546501159668, + "num_tokens": 243598121.0, + "step": 9581 + }, + { + "epoch": 1.0522732264441028, + "grad_norm": 2.1881210803985596, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7215234041213989, + "num_tokens": 243624821.0, + "step": 9582 + }, + { + "epoch": 1.0523830441467164, + "grad_norm": 2.2052431106567383, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6933832764625549, + "num_tokens": 243652140.0, + "step": 9583 + }, + { + "epoch": 1.05249286184933, + "grad_norm": 2.08516263961792, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.6966440677642822, + "num_tokens": 243682239.0, + "step": 9584 + }, + { + "epoch": 1.0526026795519439, + "grad_norm": 2.1635711193084717, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7230132818222046, + "num_tokens": 243708543.0, + "step": 9585 + }, + { + "epoch": 1.0527124972545574, + "grad_norm": 2.5535457134246826, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.709419846534729, + "num_tokens": 243729801.0, + "step": 9586 + }, + { + "epoch": 1.0528223149571712, + "grad_norm": 2.4675183296203613, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7183136940002441, + "num_tokens": 243752225.0, + "step": 9587 + }, + { + "epoch": 1.0529321326597847, + "grad_norm": 1.9809377193450928, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7202128171920776, + "num_tokens": 243782368.0, + "step": 9588 + }, + { + "epoch": 1.0530419503623984, + "grad_norm": 2.5382096767425537, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.742286205291748, + "num_tokens": 243803185.0, + "step": 9589 + }, + { + "epoch": 1.053151768065012, + "grad_norm": 1.9940091371536255, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6900560855865479, + "num_tokens": 243835830.0, + "step": 9590 + }, + { + "epoch": 1.0532615857676257, + "grad_norm": 2.469839096069336, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.724530816078186, + "num_tokens": 243859080.0, + "step": 9591 + }, + { + "epoch": 1.0533714034702395, + "grad_norm": 2.3379147052764893, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7063596844673157, + "num_tokens": 243883039.0, + "step": 9592 + }, + { + "epoch": 1.053481221172853, + "grad_norm": 2.111868381500244, + "learning_rate": 1e-06, + "loss": 0.7795, + "mean_token_accuracy": 0.7522251009941101, + "num_tokens": 243909076.0, + "step": 9593 + }, + { + "epoch": 1.0535910388754668, + "grad_norm": 2.154756784439087, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7017413377761841, + "num_tokens": 243937487.0, + "step": 9594 + }, + { + "epoch": 1.0537008565780803, + "grad_norm": 1.9613648653030396, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.6830384135246277, + "num_tokens": 243971177.0, + "step": 9595 + }, + { + "epoch": 1.053810674280694, + "grad_norm": 2.068819284439087, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7317723035812378, + "num_tokens": 243999329.0, + "step": 9596 + }, + { + "epoch": 1.0539204919833076, + "grad_norm": 2.2209742069244385, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7311118841171265, + "num_tokens": 244024016.0, + "step": 9597 + }, + { + "epoch": 1.0540303096859214, + "grad_norm": 2.242875337600708, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.6961769461631775, + "num_tokens": 244049770.0, + "step": 9598 + }, + { + "epoch": 1.0541401273885351, + "grad_norm": 2.0022900104522705, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7066479921340942, + "num_tokens": 244080582.0, + "step": 9599 + }, + { + "epoch": 1.0542499450911487, + "grad_norm": 1.763818621635437, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6862679719924927, + "num_tokens": 244118145.0, + "step": 9600 + }, + { + "epoch": 1.0543597627937624, + "grad_norm": 2.4790050983428955, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7184480428695679, + "num_tokens": 244139686.0, + "step": 9601 + }, + { + "epoch": 1.054469580496376, + "grad_norm": 2.4143776893615723, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7234828472137451, + "num_tokens": 244163193.0, + "step": 9602 + }, + { + "epoch": 1.0545793981989897, + "grad_norm": 2.4541895389556885, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7178959846496582, + "num_tokens": 244186119.0, + "step": 9603 + }, + { + "epoch": 1.0546892159016032, + "grad_norm": 2.5462567806243896, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7275954484939575, + "num_tokens": 244210373.0, + "step": 9604 + }, + { + "epoch": 1.054799033604217, + "grad_norm": 2.5864243507385254, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7124853730201721, + "num_tokens": 244232192.0, + "step": 9605 + }, + { + "epoch": 1.0549088513068308, + "grad_norm": 2.4525859355926514, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.729253888130188, + "num_tokens": 244253969.0, + "step": 9606 + }, + { + "epoch": 1.0550186690094443, + "grad_norm": 2.317060708999634, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7329508066177368, + "num_tokens": 244279028.0, + "step": 9607 + }, + { + "epoch": 1.055128486712058, + "grad_norm": 2.130718469619751, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.6992205381393433, + "num_tokens": 244308524.0, + "step": 9608 + }, + { + "epoch": 1.0552383044146716, + "grad_norm": 2.086886405944824, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7395976781845093, + "num_tokens": 244337481.0, + "step": 9609 + }, + { + "epoch": 1.0553481221172853, + "grad_norm": 2.3983373641967773, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7338509559631348, + "num_tokens": 244358484.0, + "step": 9610 + }, + { + "epoch": 1.0554579398198989, + "grad_norm": 2.283406972885132, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.719133734703064, + "num_tokens": 244384467.0, + "step": 9611 + }, + { + "epoch": 1.0555677575225126, + "grad_norm": 2.3535072803497314, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7141416072845459, + "num_tokens": 244407447.0, + "step": 9612 + }, + { + "epoch": 1.0556775752251264, + "grad_norm": 2.230152130126953, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7353571653366089, + "num_tokens": 244432298.0, + "step": 9613 + }, + { + "epoch": 1.05578739292774, + "grad_norm": 2.1688857078552246, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.695542573928833, + "num_tokens": 244460090.0, + "step": 9614 + }, + { + "epoch": 1.0558972106303537, + "grad_norm": 2.616185188293457, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7135089635848999, + "num_tokens": 244478483.0, + "step": 9615 + }, + { + "epoch": 1.0560070283329672, + "grad_norm": 2.1357614994049072, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7305148839950562, + "num_tokens": 244507658.0, + "step": 9616 + }, + { + "epoch": 1.056116846035581, + "grad_norm": 2.1619837284088135, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7124621868133545, + "num_tokens": 244535843.0, + "step": 9617 + }, + { + "epoch": 1.0562266637381945, + "grad_norm": 2.1085376739501953, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7081311941146851, + "num_tokens": 244565281.0, + "step": 9618 + }, + { + "epoch": 1.0563364814408083, + "grad_norm": 2.282020092010498, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7199721336364746, + "num_tokens": 244588252.0, + "step": 9619 + }, + { + "epoch": 1.0564462991434218, + "grad_norm": 1.9891117811203003, + "learning_rate": 1e-06, + "loss": 1.0096, + "mean_token_accuracy": 0.6905125379562378, + "num_tokens": 244619427.0, + "step": 9620 + }, + { + "epoch": 1.0565561168460356, + "grad_norm": 2.2490665912628174, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7111196517944336, + "num_tokens": 244644627.0, + "step": 9621 + }, + { + "epoch": 1.0566659345486493, + "grad_norm": 2.1235597133636475, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7096755504608154, + "num_tokens": 244673425.0, + "step": 9622 + }, + { + "epoch": 1.0567757522512629, + "grad_norm": 2.2967512607574463, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7311393022537231, + "num_tokens": 244696023.0, + "step": 9623 + }, + { + "epoch": 1.0568855699538766, + "grad_norm": 2.44927716255188, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7129291296005249, + "num_tokens": 244718745.0, + "step": 9624 + }, + { + "epoch": 1.0569953876564901, + "grad_norm": 1.9321181774139404, + "learning_rate": 1e-06, + "loss": 0.9824, + "mean_token_accuracy": 0.6993849277496338, + "num_tokens": 244752838.0, + "step": 9625 + }, + { + "epoch": 1.057105205359104, + "grad_norm": 2.348442792892456, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7205952405929565, + "num_tokens": 244776347.0, + "step": 9626 + }, + { + "epoch": 1.0572150230617177, + "grad_norm": 2.2394347190856934, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7121110558509827, + "num_tokens": 244802611.0, + "step": 9627 + }, + { + "epoch": 1.0573248407643312, + "grad_norm": 2.424088716506958, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7376968860626221, + "num_tokens": 244825176.0, + "step": 9628 + }, + { + "epoch": 1.057434658466945, + "grad_norm": 2.383192777633667, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7309478521347046, + "num_tokens": 244847847.0, + "step": 9629 + }, + { + "epoch": 1.0575444761695585, + "grad_norm": 2.104639768600464, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.7383878827095032, + "num_tokens": 244874355.0, + "step": 9630 + }, + { + "epoch": 1.0576542938721722, + "grad_norm": 2.382725238800049, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7406930327415466, + "num_tokens": 244894825.0, + "step": 9631 + }, + { + "epoch": 1.0577641115747858, + "grad_norm": 2.2319090366363525, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7111476063728333, + "num_tokens": 244920355.0, + "step": 9632 + }, + { + "epoch": 1.0578739292773995, + "grad_norm": 2.5410211086273193, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7088095545768738, + "num_tokens": 244943985.0, + "step": 9633 + }, + { + "epoch": 1.057983746980013, + "grad_norm": 2.1707465648651123, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.702580451965332, + "num_tokens": 244972193.0, + "step": 9634 + }, + { + "epoch": 1.0580935646826268, + "grad_norm": 2.3366446495056152, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7504755854606628, + "num_tokens": 244994349.0, + "step": 9635 + }, + { + "epoch": 1.0582033823852406, + "grad_norm": 2.313624620437622, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.6984917521476746, + "num_tokens": 245018790.0, + "step": 9636 + }, + { + "epoch": 1.0583132000878541, + "grad_norm": 2.1589102745056152, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7110037803649902, + "num_tokens": 245045364.0, + "step": 9637 + }, + { + "epoch": 1.0584230177904679, + "grad_norm": 2.2662034034729004, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7245522737503052, + "num_tokens": 245068637.0, + "step": 9638 + }, + { + "epoch": 1.0585328354930814, + "grad_norm": 2.265004873275757, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7114046812057495, + "num_tokens": 245095005.0, + "step": 9639 + }, + { + "epoch": 1.0586426531956952, + "grad_norm": 2.0129404067993164, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.705376148223877, + "num_tokens": 245126225.0, + "step": 9640 + }, + { + "epoch": 1.0587524708983087, + "grad_norm": 2.4223361015319824, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7376479506492615, + "num_tokens": 245148191.0, + "step": 9641 + }, + { + "epoch": 1.0588622886009225, + "grad_norm": 2.262852191925049, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7226759791374207, + "num_tokens": 245173340.0, + "step": 9642 + }, + { + "epoch": 1.0589721063035362, + "grad_norm": 2.487064838409424, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7375681400299072, + "num_tokens": 245195606.0, + "step": 9643 + }, + { + "epoch": 1.0590819240061498, + "grad_norm": 2.592207908630371, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.737305760383606, + "num_tokens": 245216021.0, + "step": 9644 + }, + { + "epoch": 1.0591917417087635, + "grad_norm": 2.1247143745422363, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7350291013717651, + "num_tokens": 245244174.0, + "step": 9645 + }, + { + "epoch": 1.059301559411377, + "grad_norm": 2.2905497550964355, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.6999144554138184, + "num_tokens": 245270472.0, + "step": 9646 + }, + { + "epoch": 1.0594113771139908, + "grad_norm": 2.301010847091675, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.723156750202179, + "num_tokens": 245294951.0, + "step": 9647 + }, + { + "epoch": 1.0595211948166043, + "grad_norm": 2.1906681060791016, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7215367555618286, + "num_tokens": 245320211.0, + "step": 9648 + }, + { + "epoch": 1.059631012519218, + "grad_norm": 2.1912715435028076, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.6923868060112, + "num_tokens": 245348389.0, + "step": 9649 + }, + { + "epoch": 1.0597408302218319, + "grad_norm": 2.5702919960021973, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.707779586315155, + "num_tokens": 245368618.0, + "step": 9650 + }, + { + "epoch": 1.0598506479244454, + "grad_norm": 2.0054705142974854, + "learning_rate": 1e-06, + "loss": 0.9715, + "mean_token_accuracy": 0.699486255645752, + "num_tokens": 245402195.0, + "step": 9651 + }, + { + "epoch": 1.0599604656270591, + "grad_norm": 2.292837381362915, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7219602465629578, + "num_tokens": 245428106.0, + "step": 9652 + }, + { + "epoch": 1.0600702833296727, + "grad_norm": 1.8430707454681396, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7041423320770264, + "num_tokens": 245463476.0, + "step": 9653 + }, + { + "epoch": 1.0601801010322864, + "grad_norm": 2.374887704849243, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7156461477279663, + "num_tokens": 245486891.0, + "step": 9654 + }, + { + "epoch": 1.0602899187349, + "grad_norm": 2.4270238876342773, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7111750841140747, + "num_tokens": 245509974.0, + "step": 9655 + }, + { + "epoch": 1.0603997364375137, + "grad_norm": 2.4881374835968018, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7120751738548279, + "num_tokens": 245533411.0, + "step": 9656 + }, + { + "epoch": 1.0605095541401275, + "grad_norm": 2.1539688110351562, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7205905914306641, + "num_tokens": 245560863.0, + "step": 9657 + }, + { + "epoch": 1.060619371842741, + "grad_norm": 2.09454083442688, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7104356288909912, + "num_tokens": 245587866.0, + "step": 9658 + }, + { + "epoch": 1.0607291895453548, + "grad_norm": 2.7496531009674072, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7231271862983704, + "num_tokens": 245606947.0, + "step": 9659 + }, + { + "epoch": 1.0608390072479683, + "grad_norm": 2.1979281902313232, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7047337293624878, + "num_tokens": 245635519.0, + "step": 9660 + }, + { + "epoch": 1.060948824950582, + "grad_norm": 2.4760539531707764, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7271196842193604, + "num_tokens": 245656158.0, + "step": 9661 + }, + { + "epoch": 1.0610586426531956, + "grad_norm": 2.04378342628479, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7012239694595337, + "num_tokens": 245686065.0, + "step": 9662 + }, + { + "epoch": 1.0611684603558094, + "grad_norm": 2.1866745948791504, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7318360209465027, + "num_tokens": 245710267.0, + "step": 9663 + }, + { + "epoch": 1.0612782780584231, + "grad_norm": 2.2402772903442383, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7291549444198608, + "num_tokens": 245734660.0, + "step": 9664 + }, + { + "epoch": 1.0613880957610367, + "grad_norm": 2.065450668334961, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7103644609451294, + "num_tokens": 245766532.0, + "step": 9665 + }, + { + "epoch": 1.0614979134636504, + "grad_norm": 2.48580265045166, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7110627889633179, + "num_tokens": 245789715.0, + "step": 9666 + }, + { + "epoch": 1.061607731166264, + "grad_norm": 2.154808282852173, + "learning_rate": 1e-06, + "loss": 0.834, + "mean_token_accuracy": 0.7333762049674988, + "num_tokens": 245816993.0, + "step": 9667 + }, + { + "epoch": 1.0617175488688777, + "grad_norm": 2.2108538150787354, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7106395363807678, + "num_tokens": 245843877.0, + "step": 9668 + }, + { + "epoch": 1.0618273665714912, + "grad_norm": 2.298292398452759, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7229721546173096, + "num_tokens": 245866811.0, + "step": 9669 + }, + { + "epoch": 1.061937184274105, + "grad_norm": 2.325192928314209, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7215746641159058, + "num_tokens": 245889794.0, + "step": 9670 + }, + { + "epoch": 1.0620470019767188, + "grad_norm": 2.297489643096924, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7400232553482056, + "num_tokens": 245912246.0, + "step": 9671 + }, + { + "epoch": 1.0621568196793323, + "grad_norm": 2.198153495788574, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7310447692871094, + "num_tokens": 245938646.0, + "step": 9672 + }, + { + "epoch": 1.062266637381946, + "grad_norm": 2.0626285076141357, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7162915468215942, + "num_tokens": 245966822.0, + "step": 9673 + }, + { + "epoch": 1.0623764550845596, + "grad_norm": 2.180180788040161, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.6939407587051392, + "num_tokens": 245997621.0, + "step": 9674 + }, + { + "epoch": 1.0624862727871733, + "grad_norm": 2.657670497894287, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7194825410842896, + "num_tokens": 246018919.0, + "step": 9675 + }, + { + "epoch": 1.0625960904897869, + "grad_norm": 2.522775650024414, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7338023781776428, + "num_tokens": 246040418.0, + "step": 9676 + }, + { + "epoch": 1.0627059081924006, + "grad_norm": 2.3601953983306885, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7132058143615723, + "num_tokens": 246064284.0, + "step": 9677 + }, + { + "epoch": 1.0628157258950144, + "grad_norm": 2.218989133834839, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7095334529876709, + "num_tokens": 246089479.0, + "step": 9678 + }, + { + "epoch": 1.062925543597628, + "grad_norm": 1.8821685314178467, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7137105464935303, + "num_tokens": 246124355.0, + "step": 9679 + }, + { + "epoch": 1.0630353613002417, + "grad_norm": 2.6132023334503174, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7370923757553101, + "num_tokens": 246145791.0, + "step": 9680 + }, + { + "epoch": 1.0631451790028552, + "grad_norm": 2.510113477706909, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7186523675918579, + "num_tokens": 246168026.0, + "step": 9681 + }, + { + "epoch": 1.063254996705469, + "grad_norm": 2.1972835063934326, + "learning_rate": 1e-06, + "loss": 1.0547, + "mean_token_accuracy": 0.6875091195106506, + "num_tokens": 246198053.0, + "step": 9682 + }, + { + "epoch": 1.0633648144080825, + "grad_norm": 2.1289377212524414, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7086265683174133, + "num_tokens": 246227562.0, + "step": 9683 + }, + { + "epoch": 1.0634746321106963, + "grad_norm": 2.095709800720215, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.6971774101257324, + "num_tokens": 246255411.0, + "step": 9684 + }, + { + "epoch": 1.0635844498133098, + "grad_norm": 2.3679213523864746, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7176137566566467, + "num_tokens": 246279599.0, + "step": 9685 + }, + { + "epoch": 1.0636942675159236, + "grad_norm": 2.721774101257324, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7267753481864929, + "num_tokens": 246299691.0, + "step": 9686 + }, + { + "epoch": 1.0638040852185373, + "grad_norm": 2.173863172531128, + "learning_rate": 1e-06, + "loss": 1.0333, + "mean_token_accuracy": 0.6839011907577515, + "num_tokens": 246327948.0, + "step": 9687 + }, + { + "epoch": 1.0639139029211508, + "grad_norm": 2.361128091812134, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7214484214782715, + "num_tokens": 246352584.0, + "step": 9688 + }, + { + "epoch": 1.0640237206237646, + "grad_norm": 2.0862550735473633, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7334200143814087, + "num_tokens": 246381227.0, + "step": 9689 + }, + { + "epoch": 1.0641335383263781, + "grad_norm": 2.6572794914245605, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7338101267814636, + "num_tokens": 246400170.0, + "step": 9690 + }, + { + "epoch": 1.064243356028992, + "grad_norm": 2.344231128692627, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7321453094482422, + "num_tokens": 246421767.0, + "step": 9691 + }, + { + "epoch": 1.0643531737316057, + "grad_norm": 2.5287044048309326, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.731956958770752, + "num_tokens": 246441840.0, + "step": 9692 + }, + { + "epoch": 1.0644629914342192, + "grad_norm": 2.4672434329986572, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7052047252655029, + "num_tokens": 246464902.0, + "step": 9693 + }, + { + "epoch": 1.064572809136833, + "grad_norm": 2.1425812244415283, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7209466099739075, + "num_tokens": 246494057.0, + "step": 9694 + }, + { + "epoch": 1.0646826268394465, + "grad_norm": 2.6685845851898193, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.7450546026229858, + "num_tokens": 246512172.0, + "step": 9695 + }, + { + "epoch": 1.0647924445420602, + "grad_norm": 2.506317377090454, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7262943387031555, + "num_tokens": 246532028.0, + "step": 9696 + }, + { + "epoch": 1.0649022622446738, + "grad_norm": 2.197270154953003, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7217109799385071, + "num_tokens": 246557131.0, + "step": 9697 + }, + { + "epoch": 1.0650120799472875, + "grad_norm": 2.6486711502075195, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7149364948272705, + "num_tokens": 246576403.0, + "step": 9698 + }, + { + "epoch": 1.065121897649901, + "grad_norm": 2.62739634513855, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7151738405227661, + "num_tokens": 246595904.0, + "step": 9699 + }, + { + "epoch": 1.0652317153525148, + "grad_norm": 2.22432279586792, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7314308285713196, + "num_tokens": 246622341.0, + "step": 9700 + }, + { + "epoch": 1.0653415330551286, + "grad_norm": 2.096017599105835, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7022230625152588, + "num_tokens": 246651358.0, + "step": 9701 + }, + { + "epoch": 1.065451350757742, + "grad_norm": 1.9924622774124146, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7351894378662109, + "num_tokens": 246681763.0, + "step": 9702 + }, + { + "epoch": 1.0655611684603559, + "grad_norm": 2.1207964420318604, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.744780421257019, + "num_tokens": 246707518.0, + "step": 9703 + }, + { + "epoch": 1.0656709861629694, + "grad_norm": 1.982054352760315, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7340877056121826, + "num_tokens": 246738052.0, + "step": 9704 + }, + { + "epoch": 1.0657808038655832, + "grad_norm": 2.5360701084136963, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7342756986618042, + "num_tokens": 246758021.0, + "step": 9705 + }, + { + "epoch": 1.0658906215681967, + "grad_norm": 2.1526987552642822, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.741069495677948, + "num_tokens": 246784971.0, + "step": 9706 + }, + { + "epoch": 1.0660004392708105, + "grad_norm": 2.7395191192626953, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7368868589401245, + "num_tokens": 246802976.0, + "step": 9707 + }, + { + "epoch": 1.0661102569734242, + "grad_norm": 1.9051904678344727, + "learning_rate": 1e-06, + "loss": 1.0158, + "mean_token_accuracy": 0.6887770891189575, + "num_tokens": 246840664.0, + "step": 9708 + }, + { + "epoch": 1.0662200746760377, + "grad_norm": 1.9934319257736206, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7181490659713745, + "num_tokens": 246871941.0, + "step": 9709 + }, + { + "epoch": 1.0663298923786515, + "grad_norm": 2.4007222652435303, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7154083251953125, + "num_tokens": 246894185.0, + "step": 9710 + }, + { + "epoch": 1.066439710081265, + "grad_norm": 2.6145851612091064, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7149502038955688, + "num_tokens": 246916021.0, + "step": 9711 + }, + { + "epoch": 1.0665495277838788, + "grad_norm": 2.0805304050445557, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7303433418273926, + "num_tokens": 246942747.0, + "step": 9712 + }, + { + "epoch": 1.0666593454864923, + "grad_norm": 2.3281362056732178, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7286244034767151, + "num_tokens": 246967329.0, + "step": 9713 + }, + { + "epoch": 1.066769163189106, + "grad_norm": 2.482754707336426, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7210890054702759, + "num_tokens": 246988957.0, + "step": 9714 + }, + { + "epoch": 1.0668789808917198, + "grad_norm": 2.492067575454712, + "learning_rate": 1e-06, + "loss": 0.97, + "mean_token_accuracy": 0.7020423412322998, + "num_tokens": 247012616.0, + "step": 9715 + }, + { + "epoch": 1.0669887985943334, + "grad_norm": 2.0454905033111572, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7102124691009521, + "num_tokens": 247041871.0, + "step": 9716 + }, + { + "epoch": 1.0670986162969471, + "grad_norm": 2.321850061416626, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7155035734176636, + "num_tokens": 247066756.0, + "step": 9717 + }, + { + "epoch": 1.0672084339995607, + "grad_norm": 1.9823628664016724, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7074308395385742, + "num_tokens": 247096785.0, + "step": 9718 + }, + { + "epoch": 1.0673182517021744, + "grad_norm": 2.396608352661133, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7272348403930664, + "num_tokens": 247118498.0, + "step": 9719 + }, + { + "epoch": 1.067428069404788, + "grad_norm": 2.244793176651001, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7046308517456055, + "num_tokens": 247143218.0, + "step": 9720 + }, + { + "epoch": 1.0675378871074017, + "grad_norm": 2.267958164215088, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7181942462921143, + "num_tokens": 247167986.0, + "step": 9721 + }, + { + "epoch": 1.0676477048100155, + "grad_norm": 2.5343587398529053, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.715354323387146, + "num_tokens": 247190617.0, + "step": 9722 + }, + { + "epoch": 1.067757522512629, + "grad_norm": 2.475172519683838, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7074720859527588, + "num_tokens": 247213461.0, + "step": 9723 + }, + { + "epoch": 1.0678673402152428, + "grad_norm": 2.3456552028656006, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7135007977485657, + "num_tokens": 247237433.0, + "step": 9724 + }, + { + "epoch": 1.0679771579178563, + "grad_norm": 1.9808828830718994, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7031561732292175, + "num_tokens": 247266353.0, + "step": 9725 + }, + { + "epoch": 1.06808697562047, + "grad_norm": 2.3842790126800537, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7304917573928833, + "num_tokens": 247289710.0, + "step": 9726 + }, + { + "epoch": 1.0681967933230836, + "grad_norm": 2.2642691135406494, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7169471979141235, + "num_tokens": 247316898.0, + "step": 9727 + }, + { + "epoch": 1.0683066110256974, + "grad_norm": 2.208986759185791, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7209070920944214, + "num_tokens": 247343298.0, + "step": 9728 + }, + { + "epoch": 1.068416428728311, + "grad_norm": 2.1310746669769287, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7037778496742249, + "num_tokens": 247371103.0, + "step": 9729 + }, + { + "epoch": 1.0685262464309246, + "grad_norm": 2.397177219390869, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7311749458312988, + "num_tokens": 247393817.0, + "step": 9730 + }, + { + "epoch": 1.0686360641335384, + "grad_norm": 2.238462209701538, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7299035787582397, + "num_tokens": 247417709.0, + "step": 9731 + }, + { + "epoch": 1.068745881836152, + "grad_norm": 2.544147491455078, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7197800874710083, + "num_tokens": 247440413.0, + "step": 9732 + }, + { + "epoch": 1.0688556995387657, + "grad_norm": 2.3503925800323486, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.700664758682251, + "num_tokens": 247466314.0, + "step": 9733 + }, + { + "epoch": 1.0689655172413792, + "grad_norm": 2.328299045562744, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.727816641330719, + "num_tokens": 247489955.0, + "step": 9734 + }, + { + "epoch": 1.069075334943993, + "grad_norm": 2.3652684688568115, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7237639427185059, + "num_tokens": 247514438.0, + "step": 9735 + }, + { + "epoch": 1.0691851526466065, + "grad_norm": 2.3745949268341064, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.703237771987915, + "num_tokens": 247538299.0, + "step": 9736 + }, + { + "epoch": 1.0692949703492203, + "grad_norm": 2.1232898235321045, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7087465524673462, + "num_tokens": 247566300.0, + "step": 9737 + }, + { + "epoch": 1.069404788051834, + "grad_norm": 2.667612075805664, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.724034309387207, + "num_tokens": 247585157.0, + "step": 9738 + }, + { + "epoch": 1.0695146057544476, + "grad_norm": 2.536015748977661, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7365791201591492, + "num_tokens": 247605509.0, + "step": 9739 + }, + { + "epoch": 1.0696244234570613, + "grad_norm": 2.4758260250091553, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7282298803329468, + "num_tokens": 247629867.0, + "step": 9740 + }, + { + "epoch": 1.0697342411596749, + "grad_norm": 2.3206324577331543, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7196646928787231, + "num_tokens": 247656159.0, + "step": 9741 + }, + { + "epoch": 1.0698440588622886, + "grad_norm": 2.1458346843719482, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7168523073196411, + "num_tokens": 247684217.0, + "step": 9742 + }, + { + "epoch": 1.0699538765649024, + "grad_norm": 2.0583362579345703, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7232496738433838, + "num_tokens": 247712525.0, + "step": 9743 + }, + { + "epoch": 1.070063694267516, + "grad_norm": 1.9848593473434448, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7080549001693726, + "num_tokens": 247742144.0, + "step": 9744 + }, + { + "epoch": 1.0701735119701297, + "grad_norm": 2.3569045066833496, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7133204936981201, + "num_tokens": 247763789.0, + "step": 9745 + }, + { + "epoch": 1.0702833296727432, + "grad_norm": 2.2395472526550293, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7013188600540161, + "num_tokens": 247789904.0, + "step": 9746 + }, + { + "epoch": 1.070393147375357, + "grad_norm": 2.305997610092163, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7210726141929626, + "num_tokens": 247813464.0, + "step": 9747 + }, + { + "epoch": 1.0705029650779705, + "grad_norm": 2.170133352279663, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7346367835998535, + "num_tokens": 247838412.0, + "step": 9748 + }, + { + "epoch": 1.0706127827805842, + "grad_norm": 2.396001100540161, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7358725070953369, + "num_tokens": 247861261.0, + "step": 9749 + }, + { + "epoch": 1.0707226004831978, + "grad_norm": 2.409498929977417, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.7397223114967346, + "num_tokens": 247881682.0, + "step": 9750 + }, + { + "epoch": 1.0708324181858115, + "grad_norm": 2.368180751800537, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7228372097015381, + "num_tokens": 247908989.0, + "step": 9751 + }, + { + "epoch": 1.0709422358884253, + "grad_norm": 2.7141072750091553, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7055466771125793, + "num_tokens": 247927834.0, + "step": 9752 + }, + { + "epoch": 1.0710520535910388, + "grad_norm": 1.9195094108581543, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7072517275810242, + "num_tokens": 247964299.0, + "step": 9753 + }, + { + "epoch": 1.0711618712936526, + "grad_norm": 2.3378660678863525, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7080645561218262, + "num_tokens": 247986857.0, + "step": 9754 + }, + { + "epoch": 1.0712716889962661, + "grad_norm": 2.263597011566162, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7048442363739014, + "num_tokens": 248012477.0, + "step": 9755 + }, + { + "epoch": 1.0713815066988799, + "grad_norm": 2.483245611190796, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7400050759315491, + "num_tokens": 248032329.0, + "step": 9756 + }, + { + "epoch": 1.0714913244014934, + "grad_norm": 2.0739598274230957, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.705502986907959, + "num_tokens": 248062392.0, + "step": 9757 + }, + { + "epoch": 1.0716011421041072, + "grad_norm": 2.4663162231445312, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7194035053253174, + "num_tokens": 248082963.0, + "step": 9758 + }, + { + "epoch": 1.071710959806721, + "grad_norm": 2.310180425643921, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7258630990982056, + "num_tokens": 248105802.0, + "step": 9759 + }, + { + "epoch": 1.0718207775093345, + "grad_norm": 1.96294367313385, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7102905511856079, + "num_tokens": 248135140.0, + "step": 9760 + }, + { + "epoch": 1.0719305952119482, + "grad_norm": 2.2853970527648926, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7388911247253418, + "num_tokens": 248157654.0, + "step": 9761 + }, + { + "epoch": 1.0720404129145618, + "grad_norm": 2.4286584854125977, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7182705402374268, + "num_tokens": 248181412.0, + "step": 9762 + }, + { + "epoch": 1.0721502306171755, + "grad_norm": 2.1918489933013916, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7013747692108154, + "num_tokens": 248207369.0, + "step": 9763 + }, + { + "epoch": 1.072260048319789, + "grad_norm": 2.419875383377075, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7363501787185669, + "num_tokens": 248227130.0, + "step": 9764 + }, + { + "epoch": 1.0723698660224028, + "grad_norm": 2.52864408493042, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7211585640907288, + "num_tokens": 248247200.0, + "step": 9765 + }, + { + "epoch": 1.0724796837250166, + "grad_norm": 2.0839436054229736, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7138855457305908, + "num_tokens": 248276506.0, + "step": 9766 + }, + { + "epoch": 1.07258950142763, + "grad_norm": 2.2156131267547607, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7012912034988403, + "num_tokens": 248302273.0, + "step": 9767 + }, + { + "epoch": 1.0726993191302439, + "grad_norm": 2.117095708847046, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7022069096565247, + "num_tokens": 248330316.0, + "step": 9768 + }, + { + "epoch": 1.0728091368328574, + "grad_norm": 2.1247365474700928, + "learning_rate": 1e-06, + "loss": 0.9688, + "mean_token_accuracy": 0.7071784734725952, + "num_tokens": 248359868.0, + "step": 9769 + }, + { + "epoch": 1.0729189545354711, + "grad_norm": 2.4337339401245117, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7405747771263123, + "num_tokens": 248379612.0, + "step": 9770 + }, + { + "epoch": 1.0730287722380847, + "grad_norm": 2.123669147491455, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7135339975357056, + "num_tokens": 248408493.0, + "step": 9771 + }, + { + "epoch": 1.0731385899406984, + "grad_norm": 2.5419111251831055, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.6952201128005981, + "num_tokens": 248429925.0, + "step": 9772 + }, + { + "epoch": 1.0732484076433122, + "grad_norm": 2.2248125076293945, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.734512448310852, + "num_tokens": 248454872.0, + "step": 9773 + }, + { + "epoch": 1.0733582253459257, + "grad_norm": 2.27795672416687, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7207751274108887, + "num_tokens": 248479743.0, + "step": 9774 + }, + { + "epoch": 1.0734680430485395, + "grad_norm": 2.2263104915618896, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7284934520721436, + "num_tokens": 248503778.0, + "step": 9775 + }, + { + "epoch": 1.073577860751153, + "grad_norm": 2.20497989654541, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.722264289855957, + "num_tokens": 248529840.0, + "step": 9776 + }, + { + "epoch": 1.0736876784537668, + "grad_norm": 2.3165080547332764, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7358663082122803, + "num_tokens": 248552914.0, + "step": 9777 + }, + { + "epoch": 1.0737974961563803, + "grad_norm": 2.179490327835083, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7437810897827148, + "num_tokens": 248578276.0, + "step": 9778 + }, + { + "epoch": 1.073907313858994, + "grad_norm": 2.4334394931793213, + "learning_rate": 1e-06, + "loss": 0.7639, + "mean_token_accuracy": 0.7564764022827148, + "num_tokens": 248599613.0, + "step": 9779 + }, + { + "epoch": 1.0740171315616078, + "grad_norm": 1.9416946172714233, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7154417037963867, + "num_tokens": 248632621.0, + "step": 9780 + }, + { + "epoch": 1.0741269492642214, + "grad_norm": 2.069915533065796, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7191861867904663, + "num_tokens": 248662225.0, + "step": 9781 + }, + { + "epoch": 1.0742367669668351, + "grad_norm": 2.1005284786224365, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7288527488708496, + "num_tokens": 248689658.0, + "step": 9782 + }, + { + "epoch": 1.0743465846694487, + "grad_norm": 2.5522618293762207, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7239218950271606, + "num_tokens": 248709248.0, + "step": 9783 + }, + { + "epoch": 1.0744564023720624, + "grad_norm": 2.1864774227142334, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6880100965499878, + "num_tokens": 248738736.0, + "step": 9784 + }, + { + "epoch": 1.074566220074676, + "grad_norm": 2.1164064407348633, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7200506925582886, + "num_tokens": 248765520.0, + "step": 9785 + }, + { + "epoch": 1.0746760377772897, + "grad_norm": 2.574660539627075, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7083057165145874, + "num_tokens": 248786393.0, + "step": 9786 + }, + { + "epoch": 1.0747858554799035, + "grad_norm": 2.268134117126465, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.716381847858429, + "num_tokens": 248812605.0, + "step": 9787 + }, + { + "epoch": 1.074895673182517, + "grad_norm": 2.3574562072753906, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.703970730304718, + "num_tokens": 248839030.0, + "step": 9788 + }, + { + "epoch": 1.0750054908851308, + "grad_norm": 2.15138840675354, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.708263635635376, + "num_tokens": 248866890.0, + "step": 9789 + }, + { + "epoch": 1.0751153085877443, + "grad_norm": 2.40824294090271, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7192251086235046, + "num_tokens": 248889299.0, + "step": 9790 + }, + { + "epoch": 1.075225126290358, + "grad_norm": 2.0333988666534424, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7098848819732666, + "num_tokens": 248920640.0, + "step": 9791 + }, + { + "epoch": 1.0753349439929716, + "grad_norm": 2.142965078353882, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7083341479301453, + "num_tokens": 248947637.0, + "step": 9792 + }, + { + "epoch": 1.0754447616955853, + "grad_norm": 2.062286615371704, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7133755087852478, + "num_tokens": 248977144.0, + "step": 9793 + }, + { + "epoch": 1.075554579398199, + "grad_norm": 2.1269257068634033, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7094346880912781, + "num_tokens": 249003186.0, + "step": 9794 + }, + { + "epoch": 1.0756643971008126, + "grad_norm": 2.379258394241333, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7108515501022339, + "num_tokens": 249026923.0, + "step": 9795 + }, + { + "epoch": 1.0757742148034264, + "grad_norm": 2.3359856605529785, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7066789865493774, + "num_tokens": 249052892.0, + "step": 9796 + }, + { + "epoch": 1.07588403250604, + "grad_norm": 2.458414077758789, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7179324626922607, + "num_tokens": 249074575.0, + "step": 9797 + }, + { + "epoch": 1.0759938502086537, + "grad_norm": 2.1285321712493896, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7325239181518555, + "num_tokens": 249102666.0, + "step": 9798 + }, + { + "epoch": 1.0761036679112672, + "grad_norm": 2.356942653656006, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.6962132453918457, + "num_tokens": 249129571.0, + "step": 9799 + }, + { + "epoch": 1.076213485613881, + "grad_norm": 2.589792490005493, + "learning_rate": 1e-06, + "loss": 0.8196, + "mean_token_accuracy": 0.73698890209198, + "num_tokens": 249148065.0, + "step": 9800 + }, + { + "epoch": 1.0763233033164945, + "grad_norm": 2.3902721405029297, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7087188959121704, + "num_tokens": 249171530.0, + "step": 9801 + }, + { + "epoch": 1.0764331210191083, + "grad_norm": 2.264779567718506, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7194018959999084, + "num_tokens": 249197224.0, + "step": 9802 + }, + { + "epoch": 1.076542938721722, + "grad_norm": 2.5220253467559814, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7377862930297852, + "num_tokens": 249217799.0, + "step": 9803 + }, + { + "epoch": 1.0766527564243356, + "grad_norm": 2.1116316318511963, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7077380418777466, + "num_tokens": 249247992.0, + "step": 9804 + }, + { + "epoch": 1.0767625741269493, + "grad_norm": 2.5813865661621094, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7378914952278137, + "num_tokens": 249266889.0, + "step": 9805 + }, + { + "epoch": 1.0768723918295628, + "grad_norm": 2.443284034729004, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7215388417243958, + "num_tokens": 249290448.0, + "step": 9806 + }, + { + "epoch": 1.0769822095321766, + "grad_norm": 2.601592540740967, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.717941164970398, + "num_tokens": 249311622.0, + "step": 9807 + }, + { + "epoch": 1.0770920272347904, + "grad_norm": 2.072878122329712, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7404388189315796, + "num_tokens": 249338275.0, + "step": 9808 + }, + { + "epoch": 1.077201844937404, + "grad_norm": 1.98923659324646, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7261808514595032, + "num_tokens": 249369713.0, + "step": 9809 + }, + { + "epoch": 1.0773116626400177, + "grad_norm": 1.9835636615753174, + "learning_rate": 1e-06, + "loss": 0.832, + "mean_token_accuracy": 0.738111674785614, + "num_tokens": 249398259.0, + "step": 9810 + }, + { + "epoch": 1.0774214803426312, + "grad_norm": 1.8779610395431519, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7204241752624512, + "num_tokens": 249432329.0, + "step": 9811 + }, + { + "epoch": 1.077531298045245, + "grad_norm": 2.144066095352173, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7263475656509399, + "num_tokens": 249458095.0, + "step": 9812 + }, + { + "epoch": 1.0776411157478585, + "grad_norm": 2.1415910720825195, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7321019172668457, + "num_tokens": 249485127.0, + "step": 9813 + }, + { + "epoch": 1.0777509334504722, + "grad_norm": 2.121790647506714, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7099689245223999, + "num_tokens": 249514000.0, + "step": 9814 + }, + { + "epoch": 1.0778607511530858, + "grad_norm": 2.150155544281006, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7274807691574097, + "num_tokens": 249539296.0, + "step": 9815 + }, + { + "epoch": 1.0779705688556995, + "grad_norm": 2.2533364295959473, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7167505025863647, + "num_tokens": 249565697.0, + "step": 9816 + }, + { + "epoch": 1.0780803865583133, + "grad_norm": 2.374668598175049, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7119149565696716, + "num_tokens": 249589478.0, + "step": 9817 + }, + { + "epoch": 1.0781902042609268, + "grad_norm": 2.158921241760254, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7100518941879272, + "num_tokens": 249616245.0, + "step": 9818 + }, + { + "epoch": 1.0783000219635406, + "grad_norm": 2.2939350605010986, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7330774664878845, + "num_tokens": 249640245.0, + "step": 9819 + }, + { + "epoch": 1.0784098396661541, + "grad_norm": 2.144568681716919, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7154903411865234, + "num_tokens": 249666636.0, + "step": 9820 + }, + { + "epoch": 1.0785196573687679, + "grad_norm": 2.352138042449951, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7206209301948547, + "num_tokens": 249690827.0, + "step": 9821 + }, + { + "epoch": 1.0786294750713814, + "grad_norm": 2.1790707111358643, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.697442352771759, + "num_tokens": 249718487.0, + "step": 9822 + }, + { + "epoch": 1.0787392927739952, + "grad_norm": 2.2259521484375, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6903916001319885, + "num_tokens": 249746849.0, + "step": 9823 + }, + { + "epoch": 1.078849110476609, + "grad_norm": 1.8651930093765259, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.6891324520111084, + "num_tokens": 249781362.0, + "step": 9824 + }, + { + "epoch": 1.0789589281792225, + "grad_norm": 2.2982096672058105, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7166297435760498, + "num_tokens": 249804317.0, + "step": 9825 + }, + { + "epoch": 1.0790687458818362, + "grad_norm": 2.4263103008270264, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7303668260574341, + "num_tokens": 249826395.0, + "step": 9826 + }, + { + "epoch": 1.0791785635844497, + "grad_norm": 2.6088948249816895, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.727665901184082, + "num_tokens": 249846831.0, + "step": 9827 + }, + { + "epoch": 1.0792883812870635, + "grad_norm": 1.9817126989364624, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7113727927207947, + "num_tokens": 249878991.0, + "step": 9828 + }, + { + "epoch": 1.079398198989677, + "grad_norm": 2.321716070175171, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7383378744125366, + "num_tokens": 249904737.0, + "step": 9829 + }, + { + "epoch": 1.0795080166922908, + "grad_norm": 2.090686082839966, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.7013344764709473, + "num_tokens": 249934585.0, + "step": 9830 + }, + { + "epoch": 1.0796178343949046, + "grad_norm": 2.593614339828491, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7230815887451172, + "num_tokens": 249955958.0, + "step": 9831 + }, + { + "epoch": 1.079727652097518, + "grad_norm": 2.4761099815368652, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.725226104259491, + "num_tokens": 249977837.0, + "step": 9832 + }, + { + "epoch": 1.0798374698001318, + "grad_norm": 2.4063222408294678, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7245274782180786, + "num_tokens": 250002852.0, + "step": 9833 + }, + { + "epoch": 1.0799472875027454, + "grad_norm": 2.1361491680145264, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7188270688056946, + "num_tokens": 250030966.0, + "step": 9834 + }, + { + "epoch": 1.0800571052053591, + "grad_norm": 2.202686309814453, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7242110967636108, + "num_tokens": 250058101.0, + "step": 9835 + }, + { + "epoch": 1.0801669229079727, + "grad_norm": 2.1478946208953857, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7099236249923706, + "num_tokens": 250086973.0, + "step": 9836 + }, + { + "epoch": 1.0802767406105864, + "grad_norm": 2.424572706222534, + "learning_rate": 1e-06, + "loss": 0.8013, + "mean_token_accuracy": 0.7358700633049011, + "num_tokens": 250107200.0, + "step": 9837 + }, + { + "epoch": 1.0803865583132002, + "grad_norm": 2.5014092922210693, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7319558262825012, + "num_tokens": 250127671.0, + "step": 9838 + }, + { + "epoch": 1.0804963760158137, + "grad_norm": 2.183086395263672, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6989691257476807, + "num_tokens": 250157856.0, + "step": 9839 + }, + { + "epoch": 1.0806061937184275, + "grad_norm": 2.1223044395446777, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7117159962654114, + "num_tokens": 250186838.0, + "step": 9840 + }, + { + "epoch": 1.080716011421041, + "grad_norm": 2.1665003299713135, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.743525505065918, + "num_tokens": 250211305.0, + "step": 9841 + }, + { + "epoch": 1.0808258291236548, + "grad_norm": 2.2579126358032227, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.731942892074585, + "num_tokens": 250237499.0, + "step": 9842 + }, + { + "epoch": 1.0809356468262683, + "grad_norm": 2.113286256790161, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.728839635848999, + "num_tokens": 250266122.0, + "step": 9843 + }, + { + "epoch": 1.081045464528882, + "grad_norm": 2.3597161769866943, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7135176658630371, + "num_tokens": 250289568.0, + "step": 9844 + }, + { + "epoch": 1.0811552822314958, + "grad_norm": 2.2776358127593994, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.724261462688446, + "num_tokens": 250315581.0, + "step": 9845 + }, + { + "epoch": 1.0812650999341094, + "grad_norm": 2.614004611968994, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7377941012382507, + "num_tokens": 250335826.0, + "step": 9846 + }, + { + "epoch": 1.081374917636723, + "grad_norm": 2.358018159866333, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7356719970703125, + "num_tokens": 250358657.0, + "step": 9847 + }, + { + "epoch": 1.0814847353393366, + "grad_norm": 2.2986738681793213, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7163910865783691, + "num_tokens": 250384262.0, + "step": 9848 + }, + { + "epoch": 1.0815945530419504, + "grad_norm": 2.3667922019958496, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7132388353347778, + "num_tokens": 250407760.0, + "step": 9849 + }, + { + "epoch": 1.081704370744564, + "grad_norm": 2.0023233890533447, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7030198574066162, + "num_tokens": 250438611.0, + "step": 9850 + }, + { + "epoch": 1.0818141884471777, + "grad_norm": 2.494354009628296, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7343109250068665, + "num_tokens": 250459594.0, + "step": 9851 + }, + { + "epoch": 1.0819240061497912, + "grad_norm": 2.019944190979004, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7053682804107666, + "num_tokens": 250489774.0, + "step": 9852 + }, + { + "epoch": 1.082033823852405, + "grad_norm": 2.2422263622283936, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7453181147575378, + "num_tokens": 250518342.0, + "step": 9853 + }, + { + "epoch": 1.0821436415550187, + "grad_norm": 2.8281896114349365, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7267163991928101, + "num_tokens": 250536331.0, + "step": 9854 + }, + { + "epoch": 1.0822534592576323, + "grad_norm": 2.1868057250976562, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7178136110305786, + "num_tokens": 250563510.0, + "step": 9855 + }, + { + "epoch": 1.082363276960246, + "grad_norm": 1.9219070672988892, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7120383977890015, + "num_tokens": 250595708.0, + "step": 9856 + }, + { + "epoch": 1.0824730946628596, + "grad_norm": 2.2206034660339355, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7187577486038208, + "num_tokens": 250622081.0, + "step": 9857 + }, + { + "epoch": 1.0825829123654733, + "grad_norm": 2.060910701751709, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6904045343399048, + "num_tokens": 250654803.0, + "step": 9858 + }, + { + "epoch": 1.082692730068087, + "grad_norm": 2.4640066623687744, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7135642170906067, + "num_tokens": 250676276.0, + "step": 9859 + }, + { + "epoch": 1.0828025477707006, + "grad_norm": 1.9441033601760864, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7382710576057434, + "num_tokens": 250706956.0, + "step": 9860 + }, + { + "epoch": 1.0829123654733144, + "grad_norm": 2.3904712200164795, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.70472252368927, + "num_tokens": 250732848.0, + "step": 9861 + }, + { + "epoch": 1.083022183175928, + "grad_norm": 2.0946738719940186, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7192142009735107, + "num_tokens": 250759966.0, + "step": 9862 + }, + { + "epoch": 1.0831320008785417, + "grad_norm": 2.4173974990844727, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7327945232391357, + "num_tokens": 250783630.0, + "step": 9863 + }, + { + "epoch": 1.0832418185811552, + "grad_norm": 2.3770036697387695, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7357651591300964, + "num_tokens": 250807730.0, + "step": 9864 + }, + { + "epoch": 1.083351636283769, + "grad_norm": 2.3780834674835205, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7381538152694702, + "num_tokens": 250829910.0, + "step": 9865 + }, + { + "epoch": 1.0834614539863825, + "grad_norm": 2.4634721279144287, + "learning_rate": 1e-06, + "loss": 0.739, + "mean_token_accuracy": 0.7576518058776855, + "num_tokens": 250849436.0, + "step": 9866 + }, + { + "epoch": 1.0835712716889963, + "grad_norm": 2.2324516773223877, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7355566024780273, + "num_tokens": 250873310.0, + "step": 9867 + }, + { + "epoch": 1.08368108939161, + "grad_norm": 2.5175626277923584, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7273498773574829, + "num_tokens": 250893548.0, + "step": 9868 + }, + { + "epoch": 1.0837909070942235, + "grad_norm": 2.122124671936035, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7283969521522522, + "num_tokens": 250921817.0, + "step": 9869 + }, + { + "epoch": 1.0839007247968373, + "grad_norm": 2.2232978343963623, + "learning_rate": 1e-06, + "loss": 0.9979, + "mean_token_accuracy": 0.7063578367233276, + "num_tokens": 250947110.0, + "step": 9870 + }, + { + "epoch": 1.0840105424994508, + "grad_norm": 1.9907126426696777, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7411020994186401, + "num_tokens": 250974361.0, + "step": 9871 + }, + { + "epoch": 1.0841203602020646, + "grad_norm": 2.6010420322418213, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7041944265365601, + "num_tokens": 250994430.0, + "step": 9872 + }, + { + "epoch": 1.0842301779046783, + "grad_norm": 2.2992677688598633, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6862545609474182, + "num_tokens": 251020702.0, + "step": 9873 + }, + { + "epoch": 1.0843399956072919, + "grad_norm": 2.2672061920166016, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7145905494689941, + "num_tokens": 251047515.0, + "step": 9874 + }, + { + "epoch": 1.0844498133099056, + "grad_norm": 2.239248752593994, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6987159848213196, + "num_tokens": 251073244.0, + "step": 9875 + }, + { + "epoch": 1.0845596310125192, + "grad_norm": 2.2834744453430176, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7041049003601074, + "num_tokens": 251099660.0, + "step": 9876 + }, + { + "epoch": 1.084669448715133, + "grad_norm": 2.0452394485473633, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.699825644493103, + "num_tokens": 251131471.0, + "step": 9877 + }, + { + "epoch": 1.0847792664177465, + "grad_norm": 2.2649035453796387, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6897516250610352, + "num_tokens": 251158344.0, + "step": 9878 + }, + { + "epoch": 1.0848890841203602, + "grad_norm": 2.310917377471924, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.73678058385849, + "num_tokens": 251182585.0, + "step": 9879 + }, + { + "epoch": 1.0849989018229738, + "grad_norm": 2.0941267013549805, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7281187176704407, + "num_tokens": 251210477.0, + "step": 9880 + }, + { + "epoch": 1.0851087195255875, + "grad_norm": 2.421583890914917, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7150744199752808, + "num_tokens": 251232556.0, + "step": 9881 + }, + { + "epoch": 1.0852185372282013, + "grad_norm": 1.8785762786865234, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7052732706069946, + "num_tokens": 251270229.0, + "step": 9882 + }, + { + "epoch": 1.0853283549308148, + "grad_norm": 2.560509443283081, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7214521169662476, + "num_tokens": 251292280.0, + "step": 9883 + }, + { + "epoch": 1.0854381726334286, + "grad_norm": 2.4251670837402344, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7193036079406738, + "num_tokens": 251316296.0, + "step": 9884 + }, + { + "epoch": 1.085547990336042, + "grad_norm": 2.089216470718384, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.6985336542129517, + "num_tokens": 251346155.0, + "step": 9885 + }, + { + "epoch": 1.0856578080386559, + "grad_norm": 2.208773612976074, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.714093804359436, + "num_tokens": 251372835.0, + "step": 9886 + }, + { + "epoch": 1.0857676257412694, + "grad_norm": 2.364227056503296, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.7438592910766602, + "num_tokens": 251395630.0, + "step": 9887 + }, + { + "epoch": 1.0858774434438832, + "grad_norm": 2.294795036315918, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7108129858970642, + "num_tokens": 251420470.0, + "step": 9888 + }, + { + "epoch": 1.085987261146497, + "grad_norm": 2.5214579105377197, + "learning_rate": 1e-06, + "loss": 0.8042, + "mean_token_accuracy": 0.7486928701400757, + "num_tokens": 251440925.0, + "step": 9889 + }, + { + "epoch": 1.0860970788491104, + "grad_norm": 2.2162954807281494, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.727769672870636, + "num_tokens": 251466024.0, + "step": 9890 + }, + { + "epoch": 1.0862068965517242, + "grad_norm": 2.3304483890533447, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7025420665740967, + "num_tokens": 251489458.0, + "step": 9891 + }, + { + "epoch": 1.0863167142543377, + "grad_norm": 2.061469078063965, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.6989867687225342, + "num_tokens": 251519889.0, + "step": 9892 + }, + { + "epoch": 1.0864265319569515, + "grad_norm": 2.186098575592041, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7066769599914551, + "num_tokens": 251547257.0, + "step": 9893 + }, + { + "epoch": 1.086536349659565, + "grad_norm": 2.281090259552002, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7282699346542358, + "num_tokens": 251571424.0, + "step": 9894 + }, + { + "epoch": 1.0866461673621788, + "grad_norm": 3.046679735183716, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7400996088981628, + "num_tokens": 251587720.0, + "step": 9895 + }, + { + "epoch": 1.0867559850647925, + "grad_norm": 2.162292957305908, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.72688889503479, + "num_tokens": 251614350.0, + "step": 9896 + }, + { + "epoch": 1.086865802767406, + "grad_norm": 2.257274866104126, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7067482471466064, + "num_tokens": 251638558.0, + "step": 9897 + }, + { + "epoch": 1.0869756204700198, + "grad_norm": 2.1083245277404785, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7239271402359009, + "num_tokens": 251663957.0, + "step": 9898 + }, + { + "epoch": 1.0870854381726334, + "grad_norm": 2.054189920425415, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.706757664680481, + "num_tokens": 251695545.0, + "step": 9899 + }, + { + "epoch": 1.0871952558752471, + "grad_norm": 2.4926390647888184, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7361323833465576, + "num_tokens": 251717392.0, + "step": 9900 + }, + { + "epoch": 1.0873050735778607, + "grad_norm": 2.178873062133789, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7116639614105225, + "num_tokens": 251746557.0, + "step": 9901 + }, + { + "epoch": 1.0874148912804744, + "grad_norm": 2.2834584712982178, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7201755046844482, + "num_tokens": 251772015.0, + "step": 9902 + }, + { + "epoch": 1.0875247089830882, + "grad_norm": 2.2419402599334717, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.728635311126709, + "num_tokens": 251797845.0, + "step": 9903 + }, + { + "epoch": 1.0876345266857017, + "grad_norm": 2.118117094039917, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7187813520431519, + "num_tokens": 251827411.0, + "step": 9904 + }, + { + "epoch": 1.0877443443883155, + "grad_norm": 2.151715040206909, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7174420356750488, + "num_tokens": 251854637.0, + "step": 9905 + }, + { + "epoch": 1.087854162090929, + "grad_norm": 2.2496936321258545, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7265912294387817, + "num_tokens": 251880478.0, + "step": 9906 + }, + { + "epoch": 1.0879639797935428, + "grad_norm": 2.315880537033081, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7254399657249451, + "num_tokens": 251904799.0, + "step": 9907 + }, + { + "epoch": 1.0880737974961563, + "grad_norm": 1.9631242752075195, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7260432243347168, + "num_tokens": 251937532.0, + "step": 9908 + }, + { + "epoch": 1.08818361519877, + "grad_norm": 2.095254898071289, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.736362874507904, + "num_tokens": 251966191.0, + "step": 9909 + }, + { + "epoch": 1.0882934329013838, + "grad_norm": 2.2351043224334717, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7040863633155823, + "num_tokens": 251993241.0, + "step": 9910 + }, + { + "epoch": 1.0884032506039973, + "grad_norm": 2.1485342979431152, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7302678823471069, + "num_tokens": 252020264.0, + "step": 9911 + }, + { + "epoch": 1.088513068306611, + "grad_norm": 2.4249422550201416, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7305169105529785, + "num_tokens": 252041817.0, + "step": 9912 + }, + { + "epoch": 1.0886228860092246, + "grad_norm": 2.3164169788360596, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7228111028671265, + "num_tokens": 252064421.0, + "step": 9913 + }, + { + "epoch": 1.0887327037118384, + "grad_norm": 2.109536647796631, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7174408435821533, + "num_tokens": 252093688.0, + "step": 9914 + }, + { + "epoch": 1.088842521414452, + "grad_norm": 2.2311177253723145, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7155969142913818, + "num_tokens": 252120296.0, + "step": 9915 + }, + { + "epoch": 1.0889523391170657, + "grad_norm": 2.115565299987793, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7115271687507629, + "num_tokens": 252151658.0, + "step": 9916 + }, + { + "epoch": 1.0890621568196792, + "grad_norm": 2.1823248863220215, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7222950458526611, + "num_tokens": 252177427.0, + "step": 9917 + }, + { + "epoch": 1.089171974522293, + "grad_norm": 2.5661864280700684, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7189609408378601, + "num_tokens": 252197919.0, + "step": 9918 + }, + { + "epoch": 1.0892817922249067, + "grad_norm": 2.296185255050659, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.6983532309532166, + "num_tokens": 252224362.0, + "step": 9919 + }, + { + "epoch": 1.0893916099275203, + "grad_norm": 2.2489826679229736, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7100546360015869, + "num_tokens": 252250024.0, + "step": 9920 + }, + { + "epoch": 1.089501427630134, + "grad_norm": 2.1791911125183105, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7111563086509705, + "num_tokens": 252276878.0, + "step": 9921 + }, + { + "epoch": 1.0896112453327476, + "grad_norm": 2.0848257541656494, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7088068127632141, + "num_tokens": 252306186.0, + "step": 9922 + }, + { + "epoch": 1.0897210630353613, + "grad_norm": 2.2890124320983887, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7415006160736084, + "num_tokens": 252329668.0, + "step": 9923 + }, + { + "epoch": 1.089830880737975, + "grad_norm": 2.1976566314697266, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.718909502029419, + "num_tokens": 252355298.0, + "step": 9924 + }, + { + "epoch": 1.0899406984405886, + "grad_norm": 2.4517767429351807, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7124305367469788, + "num_tokens": 252379063.0, + "step": 9925 + }, + { + "epoch": 1.0900505161432024, + "grad_norm": 2.0888755321502686, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7104254961013794, + "num_tokens": 252407305.0, + "step": 9926 + }, + { + "epoch": 1.090160333845816, + "grad_norm": 2.1440927982330322, + "learning_rate": 1e-06, + "loss": 1.0409, + "mean_token_accuracy": 0.6781426668167114, + "num_tokens": 252438764.0, + "step": 9927 + }, + { + "epoch": 1.0902701515484297, + "grad_norm": 2.1794509887695312, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.6983273029327393, + "num_tokens": 252465835.0, + "step": 9928 + }, + { + "epoch": 1.0903799692510432, + "grad_norm": 2.385507106781006, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7370600700378418, + "num_tokens": 252487489.0, + "step": 9929 + }, + { + "epoch": 1.090489786953657, + "grad_norm": 2.231081485748291, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7158265709877014, + "num_tokens": 252515647.0, + "step": 9930 + }, + { + "epoch": 1.0905996046562705, + "grad_norm": 2.0882015228271484, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7202101945877075, + "num_tokens": 252542082.0, + "step": 9931 + }, + { + "epoch": 1.0907094223588842, + "grad_norm": 2.2865936756134033, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7032558917999268, + "num_tokens": 252569046.0, + "step": 9932 + }, + { + "epoch": 1.090819240061498, + "grad_norm": 2.191425085067749, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7086372971534729, + "num_tokens": 252597250.0, + "step": 9933 + }, + { + "epoch": 1.0909290577641115, + "grad_norm": 2.1059529781341553, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7381621599197388, + "num_tokens": 252622796.0, + "step": 9934 + }, + { + "epoch": 1.0910388754667253, + "grad_norm": 2.042755365371704, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6948392987251282, + "num_tokens": 252653162.0, + "step": 9935 + }, + { + "epoch": 1.0911486931693388, + "grad_norm": 2.392133951187134, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7060205936431885, + "num_tokens": 252676695.0, + "step": 9936 + }, + { + "epoch": 1.0912585108719526, + "grad_norm": 2.2465591430664062, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7287701964378357, + "num_tokens": 252703031.0, + "step": 9937 + }, + { + "epoch": 1.0913683285745663, + "grad_norm": 2.280803918838501, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7198957204818726, + "num_tokens": 252728444.0, + "step": 9938 + }, + { + "epoch": 1.0914781462771799, + "grad_norm": 2.2873928546905518, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.713737964630127, + "num_tokens": 252754001.0, + "step": 9939 + }, + { + "epoch": 1.0915879639797936, + "grad_norm": 2.329864025115967, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.6913053393363953, + "num_tokens": 252779147.0, + "step": 9940 + }, + { + "epoch": 1.0916977816824072, + "grad_norm": 2.563275098800659, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7158864736557007, + "num_tokens": 252800886.0, + "step": 9941 + }, + { + "epoch": 1.091807599385021, + "grad_norm": 2.0708167552948, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7049906253814697, + "num_tokens": 252828965.0, + "step": 9942 + }, + { + "epoch": 1.0919174170876345, + "grad_norm": 2.283194065093994, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7209200859069824, + "num_tokens": 252853974.0, + "step": 9943 + }, + { + "epoch": 1.0920272347902482, + "grad_norm": 2.245588541030884, + "learning_rate": 1e-06, + "loss": 1.0004, + "mean_token_accuracy": 0.6953381299972534, + "num_tokens": 252881770.0, + "step": 9944 + }, + { + "epoch": 1.0921370524928617, + "grad_norm": 2.2664637565612793, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7148057222366333, + "num_tokens": 252905916.0, + "step": 9945 + }, + { + "epoch": 1.0922468701954755, + "grad_norm": 2.2247445583343506, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7236881256103516, + "num_tokens": 252932465.0, + "step": 9946 + }, + { + "epoch": 1.0923566878980893, + "grad_norm": 1.9667205810546875, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7050875425338745, + "num_tokens": 252964955.0, + "step": 9947 + }, + { + "epoch": 1.0924665056007028, + "grad_norm": 2.1986947059631348, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7151644825935364, + "num_tokens": 252992340.0, + "step": 9948 + }, + { + "epoch": 1.0925763233033166, + "grad_norm": 2.204327344894409, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7319321036338806, + "num_tokens": 253017818.0, + "step": 9949 + }, + { + "epoch": 1.09268614100593, + "grad_norm": 2.484755754470825, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7221536040306091, + "num_tokens": 253038699.0, + "step": 9950 + }, + { + "epoch": 1.0927959587085438, + "grad_norm": 2.448971748352051, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7243886590003967, + "num_tokens": 253059778.0, + "step": 9951 + }, + { + "epoch": 1.0929057764111574, + "grad_norm": 2.270305871963501, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7252092361450195, + "num_tokens": 253083543.0, + "step": 9952 + }, + { + "epoch": 1.0930155941137711, + "grad_norm": 2.361527442932129, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7182985544204712, + "num_tokens": 253107045.0, + "step": 9953 + }, + { + "epoch": 1.093125411816385, + "grad_norm": 2.32757306098938, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.711651086807251, + "num_tokens": 253130944.0, + "step": 9954 + }, + { + "epoch": 1.0932352295189984, + "grad_norm": 2.731273651123047, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.743313193321228, + "num_tokens": 253148387.0, + "step": 9955 + }, + { + "epoch": 1.0933450472216122, + "grad_norm": 2.677119016647339, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7098722457885742, + "num_tokens": 253169867.0, + "step": 9956 + }, + { + "epoch": 1.0934548649242257, + "grad_norm": 2.189607858657837, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.707068145275116, + "num_tokens": 253198502.0, + "step": 9957 + }, + { + "epoch": 1.0935646826268395, + "grad_norm": 2.1833064556121826, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7159473896026611, + "num_tokens": 253224966.0, + "step": 9958 + }, + { + "epoch": 1.093674500329453, + "grad_norm": 2.050137519836426, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7214251756668091, + "num_tokens": 253256462.0, + "step": 9959 + }, + { + "epoch": 1.0937843180320668, + "grad_norm": 2.4253671169281006, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7284830808639526, + "num_tokens": 253278906.0, + "step": 9960 + }, + { + "epoch": 1.0938941357346805, + "grad_norm": 2.1614270210266113, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7133122086524963, + "num_tokens": 253307364.0, + "step": 9961 + }, + { + "epoch": 1.094003953437294, + "grad_norm": 2.4222633838653564, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7271593809127808, + "num_tokens": 253330266.0, + "step": 9962 + }, + { + "epoch": 1.0941137711399078, + "grad_norm": 2.240885019302368, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7253965735435486, + "num_tokens": 253354961.0, + "step": 9963 + }, + { + "epoch": 1.0942235888425214, + "grad_norm": 2.4434289932250977, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7216038107872009, + "num_tokens": 253377850.0, + "step": 9964 + }, + { + "epoch": 1.0943334065451351, + "grad_norm": 2.1756646633148193, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7059744596481323, + "num_tokens": 253407007.0, + "step": 9965 + }, + { + "epoch": 1.0944432242477486, + "grad_norm": 2.5302822589874268, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.717069685459137, + "num_tokens": 253428668.0, + "step": 9966 + }, + { + "epoch": 1.0945530419503624, + "grad_norm": 2.180811882019043, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7139895558357239, + "num_tokens": 253455223.0, + "step": 9967 + }, + { + "epoch": 1.0946628596529762, + "grad_norm": 2.015378475189209, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7141110897064209, + "num_tokens": 253484857.0, + "step": 9968 + }, + { + "epoch": 1.0947726773555897, + "grad_norm": 2.3142952919006348, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7245562076568604, + "num_tokens": 253509319.0, + "step": 9969 + }, + { + "epoch": 1.0948824950582035, + "grad_norm": 2.2854011058807373, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7222371697425842, + "num_tokens": 253532037.0, + "step": 9970 + }, + { + "epoch": 1.094992312760817, + "grad_norm": 2.3447790145874023, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7234780788421631, + "num_tokens": 253557931.0, + "step": 9971 + }, + { + "epoch": 1.0951021304634307, + "grad_norm": 2.578408718109131, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7304838299751282, + "num_tokens": 253577951.0, + "step": 9972 + }, + { + "epoch": 1.0952119481660443, + "grad_norm": 2.347336769104004, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7229717373847961, + "num_tokens": 253601909.0, + "step": 9973 + }, + { + "epoch": 1.095321765868658, + "grad_norm": 2.252183198928833, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7159122824668884, + "num_tokens": 253627127.0, + "step": 9974 + }, + { + "epoch": 1.0954315835712718, + "grad_norm": 2.3220369815826416, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.6977004408836365, + "num_tokens": 253654825.0, + "step": 9975 + }, + { + "epoch": 1.0955414012738853, + "grad_norm": 2.180819272994995, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7167364358901978, + "num_tokens": 253683618.0, + "step": 9976 + }, + { + "epoch": 1.095651218976499, + "grad_norm": 2.2463746070861816, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.714977502822876, + "num_tokens": 253710798.0, + "step": 9977 + }, + { + "epoch": 1.0957610366791126, + "grad_norm": 2.4691262245178223, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7201821208000183, + "num_tokens": 253733272.0, + "step": 9978 + }, + { + "epoch": 1.0958708543817264, + "grad_norm": 2.25335955619812, + "learning_rate": 1e-06, + "loss": 0.8125, + "mean_token_accuracy": 0.7421196103096008, + "num_tokens": 253758747.0, + "step": 9979 + }, + { + "epoch": 1.09598067208434, + "grad_norm": 2.1451854705810547, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7027393579483032, + "num_tokens": 253787771.0, + "step": 9980 + }, + { + "epoch": 1.0960904897869537, + "grad_norm": 2.0275075435638428, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7070518732070923, + "num_tokens": 253818231.0, + "step": 9981 + }, + { + "epoch": 1.0962003074895672, + "grad_norm": 2.3073678016662598, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7290316224098206, + "num_tokens": 253841749.0, + "step": 9982 + }, + { + "epoch": 1.096310125192181, + "grad_norm": 2.289393663406372, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7271893620491028, + "num_tokens": 253864778.0, + "step": 9983 + }, + { + "epoch": 1.0964199428947947, + "grad_norm": 2.035907745361328, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7113988399505615, + "num_tokens": 253896805.0, + "step": 9984 + }, + { + "epoch": 1.0965297605974083, + "grad_norm": 2.1005706787109375, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7056781649589539, + "num_tokens": 253928345.0, + "step": 9985 + }, + { + "epoch": 1.096639578300022, + "grad_norm": 2.101952075958252, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.72206711769104, + "num_tokens": 253957467.0, + "step": 9986 + }, + { + "epoch": 1.0967493960026355, + "grad_norm": 2.438223123550415, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7062812447547913, + "num_tokens": 253980959.0, + "step": 9987 + }, + { + "epoch": 1.0968592137052493, + "grad_norm": 1.9152989387512207, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7210549712181091, + "num_tokens": 254011679.0, + "step": 9988 + }, + { + "epoch": 1.096969031407863, + "grad_norm": 2.431424856185913, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7188711166381836, + "num_tokens": 254034718.0, + "step": 9989 + }, + { + "epoch": 1.0970788491104766, + "grad_norm": 2.4704675674438477, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.70643150806427, + "num_tokens": 254058828.0, + "step": 9990 + }, + { + "epoch": 1.0971886668130904, + "grad_norm": 2.4167492389678955, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7087270617485046, + "num_tokens": 254084341.0, + "step": 9991 + }, + { + "epoch": 1.0972984845157039, + "grad_norm": 2.107576847076416, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.7480042576789856, + "num_tokens": 254112562.0, + "step": 9992 + }, + { + "epoch": 1.0974083022183176, + "grad_norm": 2.7358391284942627, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7144819498062134, + "num_tokens": 254132483.0, + "step": 9993 + }, + { + "epoch": 1.0975181199209312, + "grad_norm": 2.0825586318969727, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7202579975128174, + "num_tokens": 254159681.0, + "step": 9994 + }, + { + "epoch": 1.097627937623545, + "grad_norm": 1.925986409187317, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7286216616630554, + "num_tokens": 254191514.0, + "step": 9995 + }, + { + "epoch": 1.0977377553261585, + "grad_norm": 2.5566680431365967, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7373438477516174, + "num_tokens": 254210740.0, + "step": 9996 + }, + { + "epoch": 1.0978475730287722, + "grad_norm": 2.6984565258026123, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7224537134170532, + "num_tokens": 254231812.0, + "step": 9997 + }, + { + "epoch": 1.097957390731386, + "grad_norm": 2.2107157707214355, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7160211801528931, + "num_tokens": 254258150.0, + "step": 9998 + }, + { + "epoch": 1.0980672084339995, + "grad_norm": 2.086766242980957, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7177691459655762, + "num_tokens": 254287781.0, + "step": 9999 + }, + { + "epoch": 1.0981770261366133, + "grad_norm": 2.1430468559265137, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7300920486450195, + "num_tokens": 254314843.0, + "step": 10000 + }, + { + "epoch": 1.0982868438392268, + "grad_norm": 2.4448370933532715, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.722355306148529, + "num_tokens": 254337026.0, + "step": 10001 + }, + { + "epoch": 1.0983966615418406, + "grad_norm": 2.3864400386810303, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.694507360458374, + "num_tokens": 254362825.0, + "step": 10002 + }, + { + "epoch": 1.098506479244454, + "grad_norm": 1.744097352027893, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.723069429397583, + "num_tokens": 254402118.0, + "step": 10003 + }, + { + "epoch": 1.0986162969470679, + "grad_norm": 2.5354950428009033, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7219640016555786, + "num_tokens": 254422801.0, + "step": 10004 + }, + { + "epoch": 1.0987261146496816, + "grad_norm": 2.1420538425445557, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7385532259941101, + "num_tokens": 254448264.0, + "step": 10005 + }, + { + "epoch": 1.0988359323522952, + "grad_norm": 2.5526626110076904, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7414642572402954, + "num_tokens": 254467169.0, + "step": 10006 + }, + { + "epoch": 1.098945750054909, + "grad_norm": 2.0645408630371094, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7149914503097534, + "num_tokens": 254495725.0, + "step": 10007 + }, + { + "epoch": 1.0990555677575224, + "grad_norm": 2.3460934162139893, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7356834411621094, + "num_tokens": 254518656.0, + "step": 10008 + }, + { + "epoch": 1.0991653854601362, + "grad_norm": 2.4004547595977783, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7115187048912048, + "num_tokens": 254541249.0, + "step": 10009 + }, + { + "epoch": 1.0992752031627497, + "grad_norm": 2.333888053894043, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7053279280662537, + "num_tokens": 254568653.0, + "step": 10010 + }, + { + "epoch": 1.0993850208653635, + "grad_norm": 2.1659514904022217, + "learning_rate": 1e-06, + "loss": 1.0279, + "mean_token_accuracy": 0.6843499541282654, + "num_tokens": 254600259.0, + "step": 10011 + }, + { + "epoch": 1.0994948385679773, + "grad_norm": 2.4727723598480225, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7073017954826355, + "num_tokens": 254622585.0, + "step": 10012 + }, + { + "epoch": 1.0996046562705908, + "grad_norm": 1.9855802059173584, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7163009643554688, + "num_tokens": 254651932.0, + "step": 10013 + }, + { + "epoch": 1.0997144739732045, + "grad_norm": 2.247162342071533, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.713495135307312, + "num_tokens": 254677440.0, + "step": 10014 + }, + { + "epoch": 1.099824291675818, + "grad_norm": 2.739607572555542, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7408035397529602, + "num_tokens": 254695655.0, + "step": 10015 + }, + { + "epoch": 1.0999341093784318, + "grad_norm": 2.3279616832733154, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7023042440414429, + "num_tokens": 254723085.0, + "step": 10016 + }, + { + "epoch": 1.1000439270810454, + "grad_norm": 2.3737387657165527, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7009842991828918, + "num_tokens": 254748439.0, + "step": 10017 + }, + { + "epoch": 1.1001537447836591, + "grad_norm": 2.2178683280944824, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7224754691123962, + "num_tokens": 254774848.0, + "step": 10018 + }, + { + "epoch": 1.1002635624862729, + "grad_norm": 2.4388909339904785, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7290308475494385, + "num_tokens": 254796555.0, + "step": 10019 + }, + { + "epoch": 1.1003733801888864, + "grad_norm": 2.2472848892211914, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.720905065536499, + "num_tokens": 254822524.0, + "step": 10020 + }, + { + "epoch": 1.1004831978915002, + "grad_norm": 2.4545257091522217, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7113025188446045, + "num_tokens": 254847293.0, + "step": 10021 + }, + { + "epoch": 1.1005930155941137, + "grad_norm": 2.035137414932251, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7240984439849854, + "num_tokens": 254876379.0, + "step": 10022 + }, + { + "epoch": 1.1007028332967275, + "grad_norm": 2.181140422821045, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7007520198822021, + "num_tokens": 254902310.0, + "step": 10023 + }, + { + "epoch": 1.100812650999341, + "grad_norm": 2.3423306941986084, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7174777984619141, + "num_tokens": 254925107.0, + "step": 10024 + }, + { + "epoch": 1.1009224687019548, + "grad_norm": 2.479525327682495, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7298613786697388, + "num_tokens": 254944978.0, + "step": 10025 + }, + { + "epoch": 1.1010322864045685, + "grad_norm": 2.5560078620910645, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7160302996635437, + "num_tokens": 254965670.0, + "step": 10026 + }, + { + "epoch": 1.101142104107182, + "grad_norm": 2.351931095123291, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7236407995223999, + "num_tokens": 254989450.0, + "step": 10027 + }, + { + "epoch": 1.1012519218097958, + "grad_norm": 2.651169776916504, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7130311727523804, + "num_tokens": 255009650.0, + "step": 10028 + }, + { + "epoch": 1.1013617395124093, + "grad_norm": 2.2097980976104736, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7204622030258179, + "num_tokens": 255035163.0, + "step": 10029 + }, + { + "epoch": 1.101471557215023, + "grad_norm": 1.9324103593826294, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.7007943391799927, + "num_tokens": 255067953.0, + "step": 10030 + }, + { + "epoch": 1.1015813749176366, + "grad_norm": 1.8870519399642944, + "learning_rate": 1e-06, + "loss": 1.1152, + "mean_token_accuracy": 0.663150429725647, + "num_tokens": 255103333.0, + "step": 10031 + }, + { + "epoch": 1.1016911926202504, + "grad_norm": 2.1649622917175293, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7177155017852783, + "num_tokens": 255129489.0, + "step": 10032 + }, + { + "epoch": 1.1018010103228641, + "grad_norm": 2.4207019805908203, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7170100212097168, + "num_tokens": 255151465.0, + "step": 10033 + }, + { + "epoch": 1.1019108280254777, + "grad_norm": 2.2602758407592773, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7057957649230957, + "num_tokens": 255177627.0, + "step": 10034 + }, + { + "epoch": 1.1020206457280914, + "grad_norm": 1.972583293914795, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7087599635124207, + "num_tokens": 255209252.0, + "step": 10035 + }, + { + "epoch": 1.102130463430705, + "grad_norm": 2.8006670475006104, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7189187407493591, + "num_tokens": 255227858.0, + "step": 10036 + }, + { + "epoch": 1.1022402811333187, + "grad_norm": 2.3558552265167236, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7194814682006836, + "num_tokens": 255250745.0, + "step": 10037 + }, + { + "epoch": 1.1023500988359323, + "grad_norm": 2.4431235790252686, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.732901394367218, + "num_tokens": 255273495.0, + "step": 10038 + }, + { + "epoch": 1.102459916538546, + "grad_norm": 2.0366742610931396, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.731467604637146, + "num_tokens": 255300800.0, + "step": 10039 + }, + { + "epoch": 1.1025697342411598, + "grad_norm": 2.301107883453369, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.716566801071167, + "num_tokens": 255326702.0, + "step": 10040 + }, + { + "epoch": 1.1026795519437733, + "grad_norm": 2.220489740371704, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7093530297279358, + "num_tokens": 255353083.0, + "step": 10041 + }, + { + "epoch": 1.102789369646387, + "grad_norm": 2.187896251678467, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7241111993789673, + "num_tokens": 255377367.0, + "step": 10042 + }, + { + "epoch": 1.1028991873490006, + "grad_norm": 2.4105570316314697, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7183750867843628, + "num_tokens": 255400451.0, + "step": 10043 + }, + { + "epoch": 1.1030090050516144, + "grad_norm": 2.1452112197875977, + "learning_rate": 1e-06, + "loss": 0.7696, + "mean_token_accuracy": 0.7593939304351807, + "num_tokens": 255424741.0, + "step": 10044 + }, + { + "epoch": 1.103118822754228, + "grad_norm": 2.3547170162200928, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7294417023658752, + "num_tokens": 255447506.0, + "step": 10045 + }, + { + "epoch": 1.1032286404568417, + "grad_norm": 2.4102354049682617, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7267571687698364, + "num_tokens": 255471099.0, + "step": 10046 + }, + { + "epoch": 1.1033384581594552, + "grad_norm": 2.342583656311035, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7174187898635864, + "num_tokens": 255495936.0, + "step": 10047 + }, + { + "epoch": 1.103448275862069, + "grad_norm": 2.324070692062378, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7024699449539185, + "num_tokens": 255523075.0, + "step": 10048 + }, + { + "epoch": 1.1035580935646827, + "grad_norm": 2.1989753246307373, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7066625356674194, + "num_tokens": 255550500.0, + "step": 10049 + }, + { + "epoch": 1.1036679112672962, + "grad_norm": 2.5206046104431152, + "learning_rate": 1e-06, + "loss": 0.8026, + "mean_token_accuracy": 0.7513186931610107, + "num_tokens": 255571132.0, + "step": 10050 + }, + { + "epoch": 1.10377772896991, + "grad_norm": 2.3190717697143555, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.732529878616333, + "num_tokens": 255594468.0, + "step": 10051 + }, + { + "epoch": 1.1038875466725235, + "grad_norm": 2.3088276386260986, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7084538340568542, + "num_tokens": 255618810.0, + "step": 10052 + }, + { + "epoch": 1.1039973643751373, + "grad_norm": 2.506467819213867, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.707400918006897, + "num_tokens": 255640117.0, + "step": 10053 + }, + { + "epoch": 1.104107182077751, + "grad_norm": 1.9818596839904785, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7241958975791931, + "num_tokens": 255670360.0, + "step": 10054 + }, + { + "epoch": 1.1042169997803646, + "grad_norm": 2.1594176292419434, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7138630151748657, + "num_tokens": 255699599.0, + "step": 10055 + }, + { + "epoch": 1.1043268174829783, + "grad_norm": 2.1678171157836914, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7193506956100464, + "num_tokens": 255726215.0, + "step": 10056 + }, + { + "epoch": 1.1044366351855919, + "grad_norm": 2.186171054840088, + "learning_rate": 1e-06, + "loss": 1.0161, + "mean_token_accuracy": 0.7016242742538452, + "num_tokens": 255754289.0, + "step": 10057 + }, + { + "epoch": 1.1045464528882056, + "grad_norm": 2.4464354515075684, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7104938626289368, + "num_tokens": 255780655.0, + "step": 10058 + }, + { + "epoch": 1.1046562705908192, + "grad_norm": 2.009796380996704, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7099565267562866, + "num_tokens": 255811745.0, + "step": 10059 + }, + { + "epoch": 1.104766088293433, + "grad_norm": 2.3265342712402344, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.728068470954895, + "num_tokens": 255833295.0, + "step": 10060 + }, + { + "epoch": 1.1048759059960465, + "grad_norm": 2.2695181369781494, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7208817005157471, + "num_tokens": 255857855.0, + "step": 10061 + }, + { + "epoch": 1.1049857236986602, + "grad_norm": 2.2886836528778076, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.720443069934845, + "num_tokens": 255881293.0, + "step": 10062 + }, + { + "epoch": 1.105095541401274, + "grad_norm": 2.0163278579711914, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.707177460193634, + "num_tokens": 255911920.0, + "step": 10063 + }, + { + "epoch": 1.1052053591038875, + "grad_norm": 2.269990921020508, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7251864671707153, + "num_tokens": 255937615.0, + "step": 10064 + }, + { + "epoch": 1.1053151768065013, + "grad_norm": 2.437105894088745, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7465954422950745, + "num_tokens": 255958531.0, + "step": 10065 + }, + { + "epoch": 1.1054249945091148, + "grad_norm": 2.4245223999023438, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7157271504402161, + "num_tokens": 255980675.0, + "step": 10066 + }, + { + "epoch": 1.1055348122117286, + "grad_norm": 2.351215124130249, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7323313355445862, + "num_tokens": 256000909.0, + "step": 10067 + }, + { + "epoch": 1.105644629914342, + "grad_norm": 1.9741287231445312, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.725351870059967, + "num_tokens": 256033330.0, + "step": 10068 + }, + { + "epoch": 1.1057544476169558, + "grad_norm": 2.1327176094055176, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7053273916244507, + "num_tokens": 256060866.0, + "step": 10069 + }, + { + "epoch": 1.1058642653195696, + "grad_norm": 2.166163206100464, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7213170528411865, + "num_tokens": 256088077.0, + "step": 10070 + }, + { + "epoch": 1.1059740830221831, + "grad_norm": 2.0934653282165527, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7166322469711304, + "num_tokens": 256116373.0, + "step": 10071 + }, + { + "epoch": 1.106083900724797, + "grad_norm": 2.161285161972046, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7170071005821228, + "num_tokens": 256142240.0, + "step": 10072 + }, + { + "epoch": 1.1061937184274104, + "grad_norm": 1.991278886795044, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7295417785644531, + "num_tokens": 256173581.0, + "step": 10073 + }, + { + "epoch": 1.1063035361300242, + "grad_norm": 2.0697133541107178, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7271198034286499, + "num_tokens": 256203892.0, + "step": 10074 + }, + { + "epoch": 1.1064133538326377, + "grad_norm": 2.09181809425354, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7113319635391235, + "num_tokens": 256232754.0, + "step": 10075 + }, + { + "epoch": 1.1065231715352515, + "grad_norm": 2.517869710922241, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7235667705535889, + "num_tokens": 256253351.0, + "step": 10076 + }, + { + "epoch": 1.1066329892378652, + "grad_norm": 2.239163875579834, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7144206762313843, + "num_tokens": 256280062.0, + "step": 10077 + }, + { + "epoch": 1.1067428069404788, + "grad_norm": 2.378375768661499, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7330447435379028, + "num_tokens": 256301866.0, + "step": 10078 + }, + { + "epoch": 1.1068526246430925, + "grad_norm": 2.2756259441375732, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7253166437149048, + "num_tokens": 256326812.0, + "step": 10079 + }, + { + "epoch": 1.106962442345706, + "grad_norm": 2.6006321907043457, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7102921009063721, + "num_tokens": 256346671.0, + "step": 10080 + }, + { + "epoch": 1.1070722600483198, + "grad_norm": 2.4809670448303223, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7318265438079834, + "num_tokens": 256367870.0, + "step": 10081 + }, + { + "epoch": 1.1071820777509334, + "grad_norm": 2.563485860824585, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7380205988883972, + "num_tokens": 256388234.0, + "step": 10082 + }, + { + "epoch": 1.1072918954535471, + "grad_norm": 2.256248712539673, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7217175960540771, + "num_tokens": 256412042.0, + "step": 10083 + }, + { + "epoch": 1.1074017131561609, + "grad_norm": 2.28292179107666, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7383304834365845, + "num_tokens": 256436131.0, + "step": 10084 + }, + { + "epoch": 1.1075115308587744, + "grad_norm": 2.3276102542877197, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7398953437805176, + "num_tokens": 256459248.0, + "step": 10085 + }, + { + "epoch": 1.1076213485613882, + "grad_norm": 2.1356496810913086, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7017725706100464, + "num_tokens": 256486326.0, + "step": 10086 + }, + { + "epoch": 1.1077311662640017, + "grad_norm": 2.325472593307495, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7202178835868835, + "num_tokens": 256511044.0, + "step": 10087 + }, + { + "epoch": 1.1078409839666155, + "grad_norm": 2.1445658206939697, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7393145561218262, + "num_tokens": 256536593.0, + "step": 10088 + }, + { + "epoch": 1.107950801669229, + "grad_norm": 2.1653130054473877, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7196226119995117, + "num_tokens": 256561178.0, + "step": 10089 + }, + { + "epoch": 1.1080606193718427, + "grad_norm": 2.5361626148223877, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7237582206726074, + "num_tokens": 256582468.0, + "step": 10090 + }, + { + "epoch": 1.1081704370744565, + "grad_norm": 2.5641932487487793, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7506004571914673, + "num_tokens": 256600559.0, + "step": 10091 + }, + { + "epoch": 1.10828025477707, + "grad_norm": 2.3434958457946777, + "learning_rate": 1e-06, + "loss": 0.8228, + "mean_token_accuracy": 0.7361711263656616, + "num_tokens": 256622250.0, + "step": 10092 + }, + { + "epoch": 1.1083900724796838, + "grad_norm": 1.991901159286499, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7262194156646729, + "num_tokens": 256649825.0, + "step": 10093 + }, + { + "epoch": 1.1084998901822973, + "grad_norm": 2.093334436416626, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.7019938230514526, + "num_tokens": 256679419.0, + "step": 10094 + }, + { + "epoch": 1.108609707884911, + "grad_norm": 2.4620776176452637, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7085428833961487, + "num_tokens": 256703474.0, + "step": 10095 + }, + { + "epoch": 1.1087195255875246, + "grad_norm": 2.196229934692383, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7182807922363281, + "num_tokens": 256729612.0, + "step": 10096 + }, + { + "epoch": 1.1088293432901384, + "grad_norm": 2.361473560333252, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.734364926815033, + "num_tokens": 256751847.0, + "step": 10097 + }, + { + "epoch": 1.108939160992752, + "grad_norm": 2.2352116107940674, + "learning_rate": 1e-06, + "loss": 1.0223, + "mean_token_accuracy": 0.6951665878295898, + "num_tokens": 256780026.0, + "step": 10098 + }, + { + "epoch": 1.1090489786953657, + "grad_norm": 2.200862169265747, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7258645296096802, + "num_tokens": 256806378.0, + "step": 10099 + }, + { + "epoch": 1.1091587963979794, + "grad_norm": 2.2777864933013916, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7219594120979309, + "num_tokens": 256830799.0, + "step": 10100 + }, + { + "epoch": 1.109268614100593, + "grad_norm": 2.242981433868408, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7067322731018066, + "num_tokens": 256855454.0, + "step": 10101 + }, + { + "epoch": 1.1093784318032067, + "grad_norm": 2.003932237625122, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6887050867080688, + "num_tokens": 256887305.0, + "step": 10102 + }, + { + "epoch": 1.1094882495058203, + "grad_norm": 2.244251251220703, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7085697054862976, + "num_tokens": 256914449.0, + "step": 10103 + }, + { + "epoch": 1.109598067208434, + "grad_norm": 2.3291306495666504, + "learning_rate": 1e-06, + "loss": 0.8126, + "mean_token_accuracy": 0.7466126084327698, + "num_tokens": 256937412.0, + "step": 10104 + }, + { + "epoch": 1.1097078849110478, + "grad_norm": 2.6213085651397705, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7082254886627197, + "num_tokens": 256958713.0, + "step": 10105 + }, + { + "epoch": 1.1098177026136613, + "grad_norm": 2.19787859916687, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.6950069665908813, + "num_tokens": 256984990.0, + "step": 10106 + }, + { + "epoch": 1.109927520316275, + "grad_norm": 2.233970880508423, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7283812761306763, + "num_tokens": 257009020.0, + "step": 10107 + }, + { + "epoch": 1.1100373380188886, + "grad_norm": 2.339413642883301, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7165117859840393, + "num_tokens": 257032780.0, + "step": 10108 + }, + { + "epoch": 1.1101471557215024, + "grad_norm": 2.432621955871582, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7063484191894531, + "num_tokens": 257061256.0, + "step": 10109 + }, + { + "epoch": 1.110256973424116, + "grad_norm": 2.1960549354553223, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.709953784942627, + "num_tokens": 257090292.0, + "step": 10110 + }, + { + "epoch": 1.1103667911267296, + "grad_norm": 2.6891255378723145, + "learning_rate": 1e-06, + "loss": 0.7988, + "mean_token_accuracy": 0.738050639629364, + "num_tokens": 257108049.0, + "step": 10111 + }, + { + "epoch": 1.1104766088293432, + "grad_norm": 2.2062103748321533, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7291556000709534, + "num_tokens": 257134686.0, + "step": 10112 + }, + { + "epoch": 1.110586426531957, + "grad_norm": 2.44543719291687, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7240588665008545, + "num_tokens": 257157235.0, + "step": 10113 + }, + { + "epoch": 1.1106962442345707, + "grad_norm": 2.25162410736084, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7123406529426575, + "num_tokens": 257183071.0, + "step": 10114 + }, + { + "epoch": 1.1108060619371842, + "grad_norm": 2.396500587463379, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7282184362411499, + "num_tokens": 257206370.0, + "step": 10115 + }, + { + "epoch": 1.110915879639798, + "grad_norm": 2.437418222427368, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7201818823814392, + "num_tokens": 257228274.0, + "step": 10116 + }, + { + "epoch": 1.1110256973424115, + "grad_norm": 2.0320098400115967, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7247451543807983, + "num_tokens": 257258517.0, + "step": 10117 + }, + { + "epoch": 1.1111355150450253, + "grad_norm": 2.2750015258789062, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7352147102355957, + "num_tokens": 257283761.0, + "step": 10118 + }, + { + "epoch": 1.111245332747639, + "grad_norm": 2.1193764209747314, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7270923256874084, + "num_tokens": 257310648.0, + "step": 10119 + }, + { + "epoch": 1.1113551504502526, + "grad_norm": 2.086439371109009, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7360290288925171, + "num_tokens": 257340383.0, + "step": 10120 + }, + { + "epoch": 1.1114649681528663, + "grad_norm": 2.066624402999878, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7227857708930969, + "num_tokens": 257368222.0, + "step": 10121 + }, + { + "epoch": 1.1115747858554799, + "grad_norm": 2.296424627304077, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.6982895731925964, + "num_tokens": 257393532.0, + "step": 10122 + }, + { + "epoch": 1.1116846035580936, + "grad_norm": 2.5308828353881836, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.734397292137146, + "num_tokens": 257412917.0, + "step": 10123 + }, + { + "epoch": 1.1117944212607072, + "grad_norm": 2.170048236846924, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6957733631134033, + "num_tokens": 257440698.0, + "step": 10124 + }, + { + "epoch": 1.111904238963321, + "grad_norm": 1.8950998783111572, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.718123197555542, + "num_tokens": 257473978.0, + "step": 10125 + }, + { + "epoch": 1.1120140566659344, + "grad_norm": 2.0415256023406982, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7229907512664795, + "num_tokens": 257502453.0, + "step": 10126 + }, + { + "epoch": 1.1121238743685482, + "grad_norm": 2.2289085388183594, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7293462753295898, + "num_tokens": 257527993.0, + "step": 10127 + }, + { + "epoch": 1.112233692071162, + "grad_norm": 2.3651487827301025, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7253228425979614, + "num_tokens": 257552266.0, + "step": 10128 + }, + { + "epoch": 1.1123435097737755, + "grad_norm": 2.3011252880096436, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6930046081542969, + "num_tokens": 257581213.0, + "step": 10129 + }, + { + "epoch": 1.1124533274763893, + "grad_norm": 2.227837085723877, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7277201414108276, + "num_tokens": 257605319.0, + "step": 10130 + }, + { + "epoch": 1.1125631451790028, + "grad_norm": 2.3591301441192627, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7147791385650635, + "num_tokens": 257628965.0, + "step": 10131 + }, + { + "epoch": 1.1126729628816165, + "grad_norm": 2.5124945640563965, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.723064124584198, + "num_tokens": 257649347.0, + "step": 10132 + }, + { + "epoch": 1.11278278058423, + "grad_norm": 2.4553487300872803, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7145037055015564, + "num_tokens": 257670545.0, + "step": 10133 + }, + { + "epoch": 1.1128925982868438, + "grad_norm": 2.434706211090088, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7129166126251221, + "num_tokens": 257693424.0, + "step": 10134 + }, + { + "epoch": 1.1130024159894576, + "grad_norm": 2.362523078918457, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7327688932418823, + "num_tokens": 257716902.0, + "step": 10135 + }, + { + "epoch": 1.1131122336920711, + "grad_norm": 2.2852134704589844, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7027631998062134, + "num_tokens": 257741793.0, + "step": 10136 + }, + { + "epoch": 1.1132220513946849, + "grad_norm": 2.454402208328247, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7235422134399414, + "num_tokens": 257763695.0, + "step": 10137 + }, + { + "epoch": 1.1133318690972984, + "grad_norm": 2.1095736026763916, + "learning_rate": 1e-06, + "loss": 1.0635, + "mean_token_accuracy": 0.6864839792251587, + "num_tokens": 257791391.0, + "step": 10138 + }, + { + "epoch": 1.1134416867999122, + "grad_norm": 2.4838767051696777, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7238979339599609, + "num_tokens": 257814237.0, + "step": 10139 + }, + { + "epoch": 1.1135515045025257, + "grad_norm": 2.162440538406372, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7364393472671509, + "num_tokens": 257840294.0, + "step": 10140 + }, + { + "epoch": 1.1136613222051395, + "grad_norm": 2.673704147338867, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7258026599884033, + "num_tokens": 257861481.0, + "step": 10141 + }, + { + "epoch": 1.1137711399077532, + "grad_norm": 2.5234174728393555, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7339000701904297, + "num_tokens": 257882565.0, + "step": 10142 + }, + { + "epoch": 1.1138809576103668, + "grad_norm": 2.583804130554199, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7185941338539124, + "num_tokens": 257904594.0, + "step": 10143 + }, + { + "epoch": 1.1139907753129805, + "grad_norm": 2.3706889152526855, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7299407720565796, + "num_tokens": 257927648.0, + "step": 10144 + }, + { + "epoch": 1.114100593015594, + "grad_norm": 2.7303948402404785, + "learning_rate": 1e-06, + "loss": 0.7687, + "mean_token_accuracy": 0.7487861514091492, + "num_tokens": 257944733.0, + "step": 10145 + }, + { + "epoch": 1.1142104107182078, + "grad_norm": 2.1975386142730713, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7085276246070862, + "num_tokens": 257973386.0, + "step": 10146 + }, + { + "epoch": 1.1143202284208213, + "grad_norm": 2.193652629852295, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7142354249954224, + "num_tokens": 257997414.0, + "step": 10147 + }, + { + "epoch": 1.114430046123435, + "grad_norm": 2.1552388668060303, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7171969413757324, + "num_tokens": 258025604.0, + "step": 10148 + }, + { + "epoch": 1.1145398638260489, + "grad_norm": 2.2420969009399414, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7191076278686523, + "num_tokens": 258050765.0, + "step": 10149 + }, + { + "epoch": 1.1146496815286624, + "grad_norm": 2.278510332107544, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7357043623924255, + "num_tokens": 258073939.0, + "step": 10150 + }, + { + "epoch": 1.1147594992312762, + "grad_norm": 2.371229887008667, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6887987852096558, + "num_tokens": 258099971.0, + "step": 10151 + }, + { + "epoch": 1.1148693169338897, + "grad_norm": 2.281304359436035, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7173824310302734, + "num_tokens": 258127443.0, + "step": 10152 + }, + { + "epoch": 1.1149791346365034, + "grad_norm": 2.4779767990112305, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7235673666000366, + "num_tokens": 258150449.0, + "step": 10153 + }, + { + "epoch": 1.115088952339117, + "grad_norm": 2.19868803024292, + "learning_rate": 1e-06, + "loss": 1.0519, + "mean_token_accuracy": 0.6894908547401428, + "num_tokens": 258179387.0, + "step": 10154 + }, + { + "epoch": 1.1151987700417307, + "grad_norm": 2.258546829223633, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7137628793716431, + "num_tokens": 258205088.0, + "step": 10155 + }, + { + "epoch": 1.1153085877443445, + "grad_norm": 2.273078203201294, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7049934267997742, + "num_tokens": 258232858.0, + "step": 10156 + }, + { + "epoch": 1.115418405446958, + "grad_norm": 2.3792946338653564, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7281148433685303, + "num_tokens": 258255282.0, + "step": 10157 + }, + { + "epoch": 1.1155282231495718, + "grad_norm": 2.3705852031707764, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7334953546524048, + "num_tokens": 258277609.0, + "step": 10158 + }, + { + "epoch": 1.1156380408521853, + "grad_norm": 2.367675304412842, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7178612351417542, + "num_tokens": 258300391.0, + "step": 10159 + }, + { + "epoch": 1.115747858554799, + "grad_norm": 2.0442724227905273, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7112269401550293, + "num_tokens": 258328469.0, + "step": 10160 + }, + { + "epoch": 1.1158576762574126, + "grad_norm": 2.596710205078125, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7159972190856934, + "num_tokens": 258349510.0, + "step": 10161 + }, + { + "epoch": 1.1159674939600264, + "grad_norm": 2.542048692703247, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7224136590957642, + "num_tokens": 258371325.0, + "step": 10162 + }, + { + "epoch": 1.11607731166264, + "grad_norm": 2.2265567779541016, + "learning_rate": 1e-06, + "loss": 0.9729, + "mean_token_accuracy": 0.7009077668190002, + "num_tokens": 258398609.0, + "step": 10163 + }, + { + "epoch": 1.1161871293652537, + "grad_norm": 2.0523030757904053, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7050207257270813, + "num_tokens": 258428415.0, + "step": 10164 + }, + { + "epoch": 1.1162969470678674, + "grad_norm": 2.190304756164551, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7145260572433472, + "num_tokens": 258454748.0, + "step": 10165 + }, + { + "epoch": 1.116406764770481, + "grad_norm": 2.222757339477539, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7124935984611511, + "num_tokens": 258482233.0, + "step": 10166 + }, + { + "epoch": 1.1165165824730947, + "grad_norm": 2.3708603382110596, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7023015022277832, + "num_tokens": 258503441.0, + "step": 10167 + }, + { + "epoch": 1.1166264001757082, + "grad_norm": 2.5380256175994873, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7160869240760803, + "num_tokens": 258527009.0, + "step": 10168 + }, + { + "epoch": 1.116736217878322, + "grad_norm": 2.3161063194274902, + "learning_rate": 1e-06, + "loss": 1.022, + "mean_token_accuracy": 0.6852843761444092, + "num_tokens": 258554539.0, + "step": 10169 + }, + { + "epoch": 1.1168460355809358, + "grad_norm": 2.2963945865631104, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7280805110931396, + "num_tokens": 258582159.0, + "step": 10170 + }, + { + "epoch": 1.1169558532835493, + "grad_norm": 2.1968629360198975, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7059668302536011, + "num_tokens": 258609543.0, + "step": 10171 + }, + { + "epoch": 1.117065670986163, + "grad_norm": 2.632981538772583, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7405261993408203, + "num_tokens": 258629279.0, + "step": 10172 + }, + { + "epoch": 1.1171754886887766, + "grad_norm": 2.5787713527679443, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7173936367034912, + "num_tokens": 258651751.0, + "step": 10173 + }, + { + "epoch": 1.1172853063913903, + "grad_norm": 2.3448898792266846, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7018142342567444, + "num_tokens": 258677449.0, + "step": 10174 + }, + { + "epoch": 1.1173951240940039, + "grad_norm": 2.289403200149536, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.714016318321228, + "num_tokens": 258704235.0, + "step": 10175 + }, + { + "epoch": 1.1175049417966176, + "grad_norm": 2.1610310077667236, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7324269413948059, + "num_tokens": 258731278.0, + "step": 10176 + }, + { + "epoch": 1.1176147594992312, + "grad_norm": 2.5144641399383545, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7438760995864868, + "num_tokens": 258752843.0, + "step": 10177 + }, + { + "epoch": 1.117724577201845, + "grad_norm": 2.483391284942627, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7325104475021362, + "num_tokens": 258775239.0, + "step": 10178 + }, + { + "epoch": 1.1178343949044587, + "grad_norm": 2.38080096244812, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.739469051361084, + "num_tokens": 258799616.0, + "step": 10179 + }, + { + "epoch": 1.1179442126070722, + "grad_norm": 2.198728561401367, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7030184864997864, + "num_tokens": 258828510.0, + "step": 10180 + }, + { + "epoch": 1.118054030309686, + "grad_norm": 2.369246006011963, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7064220309257507, + "num_tokens": 258854566.0, + "step": 10181 + }, + { + "epoch": 1.1181638480122995, + "grad_norm": 2.48626708984375, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7095618844032288, + "num_tokens": 258876228.0, + "step": 10182 + }, + { + "epoch": 1.1182736657149133, + "grad_norm": 2.083024501800537, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7299560308456421, + "num_tokens": 258905107.0, + "step": 10183 + }, + { + "epoch": 1.1183834834175268, + "grad_norm": 2.457932472229004, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.6998296976089478, + "num_tokens": 258927083.0, + "step": 10184 + }, + { + "epoch": 1.1184933011201406, + "grad_norm": 2.28877329826355, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7237379550933838, + "num_tokens": 258950648.0, + "step": 10185 + }, + { + "epoch": 1.1186031188227543, + "grad_norm": 2.3771820068359375, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7148092985153198, + "num_tokens": 258974962.0, + "step": 10186 + }, + { + "epoch": 1.1187129365253679, + "grad_norm": 2.3511624336242676, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6919798851013184, + "num_tokens": 259000714.0, + "step": 10187 + }, + { + "epoch": 1.1188227542279816, + "grad_norm": 2.608471155166626, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7315798997879028, + "num_tokens": 259021113.0, + "step": 10188 + }, + { + "epoch": 1.1189325719305951, + "grad_norm": 2.079475164413452, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7058849930763245, + "num_tokens": 259049423.0, + "step": 10189 + }, + { + "epoch": 1.119042389633209, + "grad_norm": 2.2941060066223145, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7155701518058777, + "num_tokens": 259073298.0, + "step": 10190 + }, + { + "epoch": 1.1191522073358224, + "grad_norm": 2.113962173461914, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7272768020629883, + "num_tokens": 259101164.0, + "step": 10191 + }, + { + "epoch": 1.1192620250384362, + "grad_norm": 2.4107179641723633, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7191306352615356, + "num_tokens": 259121808.0, + "step": 10192 + }, + { + "epoch": 1.11937184274105, + "grad_norm": 2.4782118797302246, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7370938062667847, + "num_tokens": 259143270.0, + "step": 10193 + }, + { + "epoch": 1.1194816604436635, + "grad_norm": 2.3354437351226807, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.723946213722229, + "num_tokens": 259166378.0, + "step": 10194 + }, + { + "epoch": 1.1195914781462772, + "grad_norm": 2.1579537391662598, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.6985455751419067, + "num_tokens": 259195073.0, + "step": 10195 + }, + { + "epoch": 1.1197012958488908, + "grad_norm": 2.4057013988494873, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7164784669876099, + "num_tokens": 259217368.0, + "step": 10196 + }, + { + "epoch": 1.1198111135515045, + "grad_norm": 2.4680681228637695, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7317969799041748, + "num_tokens": 259241112.0, + "step": 10197 + }, + { + "epoch": 1.119920931254118, + "grad_norm": 2.190352439880371, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.726544976234436, + "num_tokens": 259267488.0, + "step": 10198 + }, + { + "epoch": 1.1200307489567318, + "grad_norm": 2.1778688430786133, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.704287588596344, + "num_tokens": 259296110.0, + "step": 10199 + }, + { + "epoch": 1.1201405666593456, + "grad_norm": 2.4639134407043457, + "learning_rate": 1e-06, + "loss": 0.8202, + "mean_token_accuracy": 0.7400447130203247, + "num_tokens": 259316868.0, + "step": 10200 + }, + { + "epoch": 1.1202503843619591, + "grad_norm": 1.9709078073501587, + "learning_rate": 1e-06, + "loss": 1.0219, + "mean_token_accuracy": 0.6827074289321899, + "num_tokens": 259350679.0, + "step": 10201 + }, + { + "epoch": 1.1203602020645729, + "grad_norm": 2.1659629344940186, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.706337034702301, + "num_tokens": 259379183.0, + "step": 10202 + }, + { + "epoch": 1.1204700197671864, + "grad_norm": 2.374915361404419, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7075629234313965, + "num_tokens": 259406393.0, + "step": 10203 + }, + { + "epoch": 1.1205798374698002, + "grad_norm": 2.297043561935425, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.724851131439209, + "num_tokens": 259432237.0, + "step": 10204 + }, + { + "epoch": 1.1206896551724137, + "grad_norm": 2.117847204208374, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.699199914932251, + "num_tokens": 259461489.0, + "step": 10205 + }, + { + "epoch": 1.1207994728750275, + "grad_norm": 2.1022214889526367, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7091354131698608, + "num_tokens": 259491188.0, + "step": 10206 + }, + { + "epoch": 1.1209092905776412, + "grad_norm": 2.5660336017608643, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7432355880737305, + "num_tokens": 259511417.0, + "step": 10207 + }, + { + "epoch": 1.1210191082802548, + "grad_norm": 2.3767542839050293, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.7020668387413025, + "num_tokens": 259535359.0, + "step": 10208 + }, + { + "epoch": 1.1211289259828685, + "grad_norm": 2.2562012672424316, + "learning_rate": 1e-06, + "loss": 0.9669, + "mean_token_accuracy": 0.7094095945358276, + "num_tokens": 259563221.0, + "step": 10209 + }, + { + "epoch": 1.121238743685482, + "grad_norm": 2.3484110832214355, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7142361402511597, + "num_tokens": 259587045.0, + "step": 10210 + }, + { + "epoch": 1.1213485613880958, + "grad_norm": 2.219513177871704, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7202539443969727, + "num_tokens": 259612951.0, + "step": 10211 + }, + { + "epoch": 1.1214583790907093, + "grad_norm": 2.1931817531585693, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7110614776611328, + "num_tokens": 259639548.0, + "step": 10212 + }, + { + "epoch": 1.121568196793323, + "grad_norm": 2.322141408920288, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7309330105781555, + "num_tokens": 259661823.0, + "step": 10213 + }, + { + "epoch": 1.1216780144959368, + "grad_norm": 2.034468650817871, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7104920148849487, + "num_tokens": 259691490.0, + "step": 10214 + }, + { + "epoch": 1.1217878321985504, + "grad_norm": 2.405460834503174, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7206324934959412, + "num_tokens": 259714388.0, + "step": 10215 + }, + { + "epoch": 1.1218976499011641, + "grad_norm": 2.209247350692749, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7006279230117798, + "num_tokens": 259739548.0, + "step": 10216 + }, + { + "epoch": 1.1220074676037777, + "grad_norm": 2.5028178691864014, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7251215577125549, + "num_tokens": 259759079.0, + "step": 10217 + }, + { + "epoch": 1.1221172853063914, + "grad_norm": 2.1839542388916016, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6911977529525757, + "num_tokens": 259789309.0, + "step": 10218 + }, + { + "epoch": 1.122227103009005, + "grad_norm": 2.446784496307373, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7342200875282288, + "num_tokens": 259811231.0, + "step": 10219 + }, + { + "epoch": 1.1223369207116187, + "grad_norm": 2.1600825786590576, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7054459452629089, + "num_tokens": 259837206.0, + "step": 10220 + }, + { + "epoch": 1.1224467384142325, + "grad_norm": 2.135830879211426, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7144558429718018, + "num_tokens": 259863429.0, + "step": 10221 + }, + { + "epoch": 1.122556556116846, + "grad_norm": 2.1148266792297363, + "learning_rate": 1e-06, + "loss": 1.0467, + "mean_token_accuracy": 0.6806082725524902, + "num_tokens": 259894665.0, + "step": 10222 + }, + { + "epoch": 1.1226663738194598, + "grad_norm": 1.830804705619812, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7157248258590698, + "num_tokens": 259928597.0, + "step": 10223 + }, + { + "epoch": 1.1227761915220733, + "grad_norm": 2.2423229217529297, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7015243172645569, + "num_tokens": 259953345.0, + "step": 10224 + }, + { + "epoch": 1.122886009224687, + "grad_norm": 2.1350040435791016, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7084959149360657, + "num_tokens": 259983316.0, + "step": 10225 + }, + { + "epoch": 1.1229958269273006, + "grad_norm": 2.181075096130371, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.6983663439750671, + "num_tokens": 260014181.0, + "step": 10226 + }, + { + "epoch": 1.1231056446299144, + "grad_norm": 2.4837944507598877, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.736794114112854, + "num_tokens": 260034255.0, + "step": 10227 + }, + { + "epoch": 1.123215462332528, + "grad_norm": 2.189884901046753, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7246206402778625, + "num_tokens": 260061842.0, + "step": 10228 + }, + { + "epoch": 1.1233252800351416, + "grad_norm": 2.133729934692383, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.719411313533783, + "num_tokens": 260089176.0, + "step": 10229 + }, + { + "epoch": 1.1234350977377554, + "grad_norm": 2.186650276184082, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7238776087760925, + "num_tokens": 260114454.0, + "step": 10230 + }, + { + "epoch": 1.123544915440369, + "grad_norm": 2.171651840209961, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6938480734825134, + "num_tokens": 260144151.0, + "step": 10231 + }, + { + "epoch": 1.1236547331429827, + "grad_norm": 2.7029590606689453, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7226837873458862, + "num_tokens": 260164440.0, + "step": 10232 + }, + { + "epoch": 1.1237645508455962, + "grad_norm": 2.157274007797241, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7168739438056946, + "num_tokens": 260189905.0, + "step": 10233 + }, + { + "epoch": 1.12387436854821, + "grad_norm": 2.319455862045288, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7158575057983398, + "num_tokens": 260213713.0, + "step": 10234 + }, + { + "epoch": 1.1239841862508237, + "grad_norm": 2.5385630130767822, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7246838212013245, + "num_tokens": 260234743.0, + "step": 10235 + }, + { + "epoch": 1.1240940039534373, + "grad_norm": 2.479006052017212, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7325143814086914, + "num_tokens": 260254988.0, + "step": 10236 + }, + { + "epoch": 1.124203821656051, + "grad_norm": 2.3410909175872803, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7306588292121887, + "num_tokens": 260278407.0, + "step": 10237 + }, + { + "epoch": 1.1243136393586646, + "grad_norm": 2.3492915630340576, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7147393226623535, + "num_tokens": 260301400.0, + "step": 10238 + }, + { + "epoch": 1.1244234570612783, + "grad_norm": 2.277208089828491, + "learning_rate": 1e-06, + "loss": 1.0265, + "mean_token_accuracy": 0.6982428431510925, + "num_tokens": 260326902.0, + "step": 10239 + }, + { + "epoch": 1.1245332747638919, + "grad_norm": 2.1395933628082275, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7038934230804443, + "num_tokens": 260356747.0, + "step": 10240 + }, + { + "epoch": 1.1246430924665056, + "grad_norm": 2.0898001194000244, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7151617407798767, + "num_tokens": 260385271.0, + "step": 10241 + }, + { + "epoch": 1.1247529101691192, + "grad_norm": 2.1396334171295166, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.711323082447052, + "num_tokens": 260415161.0, + "step": 10242 + }, + { + "epoch": 1.124862727871733, + "grad_norm": 2.146022319793701, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7305773496627808, + "num_tokens": 260442489.0, + "step": 10243 + }, + { + "epoch": 1.1249725455743467, + "grad_norm": 2.2404768466949463, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7244744896888733, + "num_tokens": 260467370.0, + "step": 10244 + }, + { + "epoch": 1.1250823632769602, + "grad_norm": 2.443830966949463, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7104021906852722, + "num_tokens": 260490696.0, + "step": 10245 + }, + { + "epoch": 1.125192180979574, + "grad_norm": 2.2180488109588623, + "learning_rate": 1e-06, + "loss": 1.0156, + "mean_token_accuracy": 0.687055230140686, + "num_tokens": 260520735.0, + "step": 10246 + }, + { + "epoch": 1.1253019986821875, + "grad_norm": 2.4911279678344727, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7256947159767151, + "num_tokens": 260542598.0, + "step": 10247 + }, + { + "epoch": 1.1254118163848013, + "grad_norm": 2.216818332672119, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7175620198249817, + "num_tokens": 260568781.0, + "step": 10248 + }, + { + "epoch": 1.125521634087415, + "grad_norm": 2.267918348312378, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7200271487236023, + "num_tokens": 260595549.0, + "step": 10249 + }, + { + "epoch": 1.1256314517900285, + "grad_norm": 2.421627998352051, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7327357530593872, + "num_tokens": 260617273.0, + "step": 10250 + }, + { + "epoch": 1.1257412694926423, + "grad_norm": 2.247666835784912, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.7002068161964417, + "num_tokens": 260643622.0, + "step": 10251 + }, + { + "epoch": 1.1258510871952558, + "grad_norm": 2.1752095222473145, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.718336820602417, + "num_tokens": 260671199.0, + "step": 10252 + }, + { + "epoch": 1.1259609048978696, + "grad_norm": 2.0713491439819336, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7297542691230774, + "num_tokens": 260698917.0, + "step": 10253 + }, + { + "epoch": 1.1260707226004831, + "grad_norm": 2.2577548027038574, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7200533151626587, + "num_tokens": 260723693.0, + "step": 10254 + }, + { + "epoch": 1.1261805403030969, + "grad_norm": 2.132803440093994, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7101577520370483, + "num_tokens": 260753016.0, + "step": 10255 + }, + { + "epoch": 1.1262903580057104, + "grad_norm": 2.539215087890625, + "learning_rate": 1e-06, + "loss": 0.8134, + "mean_token_accuracy": 0.7414527535438538, + "num_tokens": 260773923.0, + "step": 10256 + }, + { + "epoch": 1.1264001757083242, + "grad_norm": 2.205447196960449, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7119674682617188, + "num_tokens": 260800452.0, + "step": 10257 + }, + { + "epoch": 1.126509993410938, + "grad_norm": 2.2363903522491455, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7139592170715332, + "num_tokens": 260826767.0, + "step": 10258 + }, + { + "epoch": 1.1266198111135515, + "grad_norm": 2.2010316848754883, + "learning_rate": 1e-06, + "loss": 0.9958, + "mean_token_accuracy": 0.6917163729667664, + "num_tokens": 260856826.0, + "step": 10259 + }, + { + "epoch": 1.1267296288161652, + "grad_norm": 2.2387897968292236, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7066153287887573, + "num_tokens": 260883215.0, + "step": 10260 + }, + { + "epoch": 1.1268394465187788, + "grad_norm": 2.18532133102417, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7148690223693848, + "num_tokens": 260910611.0, + "step": 10261 + }, + { + "epoch": 1.1269492642213925, + "grad_norm": 2.3151116371154785, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.7300235629081726, + "num_tokens": 260934576.0, + "step": 10262 + }, + { + "epoch": 1.127059081924006, + "grad_norm": 2.5878803730010986, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7274661064147949, + "num_tokens": 260954265.0, + "step": 10263 + }, + { + "epoch": 1.1271688996266198, + "grad_norm": 2.490062713623047, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7223787903785706, + "num_tokens": 260976428.0, + "step": 10264 + }, + { + "epoch": 1.1272787173292333, + "grad_norm": 2.2445952892303467, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7062425017356873, + "num_tokens": 261003036.0, + "step": 10265 + }, + { + "epoch": 1.127388535031847, + "grad_norm": 2.1552000045776367, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.7518322467803955, + "num_tokens": 261027809.0, + "step": 10266 + }, + { + "epoch": 1.1274983527344609, + "grad_norm": 2.201460123062134, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7093352675437927, + "num_tokens": 261055054.0, + "step": 10267 + }, + { + "epoch": 1.1276081704370744, + "grad_norm": 2.5576748847961426, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7340285778045654, + "num_tokens": 261075419.0, + "step": 10268 + }, + { + "epoch": 1.1277179881396882, + "grad_norm": 2.6627206802368164, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7102916240692139, + "num_tokens": 261094167.0, + "step": 10269 + }, + { + "epoch": 1.1278278058423017, + "grad_norm": 2.401874542236328, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7223362922668457, + "num_tokens": 261117346.0, + "step": 10270 + }, + { + "epoch": 1.1279376235449154, + "grad_norm": 2.253661870956421, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.6969184279441833, + "num_tokens": 261142179.0, + "step": 10271 + }, + { + "epoch": 1.1280474412475292, + "grad_norm": 2.1199803352355957, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7161351442337036, + "num_tokens": 261172449.0, + "step": 10272 + }, + { + "epoch": 1.1281572589501427, + "grad_norm": 2.263298749923706, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7155661582946777, + "num_tokens": 261196795.0, + "step": 10273 + }, + { + "epoch": 1.1282670766527565, + "grad_norm": 2.4176571369171143, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7075409889221191, + "num_tokens": 261220736.0, + "step": 10274 + }, + { + "epoch": 1.12837689435537, + "grad_norm": 2.250417470932007, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7248692512512207, + "num_tokens": 261245665.0, + "step": 10275 + }, + { + "epoch": 1.1284867120579838, + "grad_norm": 2.0113654136657715, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7253672480583191, + "num_tokens": 261274077.0, + "step": 10276 + }, + { + "epoch": 1.1285965297605973, + "grad_norm": 2.7029471397399902, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7210180163383484, + "num_tokens": 261292889.0, + "step": 10277 + }, + { + "epoch": 1.128706347463211, + "grad_norm": 2.5729236602783203, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7056965827941895, + "num_tokens": 261314262.0, + "step": 10278 + }, + { + "epoch": 1.1288161651658246, + "grad_norm": 2.149834156036377, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.695393443107605, + "num_tokens": 261342854.0, + "step": 10279 + }, + { + "epoch": 1.1289259828684384, + "grad_norm": 2.6143319606781006, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.717645525932312, + "num_tokens": 261367684.0, + "step": 10280 + }, + { + "epoch": 1.1290358005710521, + "grad_norm": 2.065945625305176, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7166142463684082, + "num_tokens": 261395820.0, + "step": 10281 + }, + { + "epoch": 1.1291456182736657, + "grad_norm": 2.510796546936035, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7208531498908997, + "num_tokens": 261417655.0, + "step": 10282 + }, + { + "epoch": 1.1292554359762794, + "grad_norm": 2.1035027503967285, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.6994462013244629, + "num_tokens": 261445931.0, + "step": 10283 + }, + { + "epoch": 1.129365253678893, + "grad_norm": 2.107283115386963, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.6939480304718018, + "num_tokens": 261476201.0, + "step": 10284 + }, + { + "epoch": 1.1294750713815067, + "grad_norm": 2.038125991821289, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7258172035217285, + "num_tokens": 261507597.0, + "step": 10285 + }, + { + "epoch": 1.1295848890841205, + "grad_norm": 2.3905274868011475, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7152259945869446, + "num_tokens": 261531956.0, + "step": 10286 + }, + { + "epoch": 1.129694706786734, + "grad_norm": 2.4856350421905518, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7350360155105591, + "num_tokens": 261552465.0, + "step": 10287 + }, + { + "epoch": 1.1298045244893478, + "grad_norm": 2.239354133605957, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7305189371109009, + "num_tokens": 261577215.0, + "step": 10288 + }, + { + "epoch": 1.1299143421919613, + "grad_norm": 2.2998669147491455, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7308765649795532, + "num_tokens": 261601368.0, + "step": 10289 + }, + { + "epoch": 1.130024159894575, + "grad_norm": 2.0707058906555176, + "learning_rate": 1e-06, + "loss": 1.0367, + "mean_token_accuracy": 0.6851286888122559, + "num_tokens": 261630892.0, + "step": 10290 + }, + { + "epoch": 1.1301339775971886, + "grad_norm": 2.4005517959594727, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.719388484954834, + "num_tokens": 261653319.0, + "step": 10291 + }, + { + "epoch": 1.1302437952998023, + "grad_norm": 2.11812162399292, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7197110056877136, + "num_tokens": 261682210.0, + "step": 10292 + }, + { + "epoch": 1.1303536130024159, + "grad_norm": 2.0930490493774414, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7170117497444153, + "num_tokens": 261709595.0, + "step": 10293 + }, + { + "epoch": 1.1304634307050296, + "grad_norm": 2.71601939201355, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.6994776725769043, + "num_tokens": 261729753.0, + "step": 10294 + }, + { + "epoch": 1.1305732484076434, + "grad_norm": 2.447580099105835, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7244453430175781, + "num_tokens": 261751435.0, + "step": 10295 + }, + { + "epoch": 1.130683066110257, + "grad_norm": 1.9898731708526611, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7100616097450256, + "num_tokens": 261783300.0, + "step": 10296 + }, + { + "epoch": 1.1307928838128707, + "grad_norm": 2.3257570266723633, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7359678745269775, + "num_tokens": 261806004.0, + "step": 10297 + }, + { + "epoch": 1.1309027015154842, + "grad_norm": 2.383140802383423, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7068142294883728, + "num_tokens": 261828095.0, + "step": 10298 + }, + { + "epoch": 1.131012519218098, + "grad_norm": 2.3461735248565674, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7133916020393372, + "num_tokens": 261853456.0, + "step": 10299 + }, + { + "epoch": 1.1311223369207117, + "grad_norm": 2.338001251220703, + "learning_rate": 1e-06, + "loss": 0.9634, + "mean_token_accuracy": 0.7032384872436523, + "num_tokens": 261876928.0, + "step": 10300 + }, + { + "epoch": 1.1312321546233253, + "grad_norm": 2.284233331680298, + "learning_rate": 1e-06, + "loss": 0.9633, + "mean_token_accuracy": 0.7038133144378662, + "num_tokens": 261902245.0, + "step": 10301 + }, + { + "epoch": 1.131341972325939, + "grad_norm": 2.2456657886505127, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7204165458679199, + "num_tokens": 261926867.0, + "step": 10302 + }, + { + "epoch": 1.1314517900285526, + "grad_norm": 2.302633285522461, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7053533792495728, + "num_tokens": 261953832.0, + "step": 10303 + }, + { + "epoch": 1.1315616077311663, + "grad_norm": 2.5375683307647705, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7203031778335571, + "num_tokens": 261975550.0, + "step": 10304 + }, + { + "epoch": 1.1316714254337799, + "grad_norm": 2.345468044281006, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7307744026184082, + "num_tokens": 261998455.0, + "step": 10305 + }, + { + "epoch": 1.1317812431363936, + "grad_norm": 1.9678951501846313, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7209635972976685, + "num_tokens": 262031860.0, + "step": 10306 + }, + { + "epoch": 1.1318910608390071, + "grad_norm": 2.1559104919433594, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7081851959228516, + "num_tokens": 262059808.0, + "step": 10307 + }, + { + "epoch": 1.132000878541621, + "grad_norm": 2.5832412242889404, + "learning_rate": 1e-06, + "loss": 0.7863, + "mean_token_accuracy": 0.7445226907730103, + "num_tokens": 262078699.0, + "step": 10308 + }, + { + "epoch": 1.1321106962442347, + "grad_norm": 2.1025075912475586, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7085161209106445, + "num_tokens": 262106655.0, + "step": 10309 + }, + { + "epoch": 1.1322205139468482, + "grad_norm": 2.1479923725128174, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.737046480178833, + "num_tokens": 262132991.0, + "step": 10310 + }, + { + "epoch": 1.132330331649462, + "grad_norm": 2.1827125549316406, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7077404260635376, + "num_tokens": 262161151.0, + "step": 10311 + }, + { + "epoch": 1.1324401493520755, + "grad_norm": 2.4662184715270996, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7245147228240967, + "num_tokens": 262183584.0, + "step": 10312 + }, + { + "epoch": 1.1325499670546892, + "grad_norm": 2.09354829788208, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.7044371962547302, + "num_tokens": 262212779.0, + "step": 10313 + }, + { + "epoch": 1.132659784757303, + "grad_norm": 2.803103446960449, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7208791971206665, + "num_tokens": 262231438.0, + "step": 10314 + }, + { + "epoch": 1.1327696024599165, + "grad_norm": 2.762392282485962, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7105479836463928, + "num_tokens": 262251057.0, + "step": 10315 + }, + { + "epoch": 1.1328794201625303, + "grad_norm": 2.1676700115203857, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7217808365821838, + "num_tokens": 262278013.0, + "step": 10316 + }, + { + "epoch": 1.1329892378651438, + "grad_norm": 2.3001959323883057, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7134271264076233, + "num_tokens": 262302419.0, + "step": 10317 + }, + { + "epoch": 1.1330990555677576, + "grad_norm": 2.0368151664733887, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.708810567855835, + "num_tokens": 262331808.0, + "step": 10318 + }, + { + "epoch": 1.1332088732703711, + "grad_norm": 2.251049518585205, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7127912044525146, + "num_tokens": 262358284.0, + "step": 10319 + }, + { + "epoch": 1.1333186909729849, + "grad_norm": 2.1438169479370117, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7373231649398804, + "num_tokens": 262383791.0, + "step": 10320 + }, + { + "epoch": 1.1334285086755984, + "grad_norm": 2.3603780269622803, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7087429761886597, + "num_tokens": 262410781.0, + "step": 10321 + }, + { + "epoch": 1.1335383263782122, + "grad_norm": 2.746452808380127, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.726961612701416, + "num_tokens": 262428219.0, + "step": 10322 + }, + { + "epoch": 1.133648144080826, + "grad_norm": 2.029543161392212, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7107548117637634, + "num_tokens": 262458393.0, + "step": 10323 + }, + { + "epoch": 1.1337579617834395, + "grad_norm": 2.5243213176727295, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.709250807762146, + "num_tokens": 262480860.0, + "step": 10324 + }, + { + "epoch": 1.1338677794860532, + "grad_norm": 2.15570068359375, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.6975123286247253, + "num_tokens": 262512555.0, + "step": 10325 + }, + { + "epoch": 1.1339775971886668, + "grad_norm": 2.356509208679199, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.6989949345588684, + "num_tokens": 262538696.0, + "step": 10326 + }, + { + "epoch": 1.1340874148912805, + "grad_norm": 2.404219150543213, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7104599475860596, + "num_tokens": 262563046.0, + "step": 10327 + }, + { + "epoch": 1.134197232593894, + "grad_norm": 2.422903060913086, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7228533625602722, + "num_tokens": 262585503.0, + "step": 10328 + }, + { + "epoch": 1.1343070502965078, + "grad_norm": 2.1932733058929443, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7004165649414062, + "num_tokens": 262614444.0, + "step": 10329 + }, + { + "epoch": 1.1344168679991213, + "grad_norm": 2.456308364868164, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7292425632476807, + "num_tokens": 262636219.0, + "step": 10330 + }, + { + "epoch": 1.134526685701735, + "grad_norm": 2.212501287460327, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7238839268684387, + "num_tokens": 262661084.0, + "step": 10331 + }, + { + "epoch": 1.1346365034043489, + "grad_norm": 2.220341920852661, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.710172176361084, + "num_tokens": 262687776.0, + "step": 10332 + }, + { + "epoch": 1.1347463211069624, + "grad_norm": 2.3007187843322754, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7242153882980347, + "num_tokens": 262711459.0, + "step": 10333 + }, + { + "epoch": 1.1348561388095761, + "grad_norm": 2.2649059295654297, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7037990689277649, + "num_tokens": 262738157.0, + "step": 10334 + }, + { + "epoch": 1.1349659565121897, + "grad_norm": 2.1842100620269775, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7000702023506165, + "num_tokens": 262769133.0, + "step": 10335 + }, + { + "epoch": 1.1350757742148034, + "grad_norm": 2.249689817428589, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7116878032684326, + "num_tokens": 262795266.0, + "step": 10336 + }, + { + "epoch": 1.1351855919174172, + "grad_norm": 2.129767417907715, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7360643148422241, + "num_tokens": 262822586.0, + "step": 10337 + }, + { + "epoch": 1.1352954096200307, + "grad_norm": 2.159977674484253, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7178959846496582, + "num_tokens": 262850085.0, + "step": 10338 + }, + { + "epoch": 1.1354052273226445, + "grad_norm": 2.6405200958251953, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.735904335975647, + "num_tokens": 262869282.0, + "step": 10339 + }, + { + "epoch": 1.135515045025258, + "grad_norm": 2.269805431365967, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7016701698303223, + "num_tokens": 262894944.0, + "step": 10340 + }, + { + "epoch": 1.1356248627278718, + "grad_norm": 2.0816874504089355, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6951120495796204, + "num_tokens": 262924465.0, + "step": 10341 + }, + { + "epoch": 1.1357346804304853, + "grad_norm": 2.708603620529175, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.731662392616272, + "num_tokens": 262944609.0, + "step": 10342 + }, + { + "epoch": 1.135844498133099, + "grad_norm": 2.0162527561187744, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7168090343475342, + "num_tokens": 262975698.0, + "step": 10343 + }, + { + "epoch": 1.1359543158357126, + "grad_norm": 2.598224401473999, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7046182155609131, + "num_tokens": 263000368.0, + "step": 10344 + }, + { + "epoch": 1.1360641335383264, + "grad_norm": 2.2606117725372314, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7160618901252747, + "num_tokens": 263025898.0, + "step": 10345 + }, + { + "epoch": 1.1361739512409401, + "grad_norm": 2.1602981090545654, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7215036153793335, + "num_tokens": 263051276.0, + "step": 10346 + }, + { + "epoch": 1.1362837689435537, + "grad_norm": 2.3469552993774414, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7296400666236877, + "num_tokens": 263073889.0, + "step": 10347 + }, + { + "epoch": 1.1363935866461674, + "grad_norm": 2.151247978210449, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6953699588775635, + "num_tokens": 263104679.0, + "step": 10348 + }, + { + "epoch": 1.136503404348781, + "grad_norm": 2.3951714038848877, + "learning_rate": 1e-06, + "loss": 0.7947, + "mean_token_accuracy": 0.7510388493537903, + "num_tokens": 263126019.0, + "step": 10349 + }, + { + "epoch": 1.1366132220513947, + "grad_norm": 2.4009149074554443, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7169126272201538, + "num_tokens": 263149092.0, + "step": 10350 + }, + { + "epoch": 1.1367230397540085, + "grad_norm": 2.4926552772521973, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7257508039474487, + "num_tokens": 263169512.0, + "step": 10351 + }, + { + "epoch": 1.136832857456622, + "grad_norm": 2.383165121078491, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7185272574424744, + "num_tokens": 263193197.0, + "step": 10352 + }, + { + "epoch": 1.1369426751592357, + "grad_norm": 2.2366135120391846, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7150519490242004, + "num_tokens": 263218181.0, + "step": 10353 + }, + { + "epoch": 1.1370524928618493, + "grad_norm": 2.498828649520874, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.6974911093711853, + "num_tokens": 263241445.0, + "step": 10354 + }, + { + "epoch": 1.137162310564463, + "grad_norm": 2.1839053630828857, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.737295389175415, + "num_tokens": 263265000.0, + "step": 10355 + }, + { + "epoch": 1.1372721282670766, + "grad_norm": 2.34316349029541, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7355399131774902, + "num_tokens": 263287997.0, + "step": 10356 + }, + { + "epoch": 1.1373819459696903, + "grad_norm": 1.9771021604537964, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7316285371780396, + "num_tokens": 263318219.0, + "step": 10357 + }, + { + "epoch": 1.1374917636723039, + "grad_norm": 2.2877752780914307, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7151933908462524, + "num_tokens": 263342075.0, + "step": 10358 + }, + { + "epoch": 1.1376015813749176, + "grad_norm": 2.050196647644043, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.709854781627655, + "num_tokens": 263371617.0, + "step": 10359 + }, + { + "epoch": 1.1377113990775314, + "grad_norm": 2.1431119441986084, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7105187177658081, + "num_tokens": 263400946.0, + "step": 10360 + }, + { + "epoch": 1.137821216780145, + "grad_norm": 2.2054500579833984, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7006837129592896, + "num_tokens": 263427706.0, + "step": 10361 + }, + { + "epoch": 1.1379310344827587, + "grad_norm": 2.39544939994812, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7376886606216431, + "num_tokens": 263450983.0, + "step": 10362 + }, + { + "epoch": 1.1380408521853722, + "grad_norm": 2.6495931148529053, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7278779149055481, + "num_tokens": 263469481.0, + "step": 10363 + }, + { + "epoch": 1.138150669887986, + "grad_norm": 2.212212324142456, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.720529317855835, + "num_tokens": 263494372.0, + "step": 10364 + }, + { + "epoch": 1.1382604875905997, + "grad_norm": 2.111725091934204, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7098853588104248, + "num_tokens": 263524119.0, + "step": 10365 + }, + { + "epoch": 1.1383703052932133, + "grad_norm": 2.3457093238830566, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7159629464149475, + "num_tokens": 263547555.0, + "step": 10366 + }, + { + "epoch": 1.138480122995827, + "grad_norm": 1.944764256477356, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7172075510025024, + "num_tokens": 263580419.0, + "step": 10367 + }, + { + "epoch": 1.1385899406984406, + "grad_norm": 2.178096055984497, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7124775648117065, + "num_tokens": 263607260.0, + "step": 10368 + }, + { + "epoch": 1.1386997584010543, + "grad_norm": 2.2937021255493164, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7047457695007324, + "num_tokens": 263631167.0, + "step": 10369 + }, + { + "epoch": 1.1388095761036678, + "grad_norm": 2.2484118938446045, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.722152829170227, + "num_tokens": 263655646.0, + "step": 10370 + }, + { + "epoch": 1.1389193938062816, + "grad_norm": 2.362213134765625, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.706826388835907, + "num_tokens": 263679961.0, + "step": 10371 + }, + { + "epoch": 1.1390292115088951, + "grad_norm": 2.1967663764953613, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7334327697753906, + "num_tokens": 263705654.0, + "step": 10372 + }, + { + "epoch": 1.139139029211509, + "grad_norm": 2.0676980018615723, + "learning_rate": 1e-06, + "loss": 0.9783, + "mean_token_accuracy": 0.6968777179718018, + "num_tokens": 263734935.0, + "step": 10373 + }, + { + "epoch": 1.1392488469141226, + "grad_norm": 2.136831760406494, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.6912002563476562, + "num_tokens": 263763940.0, + "step": 10374 + }, + { + "epoch": 1.1393586646167362, + "grad_norm": 2.145810842514038, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7158709764480591, + "num_tokens": 263791823.0, + "step": 10375 + }, + { + "epoch": 1.13946848231935, + "grad_norm": 2.325507640838623, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.722064197063446, + "num_tokens": 263815761.0, + "step": 10376 + }, + { + "epoch": 1.1395783000219635, + "grad_norm": 2.3166797161102295, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7235442399978638, + "num_tokens": 263838231.0, + "step": 10377 + }, + { + "epoch": 1.1396881177245772, + "grad_norm": 2.0475571155548096, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7040781378746033, + "num_tokens": 263869090.0, + "step": 10378 + }, + { + "epoch": 1.1397979354271908, + "grad_norm": 2.190347671508789, + "learning_rate": 1e-06, + "loss": 0.7893, + "mean_token_accuracy": 0.7621946930885315, + "num_tokens": 263893563.0, + "step": 10379 + }, + { + "epoch": 1.1399077531298045, + "grad_norm": 2.7010886669158936, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7368011474609375, + "num_tokens": 263912836.0, + "step": 10380 + }, + { + "epoch": 1.1400175708324183, + "grad_norm": 2.12310791015625, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7141104340553284, + "num_tokens": 263940265.0, + "step": 10381 + }, + { + "epoch": 1.1401273885350318, + "grad_norm": 2.1978368759155273, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.715078592300415, + "num_tokens": 263966015.0, + "step": 10382 + }, + { + "epoch": 1.1402372062376456, + "grad_norm": 2.39919376373291, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7131040096282959, + "num_tokens": 263991232.0, + "step": 10383 + }, + { + "epoch": 1.140347023940259, + "grad_norm": 2.015101194381714, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7052455544471741, + "num_tokens": 264023068.0, + "step": 10384 + }, + { + "epoch": 1.1404568416428729, + "grad_norm": 2.493936538696289, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7256631851196289, + "num_tokens": 264044357.0, + "step": 10385 + }, + { + "epoch": 1.1405666593454864, + "grad_norm": 2.607717752456665, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7430177927017212, + "num_tokens": 264063315.0, + "step": 10386 + }, + { + "epoch": 1.1406764770481002, + "grad_norm": 2.224466323852539, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7129682302474976, + "num_tokens": 264089212.0, + "step": 10387 + }, + { + "epoch": 1.140786294750714, + "grad_norm": 2.186720609664917, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7120856642723083, + "num_tokens": 264115307.0, + "step": 10388 + }, + { + "epoch": 1.1408961124533274, + "grad_norm": 2.3091890811920166, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7361641526222229, + "num_tokens": 264136884.0, + "step": 10389 + }, + { + "epoch": 1.1410059301559412, + "grad_norm": 2.263611316680908, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7369388341903687, + "num_tokens": 264161466.0, + "step": 10390 + }, + { + "epoch": 1.1411157478585547, + "grad_norm": 2.019972324371338, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.706005334854126, + "num_tokens": 264194826.0, + "step": 10391 + }, + { + "epoch": 1.1412255655611685, + "grad_norm": 2.1647918224334717, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.6920422315597534, + "num_tokens": 264222247.0, + "step": 10392 + }, + { + "epoch": 1.141335383263782, + "grad_norm": 2.2012970447540283, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.6962512731552124, + "num_tokens": 264252769.0, + "step": 10393 + }, + { + "epoch": 1.1414452009663958, + "grad_norm": 2.441124677658081, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7335155010223389, + "num_tokens": 264274937.0, + "step": 10394 + }, + { + "epoch": 1.1415550186690093, + "grad_norm": 2.486827850341797, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7170617580413818, + "num_tokens": 264297458.0, + "step": 10395 + }, + { + "epoch": 1.141664836371623, + "grad_norm": 2.041210651397705, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7168452143669128, + "num_tokens": 264327437.0, + "step": 10396 + }, + { + "epoch": 1.1417746540742368, + "grad_norm": 2.4567863941192627, + "learning_rate": 1e-06, + "loss": 0.7679, + "mean_token_accuracy": 0.7622326612472534, + "num_tokens": 264348530.0, + "step": 10397 + }, + { + "epoch": 1.1418844717768504, + "grad_norm": 2.2058265209198, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7270976901054382, + "num_tokens": 264375318.0, + "step": 10398 + }, + { + "epoch": 1.1419942894794641, + "grad_norm": 2.197463035583496, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7197221517562866, + "num_tokens": 264400613.0, + "step": 10399 + }, + { + "epoch": 1.1421041071820777, + "grad_norm": 2.283184766769409, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7120904922485352, + "num_tokens": 264429881.0, + "step": 10400 + }, + { + "epoch": 1.1422139248846914, + "grad_norm": 2.5506432056427, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7126948833465576, + "num_tokens": 264456206.0, + "step": 10401 + }, + { + "epoch": 1.1423237425873052, + "grad_norm": 2.2595324516296387, + "learning_rate": 1e-06, + "loss": 1.0101, + "mean_token_accuracy": 0.6878798604011536, + "num_tokens": 264485476.0, + "step": 10402 + }, + { + "epoch": 1.1424335602899187, + "grad_norm": 2.416088104248047, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.701103925704956, + "num_tokens": 264508870.0, + "step": 10403 + }, + { + "epoch": 1.1425433779925325, + "grad_norm": 2.2994110584259033, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.706918478012085, + "num_tokens": 264534653.0, + "step": 10404 + }, + { + "epoch": 1.142653195695146, + "grad_norm": 2.390085458755493, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7156763672828674, + "num_tokens": 264556791.0, + "step": 10405 + }, + { + "epoch": 1.1427630133977598, + "grad_norm": 2.1539132595062256, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7257958650588989, + "num_tokens": 264583566.0, + "step": 10406 + }, + { + "epoch": 1.1428728311003733, + "grad_norm": 2.249370813369751, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7114212512969971, + "num_tokens": 264611459.0, + "step": 10407 + }, + { + "epoch": 1.142982648802987, + "grad_norm": 2.5891778469085693, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7138662338256836, + "num_tokens": 264634105.0, + "step": 10408 + }, + { + "epoch": 1.1430924665056006, + "grad_norm": 2.1739256381988525, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7116756439208984, + "num_tokens": 264661049.0, + "step": 10409 + }, + { + "epoch": 1.1432022842082143, + "grad_norm": 2.2515251636505127, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7228442430496216, + "num_tokens": 264686313.0, + "step": 10410 + }, + { + "epoch": 1.143312101910828, + "grad_norm": 2.5316288471221924, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7183248996734619, + "num_tokens": 264708318.0, + "step": 10411 + }, + { + "epoch": 1.1434219196134416, + "grad_norm": 2.676459550857544, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7147886157035828, + "num_tokens": 264728790.0, + "step": 10412 + }, + { + "epoch": 1.1435317373160554, + "grad_norm": 2.615414619445801, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7338931560516357, + "num_tokens": 264748428.0, + "step": 10413 + }, + { + "epoch": 1.143641555018669, + "grad_norm": 2.29589581489563, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7329890727996826, + "num_tokens": 264775026.0, + "step": 10414 + }, + { + "epoch": 1.1437513727212827, + "grad_norm": 2.2840588092803955, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7227983474731445, + "num_tokens": 264799409.0, + "step": 10415 + }, + { + "epoch": 1.1438611904238964, + "grad_norm": 2.3119282722473145, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7020213603973389, + "num_tokens": 264827493.0, + "step": 10416 + }, + { + "epoch": 1.14397100812651, + "grad_norm": 2.1313881874084473, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7281991839408875, + "num_tokens": 264855135.0, + "step": 10417 + }, + { + "epoch": 1.1440808258291237, + "grad_norm": 2.2026751041412354, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6965082883834839, + "num_tokens": 264880737.0, + "step": 10418 + }, + { + "epoch": 1.1441906435317373, + "grad_norm": 2.3620922565460205, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7382774353027344, + "num_tokens": 264903286.0, + "step": 10419 + }, + { + "epoch": 1.144300461234351, + "grad_norm": 2.0384581089019775, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7014323472976685, + "num_tokens": 264934606.0, + "step": 10420 + }, + { + "epoch": 1.1444102789369646, + "grad_norm": 2.3698103427886963, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7233725190162659, + "num_tokens": 264957842.0, + "step": 10421 + }, + { + "epoch": 1.1445200966395783, + "grad_norm": 2.0856242179870605, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7152901291847229, + "num_tokens": 264988858.0, + "step": 10422 + }, + { + "epoch": 1.1446299143421919, + "grad_norm": 2.282625436782837, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7196261882781982, + "num_tokens": 265013816.0, + "step": 10423 + }, + { + "epoch": 1.1447397320448056, + "grad_norm": 2.268371105194092, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7249928712844849, + "num_tokens": 265040123.0, + "step": 10424 + }, + { + "epoch": 1.1448495497474194, + "grad_norm": 2.482712984085083, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7205703258514404, + "num_tokens": 265062412.0, + "step": 10425 + }, + { + "epoch": 1.144959367450033, + "grad_norm": 2.5481088161468506, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7313289642333984, + "num_tokens": 265084807.0, + "step": 10426 + }, + { + "epoch": 1.1450691851526467, + "grad_norm": 2.3734476566314697, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7123067378997803, + "num_tokens": 265109668.0, + "step": 10427 + }, + { + "epoch": 1.1451790028552602, + "grad_norm": 2.5766048431396484, + "learning_rate": 1e-06, + "loss": 0.9708, + "mean_token_accuracy": 0.7061275243759155, + "num_tokens": 265129068.0, + "step": 10428 + }, + { + "epoch": 1.145288820557874, + "grad_norm": 2.246453285217285, + "learning_rate": 1e-06, + "loss": 0.8247, + "mean_token_accuracy": 0.7403028011322021, + "num_tokens": 265155127.0, + "step": 10429 + }, + { + "epoch": 1.1453986382604877, + "grad_norm": 2.2789206504821777, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.712916374206543, + "num_tokens": 265181201.0, + "step": 10430 + }, + { + "epoch": 1.1455084559631012, + "grad_norm": 2.202871561050415, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7177263498306274, + "num_tokens": 265207561.0, + "step": 10431 + }, + { + "epoch": 1.145618273665715, + "grad_norm": 2.2801382541656494, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7140688300132751, + "num_tokens": 265233432.0, + "step": 10432 + }, + { + "epoch": 1.1457280913683285, + "grad_norm": 2.461996555328369, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7268273830413818, + "num_tokens": 265254229.0, + "step": 10433 + }, + { + "epoch": 1.1458379090709423, + "grad_norm": 2.518536329269409, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7323899269104004, + "num_tokens": 265274457.0, + "step": 10434 + }, + { + "epoch": 1.1459477267735558, + "grad_norm": 2.2447659969329834, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.721916675567627, + "num_tokens": 265300279.0, + "step": 10435 + }, + { + "epoch": 1.1460575444761696, + "grad_norm": 2.1658506393432617, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7243614196777344, + "num_tokens": 265327251.0, + "step": 10436 + }, + { + "epoch": 1.1461673621787831, + "grad_norm": 2.377577543258667, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7107981443405151, + "num_tokens": 265351389.0, + "step": 10437 + }, + { + "epoch": 1.1462771798813969, + "grad_norm": 2.4035048484802246, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7217600345611572, + "num_tokens": 265374848.0, + "step": 10438 + }, + { + "epoch": 1.1463869975840106, + "grad_norm": 2.4241864681243896, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7137550115585327, + "num_tokens": 265397839.0, + "step": 10439 + }, + { + "epoch": 1.1464968152866242, + "grad_norm": 2.3930928707122803, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7282452583312988, + "num_tokens": 265420330.0, + "step": 10440 + }, + { + "epoch": 1.146606632989238, + "grad_norm": 2.3765695095062256, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7102823257446289, + "num_tokens": 265445523.0, + "step": 10441 + }, + { + "epoch": 1.1467164506918515, + "grad_norm": 2.4749438762664795, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7153030633926392, + "num_tokens": 265470104.0, + "step": 10442 + }, + { + "epoch": 1.1468262683944652, + "grad_norm": 2.371018648147583, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7060736417770386, + "num_tokens": 265493491.0, + "step": 10443 + }, + { + "epoch": 1.1469360860970788, + "grad_norm": 2.148972988128662, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.7537505626678467, + "num_tokens": 265519252.0, + "step": 10444 + }, + { + "epoch": 1.1470459037996925, + "grad_norm": 2.1631686687469482, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.736050546169281, + "num_tokens": 265543128.0, + "step": 10445 + }, + { + "epoch": 1.147155721502306, + "grad_norm": 2.191575050354004, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7089839577674866, + "num_tokens": 265569744.0, + "step": 10446 + }, + { + "epoch": 1.1472655392049198, + "grad_norm": 2.0303192138671875, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7271832227706909, + "num_tokens": 265599166.0, + "step": 10447 + }, + { + "epoch": 1.1473753569075336, + "grad_norm": 2.4517247676849365, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.6995163559913635, + "num_tokens": 265622102.0, + "step": 10448 + }, + { + "epoch": 1.147485174610147, + "grad_norm": 2.3194479942321777, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7288601398468018, + "num_tokens": 265645187.0, + "step": 10449 + }, + { + "epoch": 1.1475949923127609, + "grad_norm": 2.331639528274536, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7121204733848572, + "num_tokens": 265669826.0, + "step": 10450 + }, + { + "epoch": 1.1477048100153744, + "grad_norm": 2.2327816486358643, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7194741368293762, + "num_tokens": 265695731.0, + "step": 10451 + }, + { + "epoch": 1.1478146277179881, + "grad_norm": 2.111307382583618, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.6997218728065491, + "num_tokens": 265724015.0, + "step": 10452 + }, + { + "epoch": 1.147924445420602, + "grad_norm": 2.747119903564453, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7171823978424072, + "num_tokens": 265744440.0, + "step": 10453 + }, + { + "epoch": 1.1480342631232154, + "grad_norm": 2.4278624057769775, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7297948598861694, + "num_tokens": 265768638.0, + "step": 10454 + }, + { + "epoch": 1.1481440808258292, + "grad_norm": 2.3903145790100098, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7353435754776001, + "num_tokens": 265791308.0, + "step": 10455 + }, + { + "epoch": 1.1482538985284427, + "grad_norm": 2.2121524810791016, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7247695326805115, + "num_tokens": 265817032.0, + "step": 10456 + }, + { + "epoch": 1.1483637162310565, + "grad_norm": 2.149106025695801, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7063122987747192, + "num_tokens": 265847383.0, + "step": 10457 + }, + { + "epoch": 1.14847353393367, + "grad_norm": 2.4016544818878174, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6950300931930542, + "num_tokens": 265872064.0, + "step": 10458 + }, + { + "epoch": 1.1485833516362838, + "grad_norm": 2.1622469425201416, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6957653760910034, + "num_tokens": 265901017.0, + "step": 10459 + }, + { + "epoch": 1.1486931693388973, + "grad_norm": 2.417943239212036, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7284579873085022, + "num_tokens": 265924258.0, + "step": 10460 + }, + { + "epoch": 1.148802987041511, + "grad_norm": 2.795466899871826, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7265772819519043, + "num_tokens": 265942891.0, + "step": 10461 + }, + { + "epoch": 1.1489128047441248, + "grad_norm": 2.3853893280029297, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7275574207305908, + "num_tokens": 265963997.0, + "step": 10462 + }, + { + "epoch": 1.1490226224467384, + "grad_norm": 2.2727060317993164, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7390103340148926, + "num_tokens": 265987728.0, + "step": 10463 + }, + { + "epoch": 1.1491324401493521, + "grad_norm": 2.371403694152832, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.712127685546875, + "num_tokens": 266010437.0, + "step": 10464 + }, + { + "epoch": 1.1492422578519657, + "grad_norm": 2.4459519386291504, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7412096261978149, + "num_tokens": 266032422.0, + "step": 10465 + }, + { + "epoch": 1.1493520755545794, + "grad_norm": 2.089916944503784, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.715726375579834, + "num_tokens": 266060417.0, + "step": 10466 + }, + { + "epoch": 1.1494618932571932, + "grad_norm": 2.321624517440796, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7333301305770874, + "num_tokens": 266086538.0, + "step": 10467 + }, + { + "epoch": 1.1495717109598067, + "grad_norm": 2.2879767417907715, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7374601364135742, + "num_tokens": 266109538.0, + "step": 10468 + }, + { + "epoch": 1.1496815286624205, + "grad_norm": 2.2616796493530273, + "learning_rate": 1e-06, + "loss": 0.7631, + "mean_token_accuracy": 0.7608376145362854, + "num_tokens": 266133074.0, + "step": 10469 + }, + { + "epoch": 1.149791346365034, + "grad_norm": 2.261070966720581, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7352575063705444, + "num_tokens": 266156799.0, + "step": 10470 + }, + { + "epoch": 1.1499011640676478, + "grad_norm": 2.337320327758789, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7210745811462402, + "num_tokens": 266180906.0, + "step": 10471 + }, + { + "epoch": 1.1500109817702613, + "grad_norm": 2.3148465156555176, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7270106673240662, + "num_tokens": 266204171.0, + "step": 10472 + }, + { + "epoch": 1.150120799472875, + "grad_norm": 2.4535775184631348, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7365642786026001, + "num_tokens": 266226199.0, + "step": 10473 + }, + { + "epoch": 1.1502306171754886, + "grad_norm": 2.0478427410125732, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.717016339302063, + "num_tokens": 266256950.0, + "step": 10474 + }, + { + "epoch": 1.1503404348781023, + "grad_norm": 2.3414721488952637, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7203500270843506, + "num_tokens": 266282040.0, + "step": 10475 + }, + { + "epoch": 1.150450252580716, + "grad_norm": 2.1549229621887207, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7179704904556274, + "num_tokens": 266308534.0, + "step": 10476 + }, + { + "epoch": 1.1505600702833296, + "grad_norm": 2.1980555057525635, + "learning_rate": 1e-06, + "loss": 1.0209, + "mean_token_accuracy": 0.6871662735939026, + "num_tokens": 266336051.0, + "step": 10477 + }, + { + "epoch": 1.1506698879859434, + "grad_norm": 2.138268232345581, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7368768453598022, + "num_tokens": 266363262.0, + "step": 10478 + }, + { + "epoch": 1.150779705688557, + "grad_norm": 2.2112650871276855, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7119867205619812, + "num_tokens": 266389947.0, + "step": 10479 + }, + { + "epoch": 1.1508895233911707, + "grad_norm": 2.2046377658843994, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.709152102470398, + "num_tokens": 266416390.0, + "step": 10480 + }, + { + "epoch": 1.1509993410937844, + "grad_norm": 2.106696844100952, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7100042104721069, + "num_tokens": 266446964.0, + "step": 10481 + }, + { + "epoch": 1.151109158796398, + "grad_norm": 2.3320913314819336, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7185804843902588, + "num_tokens": 266470896.0, + "step": 10482 + }, + { + "epoch": 1.1512189764990117, + "grad_norm": 2.351349115371704, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7339421510696411, + "num_tokens": 266494362.0, + "step": 10483 + }, + { + "epoch": 1.1513287942016253, + "grad_norm": 2.3604440689086914, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7155938148498535, + "num_tokens": 266517770.0, + "step": 10484 + }, + { + "epoch": 1.151438611904239, + "grad_norm": 2.1046109199523926, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7437378764152527, + "num_tokens": 266543746.0, + "step": 10485 + }, + { + "epoch": 1.1515484296068526, + "grad_norm": 2.0330398082733154, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.7048431038856506, + "num_tokens": 266576066.0, + "step": 10486 + }, + { + "epoch": 1.1516582473094663, + "grad_norm": 2.6963624954223633, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.726224958896637, + "num_tokens": 266595386.0, + "step": 10487 + }, + { + "epoch": 1.1517680650120798, + "grad_norm": 2.3468668460845947, + "learning_rate": 1e-06, + "loss": 1.0045, + "mean_token_accuracy": 0.7001595497131348, + "num_tokens": 266620357.0, + "step": 10488 + }, + { + "epoch": 1.1518778827146936, + "grad_norm": 2.0944743156433105, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7239266633987427, + "num_tokens": 266647344.0, + "step": 10489 + }, + { + "epoch": 1.1519877004173074, + "grad_norm": 2.2583584785461426, + "learning_rate": 1e-06, + "loss": 1.0049, + "mean_token_accuracy": 0.6926483511924744, + "num_tokens": 266673972.0, + "step": 10490 + }, + { + "epoch": 1.152097518119921, + "grad_norm": 2.5247859954833984, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7360126972198486, + "num_tokens": 266693896.0, + "step": 10491 + }, + { + "epoch": 1.1522073358225347, + "grad_norm": 2.1362831592559814, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7103920578956604, + "num_tokens": 266721612.0, + "step": 10492 + }, + { + "epoch": 1.1523171535251482, + "grad_norm": 2.3885810375213623, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7154778242111206, + "num_tokens": 266744858.0, + "step": 10493 + }, + { + "epoch": 1.152426971227762, + "grad_norm": 2.2580137252807617, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7002863883972168, + "num_tokens": 266769986.0, + "step": 10494 + }, + { + "epoch": 1.1525367889303757, + "grad_norm": 2.233044385910034, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7190195322036743, + "num_tokens": 266794915.0, + "step": 10495 + }, + { + "epoch": 1.1526466066329892, + "grad_norm": 2.1283888816833496, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7090361714363098, + "num_tokens": 266825864.0, + "step": 10496 + }, + { + "epoch": 1.152756424335603, + "grad_norm": 2.2529304027557373, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7115897536277771, + "num_tokens": 266849648.0, + "step": 10497 + }, + { + "epoch": 1.1528662420382165, + "grad_norm": 2.2707738876342773, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7115825414657593, + "num_tokens": 266874394.0, + "step": 10498 + }, + { + "epoch": 1.1529760597408303, + "grad_norm": 2.489696741104126, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7225288152694702, + "num_tokens": 266896399.0, + "step": 10499 + }, + { + "epoch": 1.1530858774434438, + "grad_norm": 2.380218505859375, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7227821350097656, + "num_tokens": 266919323.0, + "step": 10500 + }, + { + "epoch": 1.1531956951460576, + "grad_norm": 2.3882551193237305, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.706186830997467, + "num_tokens": 266943391.0, + "step": 10501 + }, + { + "epoch": 1.153305512848671, + "grad_norm": 2.1310179233551025, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7150182723999023, + "num_tokens": 266971097.0, + "step": 10502 + }, + { + "epoch": 1.1534153305512849, + "grad_norm": 2.1241707801818848, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7095114588737488, + "num_tokens": 266998384.0, + "step": 10503 + }, + { + "epoch": 1.1535251482538986, + "grad_norm": 2.439620018005371, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7181811928749084, + "num_tokens": 267024508.0, + "step": 10504 + }, + { + "epoch": 1.1536349659565122, + "grad_norm": 2.242208242416382, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7183626294136047, + "num_tokens": 267049640.0, + "step": 10505 + }, + { + "epoch": 1.153744783659126, + "grad_norm": 2.285792350769043, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7433934807777405, + "num_tokens": 267072299.0, + "step": 10506 + }, + { + "epoch": 1.1538546013617395, + "grad_norm": 2.3637993335723877, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7122728824615479, + "num_tokens": 267095102.0, + "step": 10507 + }, + { + "epoch": 1.1539644190643532, + "grad_norm": 2.2642335891723633, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7202786207199097, + "num_tokens": 267119447.0, + "step": 10508 + }, + { + "epoch": 1.1540742367669667, + "grad_norm": 2.155411720275879, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7183624505996704, + "num_tokens": 267145965.0, + "step": 10509 + }, + { + "epoch": 1.1541840544695805, + "grad_norm": 2.5528697967529297, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7208508253097534, + "num_tokens": 267166280.0, + "step": 10510 + }, + { + "epoch": 1.154293872172194, + "grad_norm": 2.1163103580474854, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7322526574134827, + "num_tokens": 267194221.0, + "step": 10511 + }, + { + "epoch": 1.1544036898748078, + "grad_norm": 2.3062455654144287, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7121775150299072, + "num_tokens": 267218683.0, + "step": 10512 + }, + { + "epoch": 1.1545135075774215, + "grad_norm": 2.042238712310791, + "learning_rate": 1e-06, + "loss": 0.9902, + "mean_token_accuracy": 0.6909074187278748, + "num_tokens": 267250684.0, + "step": 10513 + }, + { + "epoch": 1.154623325280035, + "grad_norm": 2.560549020767212, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7341541051864624, + "num_tokens": 267272021.0, + "step": 10514 + }, + { + "epoch": 1.1547331429826488, + "grad_norm": 2.1746749877929688, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7158076167106628, + "num_tokens": 267298635.0, + "step": 10515 + }, + { + "epoch": 1.1548429606852624, + "grad_norm": 2.026867389678955, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7113039493560791, + "num_tokens": 267329493.0, + "step": 10516 + }, + { + "epoch": 1.1549527783878761, + "grad_norm": 2.179575204849243, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7268925309181213, + "num_tokens": 267356266.0, + "step": 10517 + }, + { + "epoch": 1.15506259609049, + "grad_norm": 2.21404767036438, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7263133525848389, + "num_tokens": 267380529.0, + "step": 10518 + }, + { + "epoch": 1.1551724137931034, + "grad_norm": 2.126913547515869, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.7022830247879028, + "num_tokens": 267412763.0, + "step": 10519 + }, + { + "epoch": 1.1552822314957172, + "grad_norm": 2.2695651054382324, + "learning_rate": 1e-06, + "loss": 0.7821, + "mean_token_accuracy": 0.7523983716964722, + "num_tokens": 267437046.0, + "step": 10520 + }, + { + "epoch": 1.1553920491983307, + "grad_norm": 2.5831117630004883, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.729425847530365, + "num_tokens": 267456836.0, + "step": 10521 + }, + { + "epoch": 1.1555018669009445, + "grad_norm": 2.13124680519104, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7344740033149719, + "num_tokens": 267483585.0, + "step": 10522 + }, + { + "epoch": 1.155611684603558, + "grad_norm": 2.0685176849365234, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7118648886680603, + "num_tokens": 267513498.0, + "step": 10523 + }, + { + "epoch": 1.1557215023061718, + "grad_norm": 2.3439481258392334, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7111239433288574, + "num_tokens": 267538760.0, + "step": 10524 + }, + { + "epoch": 1.1558313200087853, + "grad_norm": 2.296921968460083, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7177689075469971, + "num_tokens": 267562946.0, + "step": 10525 + }, + { + "epoch": 1.155941137711399, + "grad_norm": 2.059649705886841, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7148280143737793, + "num_tokens": 267592564.0, + "step": 10526 + }, + { + "epoch": 1.1560509554140128, + "grad_norm": 2.376027822494507, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.6992892026901245, + "num_tokens": 267615388.0, + "step": 10527 + }, + { + "epoch": 1.1561607731166264, + "grad_norm": 2.77180552482605, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7338778972625732, + "num_tokens": 267634209.0, + "step": 10528 + }, + { + "epoch": 1.15627059081924, + "grad_norm": 2.309067964553833, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7048958539962769, + "num_tokens": 267658025.0, + "step": 10529 + }, + { + "epoch": 1.1563804085218536, + "grad_norm": 2.3720293045043945, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7258774042129517, + "num_tokens": 267680777.0, + "step": 10530 + }, + { + "epoch": 1.1564902262244674, + "grad_norm": 2.3965003490448, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7199368476867676, + "num_tokens": 267703883.0, + "step": 10531 + }, + { + "epoch": 1.1566000439270812, + "grad_norm": 2.316643238067627, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7034018039703369, + "num_tokens": 267728475.0, + "step": 10532 + }, + { + "epoch": 1.1567098616296947, + "grad_norm": 2.35017466545105, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7354892492294312, + "num_tokens": 267751966.0, + "step": 10533 + }, + { + "epoch": 1.1568196793323084, + "grad_norm": 2.138746738433838, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7051947116851807, + "num_tokens": 267780464.0, + "step": 10534 + }, + { + "epoch": 1.156929497034922, + "grad_norm": 2.2671289443969727, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7391467094421387, + "num_tokens": 267803867.0, + "step": 10535 + }, + { + "epoch": 1.1570393147375357, + "grad_norm": 2.1697380542755127, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7071535587310791, + "num_tokens": 267832525.0, + "step": 10536 + }, + { + "epoch": 1.1571491324401493, + "grad_norm": 2.308281183242798, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7148401141166687, + "num_tokens": 267857533.0, + "step": 10537 + }, + { + "epoch": 1.157258950142763, + "grad_norm": 2.3660852909088135, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.7548487186431885, + "num_tokens": 267879071.0, + "step": 10538 + }, + { + "epoch": 1.1573687678453766, + "grad_norm": 2.4069104194641113, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.727483332157135, + "num_tokens": 267902550.0, + "step": 10539 + }, + { + "epoch": 1.1574785855479903, + "grad_norm": 2.457090377807617, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7402754426002502, + "num_tokens": 267924092.0, + "step": 10540 + }, + { + "epoch": 1.157588403250604, + "grad_norm": 2.4536590576171875, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7398417592048645, + "num_tokens": 267944807.0, + "step": 10541 + }, + { + "epoch": 1.1576982209532176, + "grad_norm": 2.4171833992004395, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7145218253135681, + "num_tokens": 267967884.0, + "step": 10542 + }, + { + "epoch": 1.1578080386558314, + "grad_norm": 2.41652250289917, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7325031757354736, + "num_tokens": 267992374.0, + "step": 10543 + }, + { + "epoch": 1.157917856358445, + "grad_norm": 2.105299234390259, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.7083060145378113, + "num_tokens": 268022548.0, + "step": 10544 + }, + { + "epoch": 1.1580276740610587, + "grad_norm": 2.11720871925354, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7002996206283569, + "num_tokens": 268050043.0, + "step": 10545 + }, + { + "epoch": 1.1581374917636724, + "grad_norm": 2.1059417724609375, + "learning_rate": 1e-06, + "loss": 0.8056, + "mean_token_accuracy": 0.7459326982498169, + "num_tokens": 268076663.0, + "step": 10546 + }, + { + "epoch": 1.158247309466286, + "grad_norm": 2.2192437648773193, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.7209678888320923, + "num_tokens": 268104763.0, + "step": 10547 + }, + { + "epoch": 1.1583571271688997, + "grad_norm": 2.291200637817383, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7096775770187378, + "num_tokens": 268129500.0, + "step": 10548 + }, + { + "epoch": 1.1584669448715132, + "grad_norm": 2.2102882862091064, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7208924293518066, + "num_tokens": 268156171.0, + "step": 10549 + }, + { + "epoch": 1.158576762574127, + "grad_norm": 2.266270637512207, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7170847058296204, + "num_tokens": 268182038.0, + "step": 10550 + }, + { + "epoch": 1.1586865802767405, + "grad_norm": 2.370128870010376, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7280479669570923, + "num_tokens": 268206865.0, + "step": 10551 + }, + { + "epoch": 1.1587963979793543, + "grad_norm": 2.3267505168914795, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7412348985671997, + "num_tokens": 268230322.0, + "step": 10552 + }, + { + "epoch": 1.1589062156819678, + "grad_norm": 2.1863958835601807, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.715364933013916, + "num_tokens": 268257637.0, + "step": 10553 + }, + { + "epoch": 1.1590160333845816, + "grad_norm": 2.2111008167266846, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7226005792617798, + "num_tokens": 268282918.0, + "step": 10554 + }, + { + "epoch": 1.1591258510871953, + "grad_norm": 2.558978796005249, + "learning_rate": 1e-06, + "loss": 0.8208, + "mean_token_accuracy": 0.7380090951919556, + "num_tokens": 268302288.0, + "step": 10555 + }, + { + "epoch": 1.1592356687898089, + "grad_norm": 2.1470417976379395, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7298542261123657, + "num_tokens": 268330298.0, + "step": 10556 + }, + { + "epoch": 1.1593454864924226, + "grad_norm": 2.2850966453552246, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7114711403846741, + "num_tokens": 268354893.0, + "step": 10557 + }, + { + "epoch": 1.1594553041950362, + "grad_norm": 2.142768383026123, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.6954163312911987, + "num_tokens": 268381524.0, + "step": 10558 + }, + { + "epoch": 1.15956512189765, + "grad_norm": 2.408777952194214, + "learning_rate": 1e-06, + "loss": 0.8201, + "mean_token_accuracy": 0.7458956241607666, + "num_tokens": 268404454.0, + "step": 10559 + }, + { + "epoch": 1.1596749396002635, + "grad_norm": 2.518235921859741, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7264137864112854, + "num_tokens": 268426890.0, + "step": 10560 + }, + { + "epoch": 1.1597847573028772, + "grad_norm": 2.521883249282837, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7439597249031067, + "num_tokens": 268449000.0, + "step": 10561 + }, + { + "epoch": 1.159894575005491, + "grad_norm": 2.405702590942383, + "learning_rate": 1e-06, + "loss": 0.803, + "mean_token_accuracy": 0.7444767951965332, + "num_tokens": 268472193.0, + "step": 10562 + }, + { + "epoch": 1.1600043927081045, + "grad_norm": 2.3711135387420654, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7224132418632507, + "num_tokens": 268495644.0, + "step": 10563 + }, + { + "epoch": 1.1601142104107183, + "grad_norm": 2.223102569580078, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7064778804779053, + "num_tokens": 268521500.0, + "step": 10564 + }, + { + "epoch": 1.1602240281133318, + "grad_norm": 2.187533378601074, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7201453447341919, + "num_tokens": 268548729.0, + "step": 10565 + }, + { + "epoch": 1.1603338458159456, + "grad_norm": 2.1989011764526367, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6928489208221436, + "num_tokens": 268576644.0, + "step": 10566 + }, + { + "epoch": 1.160443663518559, + "grad_norm": 2.184865713119507, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7217410206794739, + "num_tokens": 268605516.0, + "step": 10567 + }, + { + "epoch": 1.1605534812211729, + "grad_norm": 2.6428098678588867, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7200173139572144, + "num_tokens": 268625839.0, + "step": 10568 + }, + { + "epoch": 1.1606632989237866, + "grad_norm": 2.157254934310913, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7193798422813416, + "num_tokens": 268652849.0, + "step": 10569 + }, + { + "epoch": 1.1607731166264001, + "grad_norm": 2.483337163925171, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7255158424377441, + "num_tokens": 268672819.0, + "step": 10570 + }, + { + "epoch": 1.160882934329014, + "grad_norm": 2.11452579498291, + "learning_rate": 1e-06, + "loss": 0.8036, + "mean_token_accuracy": 0.7470844388008118, + "num_tokens": 268699739.0, + "step": 10571 + }, + { + "epoch": 1.1609927520316274, + "grad_norm": 2.3058226108551025, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.6943819522857666, + "num_tokens": 268724278.0, + "step": 10572 + }, + { + "epoch": 1.1611025697342412, + "grad_norm": 2.1941051483154297, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7213611602783203, + "num_tokens": 268750562.0, + "step": 10573 + }, + { + "epoch": 1.1612123874368547, + "grad_norm": 2.342653512954712, + "learning_rate": 1e-06, + "loss": 0.7855, + "mean_token_accuracy": 0.7416396141052246, + "num_tokens": 268772766.0, + "step": 10574 + }, + { + "epoch": 1.1613222051394685, + "grad_norm": 2.149003505706787, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7295899987220764, + "num_tokens": 268799821.0, + "step": 10575 + }, + { + "epoch": 1.161432022842082, + "grad_norm": 1.9529743194580078, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7122646570205688, + "num_tokens": 268833516.0, + "step": 10576 + }, + { + "epoch": 1.1615418405446958, + "grad_norm": 2.2862303256988525, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7411470413208008, + "num_tokens": 268856506.0, + "step": 10577 + }, + { + "epoch": 1.1616516582473095, + "grad_norm": 2.4362268447875977, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7451300621032715, + "num_tokens": 268878547.0, + "step": 10578 + }, + { + "epoch": 1.161761475949923, + "grad_norm": 2.0673837661743164, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7055705785751343, + "num_tokens": 268908962.0, + "step": 10579 + }, + { + "epoch": 1.1618712936525368, + "grad_norm": 2.2364306449890137, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7146205306053162, + "num_tokens": 268936260.0, + "step": 10580 + }, + { + "epoch": 1.1619811113551504, + "grad_norm": 2.0458762645721436, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7089972496032715, + "num_tokens": 268967123.0, + "step": 10581 + }, + { + "epoch": 1.1620909290577641, + "grad_norm": 2.1866543292999268, + "learning_rate": 1e-06, + "loss": 0.8859, + "mean_token_accuracy": 0.7219207286834717, + "num_tokens": 268991904.0, + "step": 10582 + }, + { + "epoch": 1.1622007467603779, + "grad_norm": 2.7768044471740723, + "learning_rate": 1e-06, + "loss": 0.7367, + "mean_token_accuracy": 0.7551781535148621, + "num_tokens": 269008408.0, + "step": 10583 + }, + { + "epoch": 1.1623105644629914, + "grad_norm": 2.519247055053711, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7263990640640259, + "num_tokens": 269030442.0, + "step": 10584 + }, + { + "epoch": 1.1624203821656052, + "grad_norm": 2.4979331493377686, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7389618158340454, + "num_tokens": 269051802.0, + "step": 10585 + }, + { + "epoch": 1.1625301998682187, + "grad_norm": 2.2864511013031006, + "learning_rate": 1e-06, + "loss": 1.0287, + "mean_token_accuracy": 0.6898103952407837, + "num_tokens": 269079810.0, + "step": 10586 + }, + { + "epoch": 1.1626400175708325, + "grad_norm": 2.2736995220184326, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7360056638717651, + "num_tokens": 269102433.0, + "step": 10587 + }, + { + "epoch": 1.162749835273446, + "grad_norm": 2.2525553703308105, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7117525339126587, + "num_tokens": 269129515.0, + "step": 10588 + }, + { + "epoch": 1.1628596529760598, + "grad_norm": 2.1562018394470215, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7305340766906738, + "num_tokens": 269157005.0, + "step": 10589 + }, + { + "epoch": 1.1629694706786733, + "grad_norm": 2.3033201694488525, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7010315656661987, + "num_tokens": 269181671.0, + "step": 10590 + }, + { + "epoch": 1.163079288381287, + "grad_norm": 2.260071039199829, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7082628011703491, + "num_tokens": 269206615.0, + "step": 10591 + }, + { + "epoch": 1.1631891060839008, + "grad_norm": 2.2681005001068115, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7214535474777222, + "num_tokens": 269231093.0, + "step": 10592 + }, + { + "epoch": 1.1632989237865143, + "grad_norm": 2.1482784748077393, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7210279703140259, + "num_tokens": 269258134.0, + "step": 10593 + }, + { + "epoch": 1.163408741489128, + "grad_norm": 2.2933454513549805, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7330412864685059, + "num_tokens": 269281079.0, + "step": 10594 + }, + { + "epoch": 1.1635185591917416, + "grad_norm": 2.365344285964966, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7349255681037903, + "num_tokens": 269302786.0, + "step": 10595 + }, + { + "epoch": 1.1636283768943554, + "grad_norm": 2.252701759338379, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7030718326568604, + "num_tokens": 269327230.0, + "step": 10596 + }, + { + "epoch": 1.1637381945969691, + "grad_norm": 2.602388858795166, + "learning_rate": 1e-06, + "loss": 0.793, + "mean_token_accuracy": 0.7464535236358643, + "num_tokens": 269346298.0, + "step": 10597 + }, + { + "epoch": 1.1638480122995827, + "grad_norm": 2.41548752784729, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7124069929122925, + "num_tokens": 269367738.0, + "step": 10598 + }, + { + "epoch": 1.1639578300021964, + "grad_norm": 2.412217855453491, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7164106369018555, + "num_tokens": 269391027.0, + "step": 10599 + }, + { + "epoch": 1.16406764770481, + "grad_norm": 2.491027355194092, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7344194650650024, + "num_tokens": 269412991.0, + "step": 10600 + }, + { + "epoch": 1.1641774654074237, + "grad_norm": 2.477750539779663, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7071506977081299, + "num_tokens": 269435272.0, + "step": 10601 + }, + { + "epoch": 1.1642872831100373, + "grad_norm": 2.2165932655334473, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7438516020774841, + "num_tokens": 269460946.0, + "step": 10602 + }, + { + "epoch": 1.164397100812651, + "grad_norm": 2.116147041320801, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7061876058578491, + "num_tokens": 269490225.0, + "step": 10603 + }, + { + "epoch": 1.1645069185152646, + "grad_norm": 2.113041400909424, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7212525010108948, + "num_tokens": 269518411.0, + "step": 10604 + }, + { + "epoch": 1.1646167362178783, + "grad_norm": 2.188344955444336, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7259781360626221, + "num_tokens": 269543999.0, + "step": 10605 + }, + { + "epoch": 1.164726553920492, + "grad_norm": 1.9992365837097168, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.7509486079216003, + "num_tokens": 269570774.0, + "step": 10606 + }, + { + "epoch": 1.1648363716231056, + "grad_norm": 2.586714506149292, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7401024103164673, + "num_tokens": 269591274.0, + "step": 10607 + }, + { + "epoch": 1.1649461893257194, + "grad_norm": 2.1656289100646973, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.688270092010498, + "num_tokens": 269619200.0, + "step": 10608 + }, + { + "epoch": 1.165056007028333, + "grad_norm": 2.2289345264434814, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7046669721603394, + "num_tokens": 269644207.0, + "step": 10609 + }, + { + "epoch": 1.1651658247309467, + "grad_norm": 2.401472806930542, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.726504385471344, + "num_tokens": 269666291.0, + "step": 10610 + }, + { + "epoch": 1.1652756424335604, + "grad_norm": 2.621699810028076, + "learning_rate": 1e-06, + "loss": 0.7923, + "mean_token_accuracy": 0.7452734708786011, + "num_tokens": 269685904.0, + "step": 10611 + }, + { + "epoch": 1.165385460136174, + "grad_norm": 3.342015266418457, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.7188813090324402, + "num_tokens": 269710729.0, + "step": 10612 + }, + { + "epoch": 1.1654952778387877, + "grad_norm": 2.57716965675354, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7400351166725159, + "num_tokens": 269730443.0, + "step": 10613 + }, + { + "epoch": 1.1656050955414012, + "grad_norm": 2.160093069076538, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7355071306228638, + "num_tokens": 269756794.0, + "step": 10614 + }, + { + "epoch": 1.165714913244015, + "grad_norm": 2.3896405696868896, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7273610830307007, + "num_tokens": 269779637.0, + "step": 10615 + }, + { + "epoch": 1.1658247309466285, + "grad_norm": 2.329662799835205, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7391284704208374, + "num_tokens": 269803106.0, + "step": 10616 + }, + { + "epoch": 1.1659345486492423, + "grad_norm": 2.420490026473999, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7343934774398804, + "num_tokens": 269824046.0, + "step": 10617 + }, + { + "epoch": 1.1660443663518558, + "grad_norm": 2.5887014865875244, + "learning_rate": 1e-06, + "loss": 0.8636, + "mean_token_accuracy": 0.731399416923523, + "num_tokens": 269845002.0, + "step": 10618 + }, + { + "epoch": 1.1661541840544696, + "grad_norm": 2.2158920764923096, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.7410978078842163, + "num_tokens": 269871866.0, + "step": 10619 + }, + { + "epoch": 1.1662640017570833, + "grad_norm": 2.6173253059387207, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7062091827392578, + "num_tokens": 269894339.0, + "step": 10620 + }, + { + "epoch": 1.1663738194596969, + "grad_norm": 2.233464002609253, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7203459739685059, + "num_tokens": 269919976.0, + "step": 10621 + }, + { + "epoch": 1.1664836371623106, + "grad_norm": 2.075112819671631, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7291640043258667, + "num_tokens": 269948332.0, + "step": 10622 + }, + { + "epoch": 1.1665934548649242, + "grad_norm": 2.3031585216522217, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7364578247070312, + "num_tokens": 269971780.0, + "step": 10623 + }, + { + "epoch": 1.166703272567538, + "grad_norm": 2.6043012142181396, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7199814319610596, + "num_tokens": 269997126.0, + "step": 10624 + }, + { + "epoch": 1.1668130902701515, + "grad_norm": 2.32377290725708, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7307229042053223, + "num_tokens": 270020681.0, + "step": 10625 + }, + { + "epoch": 1.1669229079727652, + "grad_norm": 2.2750983238220215, + "learning_rate": 1e-06, + "loss": 0.9939, + "mean_token_accuracy": 0.6965814232826233, + "num_tokens": 270046629.0, + "step": 10626 + }, + { + "epoch": 1.167032725675379, + "grad_norm": 2.1969292163848877, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7088116407394409, + "num_tokens": 270074916.0, + "step": 10627 + }, + { + "epoch": 1.1671425433779925, + "grad_norm": 2.098900556564331, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7243971824645996, + "num_tokens": 270103256.0, + "step": 10628 + }, + { + "epoch": 1.1672523610806063, + "grad_norm": 2.08719801902771, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.709542989730835, + "num_tokens": 270132327.0, + "step": 10629 + }, + { + "epoch": 1.1673621787832198, + "grad_norm": 2.6866512298583984, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7187120914459229, + "num_tokens": 270152334.0, + "step": 10630 + }, + { + "epoch": 1.1674719964858336, + "grad_norm": 2.23336124420166, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.707450270652771, + "num_tokens": 270179051.0, + "step": 10631 + }, + { + "epoch": 1.167581814188447, + "grad_norm": 2.210726499557495, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7257606387138367, + "num_tokens": 270205331.0, + "step": 10632 + }, + { + "epoch": 1.1676916318910608, + "grad_norm": 2.241283416748047, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.725529670715332, + "num_tokens": 270229972.0, + "step": 10633 + }, + { + "epoch": 1.1678014495936746, + "grad_norm": 2.5024049282073975, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.740156888961792, + "num_tokens": 270250441.0, + "step": 10634 + }, + { + "epoch": 1.1679112672962881, + "grad_norm": 2.47152042388916, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7209566831588745, + "num_tokens": 270272272.0, + "step": 10635 + }, + { + "epoch": 1.168021084998902, + "grad_norm": 2.223184585571289, + "learning_rate": 1e-06, + "loss": 0.985, + "mean_token_accuracy": 0.7036023736000061, + "num_tokens": 270299504.0, + "step": 10636 + }, + { + "epoch": 1.1681309027015154, + "grad_norm": 2.0999162197113037, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7221632599830627, + "num_tokens": 270327734.0, + "step": 10637 + }, + { + "epoch": 1.1682407204041292, + "grad_norm": 2.0806210041046143, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7144439816474915, + "num_tokens": 270355617.0, + "step": 10638 + }, + { + "epoch": 1.1683505381067427, + "grad_norm": 2.6615548133850098, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7503101825714111, + "num_tokens": 270374193.0, + "step": 10639 + }, + { + "epoch": 1.1684603558093565, + "grad_norm": 2.3432581424713135, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7031923532485962, + "num_tokens": 270399047.0, + "step": 10640 + }, + { + "epoch": 1.16857017351197, + "grad_norm": 2.306579351425171, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7093355655670166, + "num_tokens": 270425513.0, + "step": 10641 + }, + { + "epoch": 1.1686799912145838, + "grad_norm": 2.6396799087524414, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7164843678474426, + "num_tokens": 270444489.0, + "step": 10642 + }, + { + "epoch": 1.1687898089171975, + "grad_norm": 2.058743715286255, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7094835042953491, + "num_tokens": 270475348.0, + "step": 10643 + }, + { + "epoch": 1.168899626619811, + "grad_norm": 2.4908859729766846, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7249429225921631, + "num_tokens": 270499425.0, + "step": 10644 + }, + { + "epoch": 1.1690094443224248, + "grad_norm": 2.193460464477539, + "learning_rate": 1e-06, + "loss": 1.0571, + "mean_token_accuracy": 0.6842097043991089, + "num_tokens": 270529895.0, + "step": 10645 + }, + { + "epoch": 1.1691192620250384, + "grad_norm": 2.4857125282287598, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7228870391845703, + "num_tokens": 270550767.0, + "step": 10646 + }, + { + "epoch": 1.169229079727652, + "grad_norm": 2.013720989227295, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.6986567378044128, + "num_tokens": 270582280.0, + "step": 10647 + }, + { + "epoch": 1.1693388974302659, + "grad_norm": 2.3017325401306152, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7234073281288147, + "num_tokens": 270607325.0, + "step": 10648 + }, + { + "epoch": 1.1694487151328794, + "grad_norm": 2.399336338043213, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7150804996490479, + "num_tokens": 270631053.0, + "step": 10649 + }, + { + "epoch": 1.1695585328354932, + "grad_norm": 2.658797025680542, + "learning_rate": 1e-06, + "loss": 0.7718, + "mean_token_accuracy": 0.7549895644187927, + "num_tokens": 270649976.0, + "step": 10650 + }, + { + "epoch": 1.1696683505381067, + "grad_norm": 2.085066080093384, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.711615800857544, + "num_tokens": 270680165.0, + "step": 10651 + }, + { + "epoch": 1.1697781682407205, + "grad_norm": 2.0292060375213623, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7293853759765625, + "num_tokens": 270710579.0, + "step": 10652 + }, + { + "epoch": 1.169887985943334, + "grad_norm": 1.875370740890503, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7308446168899536, + "num_tokens": 270744477.0, + "step": 10653 + }, + { + "epoch": 1.1699978036459477, + "grad_norm": 2.2733571529388428, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7177010178565979, + "num_tokens": 270771470.0, + "step": 10654 + }, + { + "epoch": 1.1701076213485613, + "grad_norm": 2.435819625854492, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7402289509773254, + "num_tokens": 270793317.0, + "step": 10655 + }, + { + "epoch": 1.170217439051175, + "grad_norm": 2.189666986465454, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7159773707389832, + "num_tokens": 270820609.0, + "step": 10656 + }, + { + "epoch": 1.1703272567537888, + "grad_norm": 2.2457151412963867, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7299628257751465, + "num_tokens": 270844701.0, + "step": 10657 + }, + { + "epoch": 1.1704370744564023, + "grad_norm": 2.372117757797241, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.735821008682251, + "num_tokens": 270866667.0, + "step": 10658 + }, + { + "epoch": 1.170546892159016, + "grad_norm": 2.6528024673461914, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7054332494735718, + "num_tokens": 270888861.0, + "step": 10659 + }, + { + "epoch": 1.1706567098616296, + "grad_norm": 2.5388197898864746, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7290345430374146, + "num_tokens": 270908760.0, + "step": 10660 + }, + { + "epoch": 1.1707665275642434, + "grad_norm": 2.236919641494751, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.727483868598938, + "num_tokens": 270934037.0, + "step": 10661 + }, + { + "epoch": 1.1708763452668571, + "grad_norm": 2.212167501449585, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7213329076766968, + "num_tokens": 270958856.0, + "step": 10662 + }, + { + "epoch": 1.1709861629694707, + "grad_norm": 2.2289907932281494, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7126650810241699, + "num_tokens": 270984086.0, + "step": 10663 + }, + { + "epoch": 1.1710959806720844, + "grad_norm": 2.728144884109497, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7234432697296143, + "num_tokens": 271003778.0, + "step": 10664 + }, + { + "epoch": 1.171205798374698, + "grad_norm": 2.4184367656707764, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7282634377479553, + "num_tokens": 271024952.0, + "step": 10665 + }, + { + "epoch": 1.1713156160773117, + "grad_norm": 2.136261224746704, + "learning_rate": 1e-06, + "loss": 1.0405, + "mean_token_accuracy": 0.6857339143753052, + "num_tokens": 271053786.0, + "step": 10666 + }, + { + "epoch": 1.1714254337799253, + "grad_norm": 2.172417163848877, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6994503140449524, + "num_tokens": 271085033.0, + "step": 10667 + }, + { + "epoch": 1.171535251482539, + "grad_norm": 2.1996822357177734, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7075618505477905, + "num_tokens": 271112350.0, + "step": 10668 + }, + { + "epoch": 1.1716450691851525, + "grad_norm": 2.1923751831054688, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7324333786964417, + "num_tokens": 271138813.0, + "step": 10669 + }, + { + "epoch": 1.1717548868877663, + "grad_norm": 2.2254137992858887, + "learning_rate": 1e-06, + "loss": 0.8385, + "mean_token_accuracy": 0.7317039966583252, + "num_tokens": 271162593.0, + "step": 10670 + }, + { + "epoch": 1.17186470459038, + "grad_norm": 2.536555051803589, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7423990964889526, + "num_tokens": 271183164.0, + "step": 10671 + }, + { + "epoch": 1.1719745222929936, + "grad_norm": 2.3398709297180176, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.6978464126586914, + "num_tokens": 271209660.0, + "step": 10672 + }, + { + "epoch": 1.1720843399956073, + "grad_norm": 2.186887741088867, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7053183317184448, + "num_tokens": 271237406.0, + "step": 10673 + }, + { + "epoch": 1.1721941576982209, + "grad_norm": 1.9698213338851929, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7280527353286743, + "num_tokens": 271267942.0, + "step": 10674 + }, + { + "epoch": 1.1723039754008346, + "grad_norm": 2.262953281402588, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.710023820400238, + "num_tokens": 271292309.0, + "step": 10675 + }, + { + "epoch": 1.1724137931034484, + "grad_norm": 2.423203706741333, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7514759302139282, + "num_tokens": 271312474.0, + "step": 10676 + }, + { + "epoch": 1.172523610806062, + "grad_norm": 2.053569793701172, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7145912051200867, + "num_tokens": 271344514.0, + "step": 10677 + }, + { + "epoch": 1.1726334285086757, + "grad_norm": 2.4404945373535156, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7081049084663391, + "num_tokens": 271368047.0, + "step": 10678 + }, + { + "epoch": 1.1727432462112892, + "grad_norm": 2.3800606727600098, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7193874716758728, + "num_tokens": 271391911.0, + "step": 10679 + }, + { + "epoch": 1.172853063913903, + "grad_norm": 2.1649768352508545, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7040330767631531, + "num_tokens": 271421109.0, + "step": 10680 + }, + { + "epoch": 1.1729628816165165, + "grad_norm": 2.4301249980926514, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.732518196105957, + "num_tokens": 271445958.0, + "step": 10681 + }, + { + "epoch": 1.1730726993191303, + "grad_norm": 2.5323355197906494, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7350431680679321, + "num_tokens": 271465127.0, + "step": 10682 + }, + { + "epoch": 1.1731825170217438, + "grad_norm": 2.1443850994110107, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7192617654800415, + "num_tokens": 271491672.0, + "step": 10683 + }, + { + "epoch": 1.1732923347243576, + "grad_norm": 2.26639986038208, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7110356688499451, + "num_tokens": 271517071.0, + "step": 10684 + }, + { + "epoch": 1.1734021524269713, + "grad_norm": 2.295893430709839, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7203461527824402, + "num_tokens": 271543615.0, + "step": 10685 + }, + { + "epoch": 1.1735119701295849, + "grad_norm": 2.3989412784576416, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7055915594100952, + "num_tokens": 271567799.0, + "step": 10686 + }, + { + "epoch": 1.1736217878321986, + "grad_norm": 2.3324155807495117, + "learning_rate": 1e-06, + "loss": 0.8784, + "mean_token_accuracy": 0.7280457019805908, + "num_tokens": 271589827.0, + "step": 10687 + }, + { + "epoch": 1.1737316055348122, + "grad_norm": 2.1692121028900146, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.696916937828064, + "num_tokens": 271616629.0, + "step": 10688 + }, + { + "epoch": 1.173841423237426, + "grad_norm": 2.1464309692382812, + "learning_rate": 1e-06, + "loss": 0.778, + "mean_token_accuracy": 0.7511350512504578, + "num_tokens": 271641212.0, + "step": 10689 + }, + { + "epoch": 1.1739512409400394, + "grad_norm": 2.37009596824646, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7312499284744263, + "num_tokens": 271664361.0, + "step": 10690 + }, + { + "epoch": 1.1740610586426532, + "grad_norm": 2.0660338401794434, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.706626296043396, + "num_tokens": 271692203.0, + "step": 10691 + }, + { + "epoch": 1.1741708763452667, + "grad_norm": 2.3254520893096924, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7366472482681274, + "num_tokens": 271716265.0, + "step": 10692 + }, + { + "epoch": 1.1742806940478805, + "grad_norm": 2.4606173038482666, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7404423952102661, + "num_tokens": 271738616.0, + "step": 10693 + }, + { + "epoch": 1.1743905117504942, + "grad_norm": 2.2510499954223633, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7116687297821045, + "num_tokens": 271764744.0, + "step": 10694 + }, + { + "epoch": 1.1745003294531078, + "grad_norm": 2.2687888145446777, + "learning_rate": 1e-06, + "loss": 0.9682, + "mean_token_accuracy": 0.7034578919410706, + "num_tokens": 271792064.0, + "step": 10695 + }, + { + "epoch": 1.1746101471557215, + "grad_norm": 2.425647020339966, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7066426277160645, + "num_tokens": 271816113.0, + "step": 10696 + }, + { + "epoch": 1.174719964858335, + "grad_norm": 1.962361216545105, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6925821900367737, + "num_tokens": 271847704.0, + "step": 10697 + }, + { + "epoch": 1.1748297825609488, + "grad_norm": 1.9154374599456787, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7167643308639526, + "num_tokens": 271882891.0, + "step": 10698 + }, + { + "epoch": 1.1749396002635626, + "grad_norm": 2.0455634593963623, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7154749631881714, + "num_tokens": 271914569.0, + "step": 10699 + }, + { + "epoch": 1.1750494179661761, + "grad_norm": 1.9973883628845215, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7044505476951599, + "num_tokens": 271944430.0, + "step": 10700 + }, + { + "epoch": 1.1751592356687899, + "grad_norm": 2.4364333152770996, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7286491394042969, + "num_tokens": 271967043.0, + "step": 10701 + }, + { + "epoch": 1.1752690533714034, + "grad_norm": 2.3756279945373535, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.7468163967132568, + "num_tokens": 271988436.0, + "step": 10702 + }, + { + "epoch": 1.1753788710740172, + "grad_norm": 2.115286111831665, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7289661765098572, + "num_tokens": 272015092.0, + "step": 10703 + }, + { + "epoch": 1.1754886887766307, + "grad_norm": 2.1835315227508545, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.6979395747184753, + "num_tokens": 272042173.0, + "step": 10704 + }, + { + "epoch": 1.1755985064792445, + "grad_norm": 2.4567208290100098, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7245587110519409, + "num_tokens": 272065325.0, + "step": 10705 + }, + { + "epoch": 1.175708324181858, + "grad_norm": 2.1469719409942627, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7381357550621033, + "num_tokens": 272092687.0, + "step": 10706 + }, + { + "epoch": 1.1758181418844718, + "grad_norm": 2.1330246925354004, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7201045751571655, + "num_tokens": 272120330.0, + "step": 10707 + }, + { + "epoch": 1.1759279595870855, + "grad_norm": 2.3292412757873535, + "learning_rate": 1e-06, + "loss": 0.7896, + "mean_token_accuracy": 0.7437183856964111, + "num_tokens": 272142028.0, + "step": 10708 + }, + { + "epoch": 1.176037777289699, + "grad_norm": 2.612459659576416, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7226623296737671, + "num_tokens": 272162002.0, + "step": 10709 + }, + { + "epoch": 1.1761475949923128, + "grad_norm": 2.1231818199157715, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.7021150588989258, + "num_tokens": 272190392.0, + "step": 10710 + }, + { + "epoch": 1.1762574126949263, + "grad_norm": 1.9220824241638184, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7244183421134949, + "num_tokens": 272220965.0, + "step": 10711 + }, + { + "epoch": 1.17636723039754, + "grad_norm": 2.5602097511291504, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7037907838821411, + "num_tokens": 272241663.0, + "step": 10712 + }, + { + "epoch": 1.1764770481001539, + "grad_norm": 2.1067750453948975, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7280372977256775, + "num_tokens": 272269969.0, + "step": 10713 + }, + { + "epoch": 1.1765868658027674, + "grad_norm": 2.8151156902313232, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7253932952880859, + "num_tokens": 272287736.0, + "step": 10714 + }, + { + "epoch": 1.1766966835053811, + "grad_norm": 2.036529779434204, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7093113660812378, + "num_tokens": 272316927.0, + "step": 10715 + }, + { + "epoch": 1.1768065012079947, + "grad_norm": 2.336989402770996, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7194666862487793, + "num_tokens": 272341816.0, + "step": 10716 + }, + { + "epoch": 1.1769163189106084, + "grad_norm": 2.236832618713379, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7162243127822876, + "num_tokens": 272367253.0, + "step": 10717 + }, + { + "epoch": 1.177026136613222, + "grad_norm": 2.615922689437866, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7261407375335693, + "num_tokens": 272387830.0, + "step": 10718 + }, + { + "epoch": 1.1771359543158357, + "grad_norm": 2.3939719200134277, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7147829532623291, + "num_tokens": 272411083.0, + "step": 10719 + }, + { + "epoch": 1.1772457720184493, + "grad_norm": 2.266103982925415, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7063698172569275, + "num_tokens": 272436713.0, + "step": 10720 + }, + { + "epoch": 1.177355589721063, + "grad_norm": 2.362924814224243, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7228935956954956, + "num_tokens": 272458710.0, + "step": 10721 + }, + { + "epoch": 1.1774654074236768, + "grad_norm": 2.310824155807495, + "learning_rate": 1e-06, + "loss": 0.9815, + "mean_token_accuracy": 0.6974176168441772, + "num_tokens": 272484786.0, + "step": 10722 + }, + { + "epoch": 1.1775752251262903, + "grad_norm": 2.2785165309906006, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7417508363723755, + "num_tokens": 272511564.0, + "step": 10723 + }, + { + "epoch": 1.177685042828904, + "grad_norm": 2.133275032043457, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7280958294868469, + "num_tokens": 272539502.0, + "step": 10724 + }, + { + "epoch": 1.1777948605315176, + "grad_norm": 2.4170501232147217, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.724626898765564, + "num_tokens": 272562227.0, + "step": 10725 + }, + { + "epoch": 1.1779046782341314, + "grad_norm": 2.364287853240967, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7190533876419067, + "num_tokens": 272585245.0, + "step": 10726 + }, + { + "epoch": 1.1780144959367451, + "grad_norm": 2.332273244857788, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7041780352592468, + "num_tokens": 272612060.0, + "step": 10727 + }, + { + "epoch": 1.1781243136393587, + "grad_norm": 2.077467203140259, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7125334739685059, + "num_tokens": 272640812.0, + "step": 10728 + }, + { + "epoch": 1.1782341313419724, + "grad_norm": 2.1755306720733643, + "learning_rate": 1e-06, + "loss": 0.7598, + "mean_token_accuracy": 0.7612283229827881, + "num_tokens": 272666181.0, + "step": 10729 + }, + { + "epoch": 1.178343949044586, + "grad_norm": 2.1241302490234375, + "learning_rate": 1e-06, + "loss": 0.8086, + "mean_token_accuracy": 0.7544959783554077, + "num_tokens": 272690601.0, + "step": 10730 + }, + { + "epoch": 1.1784537667471997, + "grad_norm": 2.195310115814209, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7265298366546631, + "num_tokens": 272717088.0, + "step": 10731 + }, + { + "epoch": 1.1785635844498132, + "grad_norm": 2.299351930618286, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.711772084236145, + "num_tokens": 272742275.0, + "step": 10732 + }, + { + "epoch": 1.178673402152427, + "grad_norm": 2.208895206451416, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7049050331115723, + "num_tokens": 272770647.0, + "step": 10733 + }, + { + "epoch": 1.1787832198550405, + "grad_norm": 2.2761785984039307, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7148339748382568, + "num_tokens": 272795926.0, + "step": 10734 + }, + { + "epoch": 1.1788930375576543, + "grad_norm": 2.3483667373657227, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7059308886528015, + "num_tokens": 272820532.0, + "step": 10735 + }, + { + "epoch": 1.179002855260268, + "grad_norm": 2.4803738594055176, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7323465347290039, + "num_tokens": 272844036.0, + "step": 10736 + }, + { + "epoch": 1.1791126729628816, + "grad_norm": 2.2831308841705322, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7227575778961182, + "num_tokens": 272867895.0, + "step": 10737 + }, + { + "epoch": 1.1792224906654953, + "grad_norm": 2.7552237510681152, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.720431387424469, + "num_tokens": 272886848.0, + "step": 10738 + }, + { + "epoch": 1.1793323083681089, + "grad_norm": 2.0444531440734863, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7137870788574219, + "num_tokens": 272914824.0, + "step": 10739 + }, + { + "epoch": 1.1794421260707226, + "grad_norm": 1.948769450187683, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7178622484207153, + "num_tokens": 272944965.0, + "step": 10740 + }, + { + "epoch": 1.1795519437733364, + "grad_norm": 1.8925329446792603, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6956750154495239, + "num_tokens": 272980718.0, + "step": 10741 + }, + { + "epoch": 1.17966176147595, + "grad_norm": 2.4038777351379395, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7018026113510132, + "num_tokens": 273006059.0, + "step": 10742 + }, + { + "epoch": 1.1797715791785637, + "grad_norm": 2.37117075920105, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.7418543100357056, + "num_tokens": 273027653.0, + "step": 10743 + }, + { + "epoch": 1.1798813968811772, + "grad_norm": 2.4216268062591553, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7121410965919495, + "num_tokens": 273051202.0, + "step": 10744 + }, + { + "epoch": 1.179991214583791, + "grad_norm": 2.57643723487854, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.725387692451477, + "num_tokens": 273070631.0, + "step": 10745 + }, + { + "epoch": 1.1801010322864045, + "grad_norm": 2.452737331390381, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7345459461212158, + "num_tokens": 273092525.0, + "step": 10746 + }, + { + "epoch": 1.1802108499890183, + "grad_norm": 1.9724366664886475, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7016621828079224, + "num_tokens": 273125897.0, + "step": 10747 + }, + { + "epoch": 1.1803206676916318, + "grad_norm": 1.9329464435577393, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7227829694747925, + "num_tokens": 273157449.0, + "step": 10748 + }, + { + "epoch": 1.1804304853942456, + "grad_norm": 2.557447910308838, + "learning_rate": 1e-06, + "loss": 0.7881, + "mean_token_accuracy": 0.7537096738815308, + "num_tokens": 273175724.0, + "step": 10749 + }, + { + "epoch": 1.1805403030968593, + "grad_norm": 2.2589495182037354, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7171192169189453, + "num_tokens": 273200748.0, + "step": 10750 + }, + { + "epoch": 1.1806501207994728, + "grad_norm": 1.9788129329681396, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.6980575323104858, + "num_tokens": 273233324.0, + "step": 10751 + }, + { + "epoch": 1.1807599385020866, + "grad_norm": 2.4832963943481445, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7358243465423584, + "num_tokens": 273255373.0, + "step": 10752 + }, + { + "epoch": 1.1808697562047001, + "grad_norm": 2.306605100631714, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7104840874671936, + "num_tokens": 273279421.0, + "step": 10753 + }, + { + "epoch": 1.180979573907314, + "grad_norm": 2.2524607181549072, + "learning_rate": 1e-06, + "loss": 0.9801, + "mean_token_accuracy": 0.7016701102256775, + "num_tokens": 273306933.0, + "step": 10754 + }, + { + "epoch": 1.1810893916099274, + "grad_norm": 2.1183536052703857, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.7076128721237183, + "num_tokens": 273335885.0, + "step": 10755 + }, + { + "epoch": 1.1811992093125412, + "grad_norm": 2.355825424194336, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7266037464141846, + "num_tokens": 273361582.0, + "step": 10756 + }, + { + "epoch": 1.1813090270151547, + "grad_norm": 2.1524431705474854, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7239885330200195, + "num_tokens": 273389474.0, + "step": 10757 + }, + { + "epoch": 1.1814188447177685, + "grad_norm": 2.290008783340454, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7364441156387329, + "num_tokens": 273414369.0, + "step": 10758 + }, + { + "epoch": 1.1815286624203822, + "grad_norm": 2.0296688079833984, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7351380586624146, + "num_tokens": 273443995.0, + "step": 10759 + }, + { + "epoch": 1.1816384801229958, + "grad_norm": 2.1684939861297607, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.6945047378540039, + "num_tokens": 273471719.0, + "step": 10760 + }, + { + "epoch": 1.1817482978256095, + "grad_norm": 2.218500852584839, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7053929567337036, + "num_tokens": 273499073.0, + "step": 10761 + }, + { + "epoch": 1.181858115528223, + "grad_norm": 1.9595016241073608, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7166213989257812, + "num_tokens": 273531031.0, + "step": 10762 + }, + { + "epoch": 1.1819679332308368, + "grad_norm": 2.490814447402954, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7220929861068726, + "num_tokens": 273553950.0, + "step": 10763 + }, + { + "epoch": 1.1820777509334506, + "grad_norm": 2.2603635787963867, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7252687811851501, + "num_tokens": 273577682.0, + "step": 10764 + }, + { + "epoch": 1.1821875686360641, + "grad_norm": 2.3026201725006104, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7055739760398865, + "num_tokens": 273603453.0, + "step": 10765 + }, + { + "epoch": 1.1822973863386779, + "grad_norm": 2.3042585849761963, + "learning_rate": 1e-06, + "loss": 0.8203, + "mean_token_accuracy": 0.7348737716674805, + "num_tokens": 273626709.0, + "step": 10766 + }, + { + "epoch": 1.1824072040412914, + "grad_norm": 2.5109894275665283, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.733462929725647, + "num_tokens": 273647148.0, + "step": 10767 + }, + { + "epoch": 1.1825170217439052, + "grad_norm": 2.2597713470458984, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.711506724357605, + "num_tokens": 273671561.0, + "step": 10768 + }, + { + "epoch": 1.1826268394465187, + "grad_norm": 2.25783634185791, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7306140661239624, + "num_tokens": 273694722.0, + "step": 10769 + }, + { + "epoch": 1.1827366571491325, + "grad_norm": 2.1506972312927246, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7171318531036377, + "num_tokens": 273721700.0, + "step": 10770 + }, + { + "epoch": 1.182846474851746, + "grad_norm": 2.3056447505950928, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7295467257499695, + "num_tokens": 273746311.0, + "step": 10771 + }, + { + "epoch": 1.1829562925543597, + "grad_norm": 2.0089032649993896, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7291028499603271, + "num_tokens": 273775699.0, + "step": 10772 + }, + { + "epoch": 1.1830661102569735, + "grad_norm": 2.5527584552764893, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7102517485618591, + "num_tokens": 273795419.0, + "step": 10773 + }, + { + "epoch": 1.183175927959587, + "grad_norm": 2.290969133377075, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7153112888336182, + "num_tokens": 273824265.0, + "step": 10774 + }, + { + "epoch": 1.1832857456622008, + "grad_norm": 2.398787498474121, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7063155770301819, + "num_tokens": 273846945.0, + "step": 10775 + }, + { + "epoch": 1.1833955633648143, + "grad_norm": 1.9306674003601074, + "learning_rate": 1e-06, + "loss": 0.9969, + "mean_token_accuracy": 0.6950864791870117, + "num_tokens": 273882151.0, + "step": 10776 + }, + { + "epoch": 1.183505381067428, + "grad_norm": 2.3622610569000244, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7100416421890259, + "num_tokens": 273905922.0, + "step": 10777 + }, + { + "epoch": 1.1836151987700418, + "grad_norm": 2.384542465209961, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7019785046577454, + "num_tokens": 273929383.0, + "step": 10778 + }, + { + "epoch": 1.1837250164726554, + "grad_norm": 2.219866991043091, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7290182113647461, + "num_tokens": 273955129.0, + "step": 10779 + }, + { + "epoch": 1.1838348341752691, + "grad_norm": 2.040830373764038, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7213601469993591, + "num_tokens": 273986366.0, + "step": 10780 + }, + { + "epoch": 1.1839446518778827, + "grad_norm": 2.328282356262207, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7030177116394043, + "num_tokens": 274011359.0, + "step": 10781 + }, + { + "epoch": 1.1840544695804964, + "grad_norm": 2.006789207458496, + "learning_rate": 1e-06, + "loss": 1.0147, + "mean_token_accuracy": 0.6896470785140991, + "num_tokens": 274045022.0, + "step": 10782 + }, + { + "epoch": 1.18416428728311, + "grad_norm": 2.3212878704071045, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6949846744537354, + "num_tokens": 274071400.0, + "step": 10783 + }, + { + "epoch": 1.1842741049857237, + "grad_norm": 2.5640714168548584, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7093532085418701, + "num_tokens": 274092047.0, + "step": 10784 + }, + { + "epoch": 1.1843839226883373, + "grad_norm": 2.2875661849975586, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7242381572723389, + "num_tokens": 274118450.0, + "step": 10785 + }, + { + "epoch": 1.184493740390951, + "grad_norm": 2.46333384513855, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7204606533050537, + "num_tokens": 274139582.0, + "step": 10786 + }, + { + "epoch": 1.1846035580935648, + "grad_norm": 1.9641103744506836, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.726502001285553, + "num_tokens": 274172771.0, + "step": 10787 + }, + { + "epoch": 1.1847133757961783, + "grad_norm": 2.277344226837158, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7140687704086304, + "num_tokens": 274198303.0, + "step": 10788 + }, + { + "epoch": 1.184823193498792, + "grad_norm": 2.3513152599334717, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7037352323532104, + "num_tokens": 274223178.0, + "step": 10789 + }, + { + "epoch": 1.1849330112014056, + "grad_norm": 2.4050872325897217, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7346168160438538, + "num_tokens": 274245685.0, + "step": 10790 + }, + { + "epoch": 1.1850428289040194, + "grad_norm": 2.0621120929718018, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.715682864189148, + "num_tokens": 274274838.0, + "step": 10791 + }, + { + "epoch": 1.185152646606633, + "grad_norm": 2.366071939468384, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7289185523986816, + "num_tokens": 274297057.0, + "step": 10792 + }, + { + "epoch": 1.1852624643092466, + "grad_norm": 2.237832546234131, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.706459641456604, + "num_tokens": 274322259.0, + "step": 10793 + }, + { + "epoch": 1.1853722820118604, + "grad_norm": 2.3272335529327393, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7145288586616516, + "num_tokens": 274346054.0, + "step": 10794 + }, + { + "epoch": 1.185482099714474, + "grad_norm": 2.348749876022339, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.725441575050354, + "num_tokens": 274371497.0, + "step": 10795 + }, + { + "epoch": 1.1855919174170877, + "grad_norm": 2.1853950023651123, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7155701518058777, + "num_tokens": 274396917.0, + "step": 10796 + }, + { + "epoch": 1.1857017351197012, + "grad_norm": 2.355269193649292, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7302740216255188, + "num_tokens": 274421411.0, + "step": 10797 + }, + { + "epoch": 1.185811552822315, + "grad_norm": 2.141456365585327, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7254089713096619, + "num_tokens": 274448553.0, + "step": 10798 + }, + { + "epoch": 1.1859213705249285, + "grad_norm": 2.261889696121216, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.705039381980896, + "num_tokens": 274474817.0, + "step": 10799 + }, + { + "epoch": 1.1860311882275423, + "grad_norm": 2.2486109733581543, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7169766426086426, + "num_tokens": 274502612.0, + "step": 10800 + }, + { + "epoch": 1.186141005930156, + "grad_norm": 2.188934803009033, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7315799593925476, + "num_tokens": 274529500.0, + "step": 10801 + }, + { + "epoch": 1.1862508236327696, + "grad_norm": 2.1806678771972656, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.6979354023933411, + "num_tokens": 274559545.0, + "step": 10802 + }, + { + "epoch": 1.1863606413353833, + "grad_norm": 2.3909058570861816, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7233368158340454, + "num_tokens": 274582454.0, + "step": 10803 + }, + { + "epoch": 1.1864704590379969, + "grad_norm": 2.08693790435791, + "learning_rate": 1e-06, + "loss": 0.9773, + "mean_token_accuracy": 0.6967654228210449, + "num_tokens": 274612878.0, + "step": 10804 + }, + { + "epoch": 1.1865802767406106, + "grad_norm": 2.3817758560180664, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7377976179122925, + "num_tokens": 274637904.0, + "step": 10805 + }, + { + "epoch": 1.1866900944432242, + "grad_norm": 2.4011895656585693, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7348291277885437, + "num_tokens": 274659845.0, + "step": 10806 + }, + { + "epoch": 1.186799912145838, + "grad_norm": 2.0351321697235107, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7337614297866821, + "num_tokens": 274690692.0, + "step": 10807 + }, + { + "epoch": 1.1869097298484517, + "grad_norm": 2.150996685028076, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7129062414169312, + "num_tokens": 274719334.0, + "step": 10808 + }, + { + "epoch": 1.1870195475510652, + "grad_norm": 2.439223527908325, + "learning_rate": 1e-06, + "loss": 0.7918, + "mean_token_accuracy": 0.7454417943954468, + "num_tokens": 274742027.0, + "step": 10809 + }, + { + "epoch": 1.187129365253679, + "grad_norm": 2.3227860927581787, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7239252924919128, + "num_tokens": 274767691.0, + "step": 10810 + }, + { + "epoch": 1.1872391829562925, + "grad_norm": 2.0970993041992188, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.717332124710083, + "num_tokens": 274796367.0, + "step": 10811 + }, + { + "epoch": 1.1873490006589063, + "grad_norm": 2.210576295852661, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7052814960479736, + "num_tokens": 274824045.0, + "step": 10812 + }, + { + "epoch": 1.1874588183615198, + "grad_norm": 2.3748936653137207, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7340998649597168, + "num_tokens": 274847799.0, + "step": 10813 + }, + { + "epoch": 1.1875686360641335, + "grad_norm": 2.3012163639068604, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7361259460449219, + "num_tokens": 274871713.0, + "step": 10814 + }, + { + "epoch": 1.1876784537667473, + "grad_norm": 2.4564335346221924, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7168691754341125, + "num_tokens": 274893207.0, + "step": 10815 + }, + { + "epoch": 1.1877882714693608, + "grad_norm": 2.4427683353424072, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7415947914123535, + "num_tokens": 274915175.0, + "step": 10816 + }, + { + "epoch": 1.1878980891719746, + "grad_norm": 2.530315637588501, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7120828032493591, + "num_tokens": 274936198.0, + "step": 10817 + }, + { + "epoch": 1.1880079068745881, + "grad_norm": 2.343869686126709, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7220543026924133, + "num_tokens": 274960676.0, + "step": 10818 + }, + { + "epoch": 1.1881177245772019, + "grad_norm": 2.367936134338379, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7244634628295898, + "num_tokens": 274984594.0, + "step": 10819 + }, + { + "epoch": 1.1882275422798154, + "grad_norm": 2.3583879470825195, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7409482002258301, + "num_tokens": 275007118.0, + "step": 10820 + }, + { + "epoch": 1.1883373599824292, + "grad_norm": 1.887526273727417, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7081083059310913, + "num_tokens": 275041529.0, + "step": 10821 + }, + { + "epoch": 1.1884471776850427, + "grad_norm": 2.1568822860717773, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7229974269866943, + "num_tokens": 275067133.0, + "step": 10822 + }, + { + "epoch": 1.1885569953876565, + "grad_norm": 2.5208590030670166, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7379745841026306, + "num_tokens": 275088690.0, + "step": 10823 + }, + { + "epoch": 1.1886668130902702, + "grad_norm": 2.0542352199554443, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.706823468208313, + "num_tokens": 275118840.0, + "step": 10824 + }, + { + "epoch": 1.1887766307928838, + "grad_norm": 2.2376673221588135, + "learning_rate": 1e-06, + "loss": 0.8301, + "mean_token_accuracy": 0.7427167892456055, + "num_tokens": 275142576.0, + "step": 10825 + }, + { + "epoch": 1.1888864484954975, + "grad_norm": 2.4488463401794434, + "learning_rate": 1e-06, + "loss": 0.7848, + "mean_token_accuracy": 0.7463493347167969, + "num_tokens": 275162321.0, + "step": 10826 + }, + { + "epoch": 1.188996266198111, + "grad_norm": 2.447068452835083, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7000852227210999, + "num_tokens": 275184695.0, + "step": 10827 + }, + { + "epoch": 1.1891060839007248, + "grad_norm": 2.719024896621704, + "learning_rate": 1e-06, + "loss": 0.8159, + "mean_token_accuracy": 0.7412230968475342, + "num_tokens": 275203438.0, + "step": 10828 + }, + { + "epoch": 1.1892159016033386, + "grad_norm": 2.3450770378112793, + "learning_rate": 1e-06, + "loss": 0.8136, + "mean_token_accuracy": 0.742833137512207, + "num_tokens": 275225729.0, + "step": 10829 + }, + { + "epoch": 1.189325719305952, + "grad_norm": 1.9350552558898926, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.704997181892395, + "num_tokens": 275258120.0, + "step": 10830 + }, + { + "epoch": 1.1894355370085659, + "grad_norm": 1.9692871570587158, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7115281820297241, + "num_tokens": 275291175.0, + "step": 10831 + }, + { + "epoch": 1.1895453547111794, + "grad_norm": 2.3559622764587402, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6922463178634644, + "num_tokens": 275315419.0, + "step": 10832 + }, + { + "epoch": 1.1896551724137931, + "grad_norm": 1.9905719757080078, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7162307500839233, + "num_tokens": 275349003.0, + "step": 10833 + }, + { + "epoch": 1.1897649901164067, + "grad_norm": 2.301478862762451, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7164111137390137, + "num_tokens": 275372496.0, + "step": 10834 + }, + { + "epoch": 1.1898748078190204, + "grad_norm": 2.3545491695404053, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7140012979507446, + "num_tokens": 275395941.0, + "step": 10835 + }, + { + "epoch": 1.189984625521634, + "grad_norm": 2.0615129470825195, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.6889639496803284, + "num_tokens": 275428922.0, + "step": 10836 + }, + { + "epoch": 1.1900944432242477, + "grad_norm": 2.3032114505767822, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.723385214805603, + "num_tokens": 275455856.0, + "step": 10837 + }, + { + "epoch": 1.1902042609268615, + "grad_norm": 2.5222370624542236, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7283080220222473, + "num_tokens": 275476836.0, + "step": 10838 + }, + { + "epoch": 1.190314078629475, + "grad_norm": 2.1891770362854004, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7071104049682617, + "num_tokens": 275503670.0, + "step": 10839 + }, + { + "epoch": 1.1904238963320888, + "grad_norm": 2.170109987258911, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7055120468139648, + "num_tokens": 275532331.0, + "step": 10840 + }, + { + "epoch": 1.1905337140347023, + "grad_norm": 2.1382126808166504, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.715873122215271, + "num_tokens": 275559579.0, + "step": 10841 + }, + { + "epoch": 1.190643531737316, + "grad_norm": 2.327569007873535, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7181349992752075, + "num_tokens": 275583057.0, + "step": 10842 + }, + { + "epoch": 1.1907533494399298, + "grad_norm": 2.2820515632629395, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7112042903900146, + "num_tokens": 275607351.0, + "step": 10843 + }, + { + "epoch": 1.1908631671425434, + "grad_norm": 2.140749454498291, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7014321684837341, + "num_tokens": 275636284.0, + "step": 10844 + }, + { + "epoch": 1.1909729848451571, + "grad_norm": 2.0390684604644775, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7180376052856445, + "num_tokens": 275666022.0, + "step": 10845 + }, + { + "epoch": 1.1910828025477707, + "grad_norm": 2.1248631477355957, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7409529685974121, + "num_tokens": 275691341.0, + "step": 10846 + }, + { + "epoch": 1.1911926202503844, + "grad_norm": 2.1107912063598633, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7249541878700256, + "num_tokens": 275720934.0, + "step": 10847 + }, + { + "epoch": 1.191302437952998, + "grad_norm": 2.2332191467285156, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7358711361885071, + "num_tokens": 275747546.0, + "step": 10848 + }, + { + "epoch": 1.1914122556556117, + "grad_norm": 2.370013475418091, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7141866683959961, + "num_tokens": 275771308.0, + "step": 10849 + }, + { + "epoch": 1.1915220733582252, + "grad_norm": 2.1991655826568604, + "learning_rate": 1e-06, + "loss": 0.8221, + "mean_token_accuracy": 0.7409552335739136, + "num_tokens": 275796157.0, + "step": 10850 + }, + { + "epoch": 1.191631891060839, + "grad_norm": 2.174184560775757, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7061471343040466, + "num_tokens": 275825034.0, + "step": 10851 + }, + { + "epoch": 1.1917417087634528, + "grad_norm": 2.2947328090667725, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7084239721298218, + "num_tokens": 275850206.0, + "step": 10852 + }, + { + "epoch": 1.1918515264660663, + "grad_norm": 2.3566250801086426, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7310764789581299, + "num_tokens": 275873856.0, + "step": 10853 + }, + { + "epoch": 1.19196134416868, + "grad_norm": 2.1877236366271973, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7154650688171387, + "num_tokens": 275901453.0, + "step": 10854 + }, + { + "epoch": 1.1920711618712936, + "grad_norm": 1.9489257335662842, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7267506122589111, + "num_tokens": 275929970.0, + "step": 10855 + }, + { + "epoch": 1.1921809795739073, + "grad_norm": 2.233109712600708, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7256739139556885, + "num_tokens": 275956862.0, + "step": 10856 + }, + { + "epoch": 1.192290797276521, + "grad_norm": 2.0525336265563965, + "learning_rate": 1e-06, + "loss": 1.0324, + "mean_token_accuracy": 0.6825103759765625, + "num_tokens": 275987662.0, + "step": 10857 + }, + { + "epoch": 1.1924006149791346, + "grad_norm": 2.3738090991973877, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7142781019210815, + "num_tokens": 276010767.0, + "step": 10858 + }, + { + "epoch": 1.1925104326817484, + "grad_norm": 2.4143893718719482, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7508086562156677, + "num_tokens": 276030853.0, + "step": 10859 + }, + { + "epoch": 1.192620250384362, + "grad_norm": 2.3188490867614746, + "learning_rate": 1e-06, + "loss": 1.0319, + "mean_token_accuracy": 0.6821317672729492, + "num_tokens": 276057575.0, + "step": 10860 + }, + { + "epoch": 1.1927300680869757, + "grad_norm": 2.2988064289093018, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7165716886520386, + "num_tokens": 276082026.0, + "step": 10861 + }, + { + "epoch": 1.1928398857895892, + "grad_norm": 2.1032392978668213, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6990455389022827, + "num_tokens": 276109217.0, + "step": 10862 + }, + { + "epoch": 1.192949703492203, + "grad_norm": 2.4604008197784424, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7083922624588013, + "num_tokens": 276130972.0, + "step": 10863 + }, + { + "epoch": 1.1930595211948165, + "grad_norm": 2.369696617126465, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7394220232963562, + "num_tokens": 276152990.0, + "step": 10864 + }, + { + "epoch": 1.1931693388974303, + "grad_norm": 2.1698646545410156, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7308850884437561, + "num_tokens": 276178274.0, + "step": 10865 + }, + { + "epoch": 1.193279156600044, + "grad_norm": 2.053138494491577, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7104266285896301, + "num_tokens": 276208938.0, + "step": 10866 + }, + { + "epoch": 1.1933889743026576, + "grad_norm": 2.0713560581207275, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.6944062113761902, + "num_tokens": 276239292.0, + "step": 10867 + }, + { + "epoch": 1.1934987920052713, + "grad_norm": 2.2099506855010986, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7116157412528992, + "num_tokens": 276268295.0, + "step": 10868 + }, + { + "epoch": 1.1936086097078848, + "grad_norm": 2.432161808013916, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7225803136825562, + "num_tokens": 276289046.0, + "step": 10869 + }, + { + "epoch": 1.1937184274104986, + "grad_norm": 2.255253791809082, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7213129997253418, + "num_tokens": 276315422.0, + "step": 10870 + }, + { + "epoch": 1.1938282451131121, + "grad_norm": 2.2638351917266846, + "learning_rate": 1e-06, + "loss": 1.032, + "mean_token_accuracy": 0.6877562403678894, + "num_tokens": 276340773.0, + "step": 10871 + }, + { + "epoch": 1.193938062815726, + "grad_norm": 2.3142483234405518, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7188528776168823, + "num_tokens": 276363188.0, + "step": 10872 + }, + { + "epoch": 1.1940478805183394, + "grad_norm": 2.184384822845459, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7187244296073914, + "num_tokens": 276390269.0, + "step": 10873 + }, + { + "epoch": 1.1941576982209532, + "grad_norm": 2.0658750534057617, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.707669198513031, + "num_tokens": 276421279.0, + "step": 10874 + }, + { + "epoch": 1.194267515923567, + "grad_norm": 2.2992634773254395, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7085850238800049, + "num_tokens": 276447391.0, + "step": 10875 + }, + { + "epoch": 1.1943773336261805, + "grad_norm": 2.5794286727905273, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7080328464508057, + "num_tokens": 276467818.0, + "step": 10876 + }, + { + "epoch": 1.1944871513287942, + "grad_norm": 2.1896584033966064, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7247135043144226, + "num_tokens": 276493147.0, + "step": 10877 + }, + { + "epoch": 1.1945969690314078, + "grad_norm": 2.359482526779175, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7178361415863037, + "num_tokens": 276516911.0, + "step": 10878 + }, + { + "epoch": 1.1947067867340215, + "grad_norm": 2.138582944869995, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.709686279296875, + "num_tokens": 276545485.0, + "step": 10879 + }, + { + "epoch": 1.1948166044366353, + "grad_norm": 2.3999383449554443, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7183164358139038, + "num_tokens": 276569607.0, + "step": 10880 + }, + { + "epoch": 1.1949264221392488, + "grad_norm": 2.2090680599212646, + "learning_rate": 1e-06, + "loss": 1.0097, + "mean_token_accuracy": 0.6945606470108032, + "num_tokens": 276598492.0, + "step": 10881 + }, + { + "epoch": 1.1950362398418626, + "grad_norm": 2.231172800064087, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7332488298416138, + "num_tokens": 276621538.0, + "step": 10882 + }, + { + "epoch": 1.1951460575444761, + "grad_norm": 2.5418121814727783, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7222682237625122, + "num_tokens": 276642472.0, + "step": 10883 + }, + { + "epoch": 1.1952558752470899, + "grad_norm": 2.318434715270996, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7153306007385254, + "num_tokens": 276665339.0, + "step": 10884 + }, + { + "epoch": 1.1953656929497034, + "grad_norm": 2.5191142559051514, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7127507328987122, + "num_tokens": 276687382.0, + "step": 10885 + }, + { + "epoch": 1.1954755106523172, + "grad_norm": 2.129709243774414, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7244716286659241, + "num_tokens": 276713650.0, + "step": 10886 + }, + { + "epoch": 1.1955853283549307, + "grad_norm": 2.854463577270508, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7326639890670776, + "num_tokens": 276730847.0, + "step": 10887 + }, + { + "epoch": 1.1956951460575445, + "grad_norm": 2.1992900371551514, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7098832130432129, + "num_tokens": 276758373.0, + "step": 10888 + }, + { + "epoch": 1.1958049637601582, + "grad_norm": 2.157465934753418, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.721400797367096, + "num_tokens": 276784369.0, + "step": 10889 + }, + { + "epoch": 1.1959147814627717, + "grad_norm": 2.1043293476104736, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7240657806396484, + "num_tokens": 276811449.0, + "step": 10890 + }, + { + "epoch": 1.1960245991653855, + "grad_norm": 2.1749651432037354, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7504924535751343, + "num_tokens": 276839151.0, + "step": 10891 + }, + { + "epoch": 1.196134416867999, + "grad_norm": 2.2853610515594482, + "learning_rate": 1e-06, + "loss": 0.7779, + "mean_token_accuracy": 0.7498534321784973, + "num_tokens": 276861588.0, + "step": 10892 + }, + { + "epoch": 1.1962442345706128, + "grad_norm": 2.380220890045166, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7290804386138916, + "num_tokens": 276883275.0, + "step": 10893 + }, + { + "epoch": 1.1963540522732266, + "grad_norm": 2.1096746921539307, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7078931331634521, + "num_tokens": 276911593.0, + "step": 10894 + }, + { + "epoch": 1.19646386997584, + "grad_norm": 2.4148731231689453, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7152698636054993, + "num_tokens": 276935812.0, + "step": 10895 + }, + { + "epoch": 1.1965736876784538, + "grad_norm": 2.1761064529418945, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7172300815582275, + "num_tokens": 276959841.0, + "step": 10896 + }, + { + "epoch": 1.1966835053810674, + "grad_norm": 2.227543592453003, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7181832790374756, + "num_tokens": 276984603.0, + "step": 10897 + }, + { + "epoch": 1.1967933230836811, + "grad_norm": 2.1923232078552246, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7270188331604004, + "num_tokens": 277008194.0, + "step": 10898 + }, + { + "epoch": 1.1969031407862947, + "grad_norm": 1.9867725372314453, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.702853798866272, + "num_tokens": 277040500.0, + "step": 10899 + }, + { + "epoch": 1.1970129584889084, + "grad_norm": 2.1782236099243164, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7270055413246155, + "num_tokens": 277066862.0, + "step": 10900 + }, + { + "epoch": 1.197122776191522, + "grad_norm": 2.1439337730407715, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7025176286697388, + "num_tokens": 277092370.0, + "step": 10901 + }, + { + "epoch": 1.1972325938941357, + "grad_norm": 2.326831817626953, + "learning_rate": 1e-06, + "loss": 0.7663, + "mean_token_accuracy": 0.7558344006538391, + "num_tokens": 277114034.0, + "step": 10902 + }, + { + "epoch": 1.1973424115967495, + "grad_norm": 2.3414347171783447, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.739230751991272, + "num_tokens": 277137485.0, + "step": 10903 + }, + { + "epoch": 1.197452229299363, + "grad_norm": 2.490471601486206, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7067306041717529, + "num_tokens": 277159355.0, + "step": 10904 + }, + { + "epoch": 1.1975620470019768, + "grad_norm": 2.408818244934082, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7238736152648926, + "num_tokens": 277181266.0, + "step": 10905 + }, + { + "epoch": 1.1976718647045903, + "grad_norm": 2.7876367568969727, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7170277237892151, + "num_tokens": 277199517.0, + "step": 10906 + }, + { + "epoch": 1.197781682407204, + "grad_norm": 2.052461624145508, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7316476702690125, + "num_tokens": 277227490.0, + "step": 10907 + }, + { + "epoch": 1.1978915001098178, + "grad_norm": 2.4803357124328613, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7253326177597046, + "num_tokens": 277249364.0, + "step": 10908 + }, + { + "epoch": 1.1980013178124314, + "grad_norm": 2.270061492919922, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7221052646636963, + "num_tokens": 277275965.0, + "step": 10909 + }, + { + "epoch": 1.1981111355150451, + "grad_norm": 2.320446729660034, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.696121335029602, + "num_tokens": 277299581.0, + "step": 10910 + }, + { + "epoch": 1.1982209532176586, + "grad_norm": 2.2489211559295654, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7335537672042847, + "num_tokens": 277325523.0, + "step": 10911 + }, + { + "epoch": 1.1983307709202724, + "grad_norm": 1.9615176916122437, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7176541090011597, + "num_tokens": 277358048.0, + "step": 10912 + }, + { + "epoch": 1.198440588622886, + "grad_norm": 2.3998751640319824, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.728186845779419, + "num_tokens": 277381088.0, + "step": 10913 + }, + { + "epoch": 1.1985504063254997, + "grad_norm": 1.9620513916015625, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.698743462562561, + "num_tokens": 277412572.0, + "step": 10914 + }, + { + "epoch": 1.1986602240281132, + "grad_norm": 1.9417189359664917, + "learning_rate": 1e-06, + "loss": 0.945, + "mean_token_accuracy": 0.704982340335846, + "num_tokens": 277446964.0, + "step": 10915 + }, + { + "epoch": 1.198770041730727, + "grad_norm": 2.280980110168457, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7231855392456055, + "num_tokens": 277471027.0, + "step": 10916 + }, + { + "epoch": 1.1988798594333407, + "grad_norm": 2.1859090328216553, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7267665863037109, + "num_tokens": 277496931.0, + "step": 10917 + }, + { + "epoch": 1.1989896771359543, + "grad_norm": 2.1447112560272217, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.713245153427124, + "num_tokens": 277525226.0, + "step": 10918 + }, + { + "epoch": 1.199099494838568, + "grad_norm": 2.925057888031006, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7276317477226257, + "num_tokens": 277541536.0, + "step": 10919 + }, + { + "epoch": 1.1992093125411816, + "grad_norm": 2.1601974964141846, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7043713927268982, + "num_tokens": 277569199.0, + "step": 10920 + }, + { + "epoch": 1.1993191302437953, + "grad_norm": 2.0210907459259033, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7189638018608093, + "num_tokens": 277597974.0, + "step": 10921 + }, + { + "epoch": 1.199428947946409, + "grad_norm": 2.267167329788208, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7217103242874146, + "num_tokens": 277622782.0, + "step": 10922 + }, + { + "epoch": 1.1995387656490226, + "grad_norm": 2.144008159637451, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7175154089927673, + "num_tokens": 277650482.0, + "step": 10923 + }, + { + "epoch": 1.1996485833516364, + "grad_norm": 2.2437996864318848, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7368892431259155, + "num_tokens": 277673838.0, + "step": 10924 + }, + { + "epoch": 1.19975840105425, + "grad_norm": 2.400592565536499, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7321467399597168, + "num_tokens": 277697091.0, + "step": 10925 + }, + { + "epoch": 1.1998682187568637, + "grad_norm": 2.0615012645721436, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.736553430557251, + "num_tokens": 277725861.0, + "step": 10926 + }, + { + "epoch": 1.1999780364594772, + "grad_norm": 2.1305882930755615, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7467402219772339, + "num_tokens": 277753336.0, + "step": 10927 + }, + { + "epoch": 1.200087854162091, + "grad_norm": 2.4048945903778076, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7046840190887451, + "num_tokens": 277777769.0, + "step": 10928 + }, + { + "epoch": 1.2001976718647045, + "grad_norm": 2.052212715148926, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7096080183982849, + "num_tokens": 277807898.0, + "step": 10929 + }, + { + "epoch": 1.2003074895673183, + "grad_norm": 2.198993682861328, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.6996113061904907, + "num_tokens": 277834083.0, + "step": 10930 + }, + { + "epoch": 1.200417307269932, + "grad_norm": 2.6746394634246826, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7299389839172363, + "num_tokens": 277853328.0, + "step": 10931 + }, + { + "epoch": 1.2005271249725455, + "grad_norm": 2.3548033237457275, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7274727821350098, + "num_tokens": 277876218.0, + "step": 10932 + }, + { + "epoch": 1.2006369426751593, + "grad_norm": 2.2532434463500977, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7165207862854004, + "num_tokens": 277901900.0, + "step": 10933 + }, + { + "epoch": 1.2007467603777728, + "grad_norm": 2.199774980545044, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7147642374038696, + "num_tokens": 277929167.0, + "step": 10934 + }, + { + "epoch": 1.2008565780803866, + "grad_norm": 2.5699665546417236, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.6970511674880981, + "num_tokens": 277955175.0, + "step": 10935 + }, + { + "epoch": 1.2009663957830001, + "grad_norm": 2.130311965942383, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7014224529266357, + "num_tokens": 277984282.0, + "step": 10936 + }, + { + "epoch": 1.2010762134856139, + "grad_norm": 2.5675981044769287, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7305344939231873, + "num_tokens": 278005030.0, + "step": 10937 + }, + { + "epoch": 1.2011860311882274, + "grad_norm": 2.370490312576294, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.716098427772522, + "num_tokens": 278028385.0, + "step": 10938 + }, + { + "epoch": 1.2012958488908412, + "grad_norm": 2.0983028411865234, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7331725358963013, + "num_tokens": 278054997.0, + "step": 10939 + }, + { + "epoch": 1.201405666593455, + "grad_norm": 2.091235637664795, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.6946744322776794, + "num_tokens": 278086379.0, + "step": 10940 + }, + { + "epoch": 1.2015154842960685, + "grad_norm": 2.4563536643981934, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7056933045387268, + "num_tokens": 278110204.0, + "step": 10941 + }, + { + "epoch": 1.2016253019986822, + "grad_norm": 2.0351526737213135, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7097166776657104, + "num_tokens": 278142543.0, + "step": 10942 + }, + { + "epoch": 1.2017351197012958, + "grad_norm": 2.4483726024627686, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7352050542831421, + "num_tokens": 278163846.0, + "step": 10943 + }, + { + "epoch": 1.2018449374039095, + "grad_norm": 2.224522113800049, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7292550802230835, + "num_tokens": 278188825.0, + "step": 10944 + }, + { + "epoch": 1.2019547551065233, + "grad_norm": 2.717501401901245, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7161802053451538, + "num_tokens": 278208461.0, + "step": 10945 + }, + { + "epoch": 1.2020645728091368, + "grad_norm": 2.3256542682647705, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7248618006706238, + "num_tokens": 278231134.0, + "step": 10946 + }, + { + "epoch": 1.2021743905117506, + "grad_norm": 2.272047758102417, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7412654161453247, + "num_tokens": 278255011.0, + "step": 10947 + }, + { + "epoch": 1.202284208214364, + "grad_norm": 2.544130325317383, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7240961194038391, + "num_tokens": 278277397.0, + "step": 10948 + }, + { + "epoch": 1.2023940259169779, + "grad_norm": 2.276062488555908, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7106503248214722, + "num_tokens": 278302995.0, + "step": 10949 + }, + { + "epoch": 1.2025038436195914, + "grad_norm": 2.2283754348754883, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7273354530334473, + "num_tokens": 278328680.0, + "step": 10950 + }, + { + "epoch": 1.2026136613222052, + "grad_norm": 2.0940332412719727, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7337980270385742, + "num_tokens": 278356485.0, + "step": 10951 + }, + { + "epoch": 1.2027234790248187, + "grad_norm": 2.040764808654785, + "learning_rate": 1e-06, + "loss": 1.0425, + "mean_token_accuracy": 0.6805212497711182, + "num_tokens": 278387997.0, + "step": 10952 + }, + { + "epoch": 1.2028332967274324, + "grad_norm": 2.313469409942627, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7206603288650513, + "num_tokens": 278413032.0, + "step": 10953 + }, + { + "epoch": 1.2029431144300462, + "grad_norm": 2.3587095737457275, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7402167916297913, + "num_tokens": 278436425.0, + "step": 10954 + }, + { + "epoch": 1.2030529321326597, + "grad_norm": 2.3016369342803955, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.73117995262146, + "num_tokens": 278462445.0, + "step": 10955 + }, + { + "epoch": 1.2031627498352735, + "grad_norm": 2.1820924282073975, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7175617814064026, + "num_tokens": 278489602.0, + "step": 10956 + }, + { + "epoch": 1.203272567537887, + "grad_norm": 2.1709864139556885, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7070063948631287, + "num_tokens": 278517091.0, + "step": 10957 + }, + { + "epoch": 1.2033823852405008, + "grad_norm": 2.0392773151397705, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7107239365577698, + "num_tokens": 278548648.0, + "step": 10958 + }, + { + "epoch": 1.2034922029431145, + "grad_norm": 2.3216147422790527, + "learning_rate": 1e-06, + "loss": 0.8754, + "mean_token_accuracy": 0.7294772863388062, + "num_tokens": 278571998.0, + "step": 10959 + }, + { + "epoch": 1.203602020645728, + "grad_norm": 2.0791678428649902, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7050598859786987, + "num_tokens": 278600199.0, + "step": 10960 + }, + { + "epoch": 1.2037118383483418, + "grad_norm": 2.2738685607910156, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.735691249370575, + "num_tokens": 278625058.0, + "step": 10961 + }, + { + "epoch": 1.2038216560509554, + "grad_norm": 2.1851296424865723, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7107508778572083, + "num_tokens": 278653070.0, + "step": 10962 + }, + { + "epoch": 1.2039314737535691, + "grad_norm": 2.4420859813690186, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7146927118301392, + "num_tokens": 278674437.0, + "step": 10963 + }, + { + "epoch": 1.2040412914561827, + "grad_norm": 2.4071223735809326, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7177258729934692, + "num_tokens": 278696509.0, + "step": 10964 + }, + { + "epoch": 1.2041511091587964, + "grad_norm": 2.383822202682495, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7151116132736206, + "num_tokens": 278719304.0, + "step": 10965 + }, + { + "epoch": 1.20426092686141, + "grad_norm": 2.749824047088623, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7229292392730713, + "num_tokens": 278739261.0, + "step": 10966 + }, + { + "epoch": 1.2043707445640237, + "grad_norm": 2.4380667209625244, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7236195802688599, + "num_tokens": 278761350.0, + "step": 10967 + }, + { + "epoch": 1.2044805622666375, + "grad_norm": 2.4300901889801025, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7282052040100098, + "num_tokens": 278783877.0, + "step": 10968 + }, + { + "epoch": 1.204590379969251, + "grad_norm": 2.1748998165130615, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7198078036308289, + "num_tokens": 278810816.0, + "step": 10969 + }, + { + "epoch": 1.2047001976718648, + "grad_norm": 2.303288459777832, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7009087204933167, + "num_tokens": 278835350.0, + "step": 10970 + }, + { + "epoch": 1.2048100153744783, + "grad_norm": 2.374098062515259, + "learning_rate": 1e-06, + "loss": 0.8224, + "mean_token_accuracy": 0.7339341044425964, + "num_tokens": 278859127.0, + "step": 10971 + }, + { + "epoch": 1.204919833077092, + "grad_norm": 2.3372836112976074, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7252498865127563, + "num_tokens": 278883467.0, + "step": 10972 + }, + { + "epoch": 1.2050296507797058, + "grad_norm": 2.139495611190796, + "learning_rate": 1e-06, + "loss": 0.838, + "mean_token_accuracy": 0.7335549592971802, + "num_tokens": 278910983.0, + "step": 10973 + }, + { + "epoch": 1.2051394684823193, + "grad_norm": 2.6571764945983887, + "learning_rate": 1e-06, + "loss": 0.7291, + "mean_token_accuracy": 0.7647416591644287, + "num_tokens": 278929550.0, + "step": 10974 + }, + { + "epoch": 1.205249286184933, + "grad_norm": 2.563062906265259, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.735444962978363, + "num_tokens": 278948199.0, + "step": 10975 + }, + { + "epoch": 1.2053591038875466, + "grad_norm": 2.2019364833831787, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7246094346046448, + "num_tokens": 278973399.0, + "step": 10976 + }, + { + "epoch": 1.2054689215901604, + "grad_norm": 2.2758970260620117, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7085139751434326, + "num_tokens": 279000694.0, + "step": 10977 + }, + { + "epoch": 1.205578739292774, + "grad_norm": 2.1645708084106445, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7007853984832764, + "num_tokens": 279027668.0, + "step": 10978 + }, + { + "epoch": 1.2056885569953877, + "grad_norm": 2.062049388885498, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7065404057502747, + "num_tokens": 279058533.0, + "step": 10979 + }, + { + "epoch": 1.2057983746980012, + "grad_norm": 2.254168748855591, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7219239473342896, + "num_tokens": 279084300.0, + "step": 10980 + }, + { + "epoch": 1.205908192400615, + "grad_norm": 1.9718842506408691, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7309699058532715, + "num_tokens": 279113457.0, + "step": 10981 + }, + { + "epoch": 1.2060180101032287, + "grad_norm": 2.178945541381836, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.711786150932312, + "num_tokens": 279140144.0, + "step": 10982 + }, + { + "epoch": 1.2061278278058423, + "grad_norm": 2.2462780475616455, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7162599563598633, + "num_tokens": 279166094.0, + "step": 10983 + }, + { + "epoch": 1.206237645508456, + "grad_norm": 2.080935478210449, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7104991674423218, + "num_tokens": 279193719.0, + "step": 10984 + }, + { + "epoch": 1.2063474632110696, + "grad_norm": 2.389477252960205, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7222596406936646, + "num_tokens": 279216981.0, + "step": 10985 + }, + { + "epoch": 1.2064572809136833, + "grad_norm": 2.2744476795196533, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7132614850997925, + "num_tokens": 279241587.0, + "step": 10986 + }, + { + "epoch": 1.2065670986162969, + "grad_norm": 2.3108601570129395, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7242598533630371, + "num_tokens": 279265763.0, + "step": 10987 + }, + { + "epoch": 1.2066769163189106, + "grad_norm": 2.0669798851013184, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7333554029464722, + "num_tokens": 279292611.0, + "step": 10988 + }, + { + "epoch": 1.2067867340215244, + "grad_norm": 2.4473555088043213, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7325555086135864, + "num_tokens": 279312948.0, + "step": 10989 + }, + { + "epoch": 1.206896551724138, + "grad_norm": 2.4879555702209473, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7185478210449219, + "num_tokens": 279335206.0, + "step": 10990 + }, + { + "epoch": 1.2070063694267517, + "grad_norm": 2.4520583152770996, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7171708345413208, + "num_tokens": 279357488.0, + "step": 10991 + }, + { + "epoch": 1.2071161871293652, + "grad_norm": 2.3361222743988037, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.702915370464325, + "num_tokens": 279380925.0, + "step": 10992 + }, + { + "epoch": 1.207226004831979, + "grad_norm": 2.4014320373535156, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7294340133666992, + "num_tokens": 279403092.0, + "step": 10993 + }, + { + "epoch": 1.2073358225345925, + "grad_norm": 2.2078747749328613, + "learning_rate": 1e-06, + "loss": 0.9578, + "mean_token_accuracy": 0.7213680744171143, + "num_tokens": 279428886.0, + "step": 10994 + }, + { + "epoch": 1.2074456402372062, + "grad_norm": 2.461613416671753, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7299399971961975, + "num_tokens": 279452732.0, + "step": 10995 + }, + { + "epoch": 1.20755545793982, + "grad_norm": 2.6206862926483154, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7381296157836914, + "num_tokens": 279473336.0, + "step": 10996 + }, + { + "epoch": 1.2076652756424335, + "grad_norm": 2.3246142864227295, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.709222137928009, + "num_tokens": 279498574.0, + "step": 10997 + }, + { + "epoch": 1.2077750933450473, + "grad_norm": 2.2036795616149902, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7220078706741333, + "num_tokens": 279525055.0, + "step": 10998 + }, + { + "epoch": 1.2078849110476608, + "grad_norm": 2.656651735305786, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7303907871246338, + "num_tokens": 279544859.0, + "step": 10999 + }, + { + "epoch": 1.2079947287502746, + "grad_norm": 2.092705488204956, + "learning_rate": 1e-06, + "loss": 0.7932, + "mean_token_accuracy": 0.7501327991485596, + "num_tokens": 279570249.0, + "step": 11000 + }, + { + "epoch": 1.2081045464528881, + "grad_norm": 2.272026777267456, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7222675085067749, + "num_tokens": 279594746.0, + "step": 11001 + }, + { + "epoch": 1.2082143641555019, + "grad_norm": 2.2990288734436035, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7263450622558594, + "num_tokens": 279618188.0, + "step": 11002 + }, + { + "epoch": 1.2083241818581154, + "grad_norm": 2.185361862182617, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7214071750640869, + "num_tokens": 279646132.0, + "step": 11003 + }, + { + "epoch": 1.2084339995607292, + "grad_norm": 2.311983585357666, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.692969799041748, + "num_tokens": 279671275.0, + "step": 11004 + }, + { + "epoch": 1.208543817263343, + "grad_norm": 2.343269109725952, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7103047966957092, + "num_tokens": 279697390.0, + "step": 11005 + }, + { + "epoch": 1.2086536349659565, + "grad_norm": 2.2139272689819336, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7298346161842346, + "num_tokens": 279722457.0, + "step": 11006 + }, + { + "epoch": 1.2087634526685702, + "grad_norm": 1.9467730522155762, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7086910605430603, + "num_tokens": 279752994.0, + "step": 11007 + }, + { + "epoch": 1.2088732703711838, + "grad_norm": 2.2408206462860107, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7398033142089844, + "num_tokens": 279776852.0, + "step": 11008 + }, + { + "epoch": 1.2089830880737975, + "grad_norm": 2.1811015605926514, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7157360315322876, + "num_tokens": 279804849.0, + "step": 11009 + }, + { + "epoch": 1.2090929057764113, + "grad_norm": 2.4307456016540527, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7297268509864807, + "num_tokens": 279829312.0, + "step": 11010 + }, + { + "epoch": 1.2092027234790248, + "grad_norm": 2.098905563354492, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7366647720336914, + "num_tokens": 279855584.0, + "step": 11011 + }, + { + "epoch": 1.2093125411816386, + "grad_norm": 2.2404773235321045, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.721632182598114, + "num_tokens": 279879887.0, + "step": 11012 + }, + { + "epoch": 1.209422358884252, + "grad_norm": 2.207521438598633, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.711128294467926, + "num_tokens": 279905603.0, + "step": 11013 + }, + { + "epoch": 1.2095321765868658, + "grad_norm": 2.1241865158081055, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7121911644935608, + "num_tokens": 279934006.0, + "step": 11014 + }, + { + "epoch": 1.2096419942894794, + "grad_norm": 2.2219560146331787, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7035373449325562, + "num_tokens": 279959604.0, + "step": 11015 + }, + { + "epoch": 1.2097518119920931, + "grad_norm": 2.168217658996582, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7291290760040283, + "num_tokens": 279985025.0, + "step": 11016 + }, + { + "epoch": 1.2098616296947067, + "grad_norm": 2.3935112953186035, + "learning_rate": 1e-06, + "loss": 0.7908, + "mean_token_accuracy": 0.7549430131912231, + "num_tokens": 280006863.0, + "step": 11017 + }, + { + "epoch": 1.2099714473973204, + "grad_norm": 2.3936383724212646, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.712287425994873, + "num_tokens": 280028800.0, + "step": 11018 + }, + { + "epoch": 1.2100812650999342, + "grad_norm": 1.9640400409698486, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7134166955947876, + "num_tokens": 280060624.0, + "step": 11019 + }, + { + "epoch": 1.2101910828025477, + "grad_norm": 2.319575071334839, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7214557528495789, + "num_tokens": 280086169.0, + "step": 11020 + }, + { + "epoch": 1.2103009005051615, + "grad_norm": 2.2538366317749023, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6915552020072937, + "num_tokens": 280113960.0, + "step": 11021 + }, + { + "epoch": 1.210410718207775, + "grad_norm": 2.1709797382354736, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7451918125152588, + "num_tokens": 280139844.0, + "step": 11022 + }, + { + "epoch": 1.2105205359103888, + "grad_norm": 2.0776290893554688, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6974400877952576, + "num_tokens": 280168231.0, + "step": 11023 + }, + { + "epoch": 1.2106303536130025, + "grad_norm": 2.4290969371795654, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7219762206077576, + "num_tokens": 280189713.0, + "step": 11024 + }, + { + "epoch": 1.210740171315616, + "grad_norm": 2.2476232051849365, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7266538739204407, + "num_tokens": 280213991.0, + "step": 11025 + }, + { + "epoch": 1.2108499890182298, + "grad_norm": 2.3103041648864746, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7162573337554932, + "num_tokens": 280238675.0, + "step": 11026 + }, + { + "epoch": 1.2109598067208434, + "grad_norm": 2.0887322425842285, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7091705203056335, + "num_tokens": 280267476.0, + "step": 11027 + }, + { + "epoch": 1.2110696244234571, + "grad_norm": 2.501304864883423, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7408111095428467, + "num_tokens": 280289194.0, + "step": 11028 + }, + { + "epoch": 1.2111794421260706, + "grad_norm": 2.499490976333618, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7330846786499023, + "num_tokens": 280309043.0, + "step": 11029 + }, + { + "epoch": 1.2112892598286844, + "grad_norm": 2.477693796157837, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7186176776885986, + "num_tokens": 280332104.0, + "step": 11030 + }, + { + "epoch": 1.211399077531298, + "grad_norm": 1.9108942747116089, + "learning_rate": 1e-06, + "loss": 0.989, + "mean_token_accuracy": 0.6904737949371338, + "num_tokens": 280366922.0, + "step": 11031 + }, + { + "epoch": 1.2115088952339117, + "grad_norm": 2.350714921951294, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7146729230880737, + "num_tokens": 280391188.0, + "step": 11032 + }, + { + "epoch": 1.2116187129365255, + "grad_norm": 2.3146018981933594, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7381259202957153, + "num_tokens": 280415332.0, + "step": 11033 + }, + { + "epoch": 1.211728530639139, + "grad_norm": 1.9925427436828613, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7026346921920776, + "num_tokens": 280448607.0, + "step": 11034 + }, + { + "epoch": 1.2118383483417527, + "grad_norm": 2.179208517074585, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7020798921585083, + "num_tokens": 280476128.0, + "step": 11035 + }, + { + "epoch": 1.2119481660443663, + "grad_norm": 2.4718658924102783, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7172703742980957, + "num_tokens": 280499047.0, + "step": 11036 + }, + { + "epoch": 1.21205798374698, + "grad_norm": 2.1766350269317627, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.714438259601593, + "num_tokens": 280526463.0, + "step": 11037 + }, + { + "epoch": 1.2121678014495938, + "grad_norm": 2.4335880279541016, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7173290252685547, + "num_tokens": 280549255.0, + "step": 11038 + }, + { + "epoch": 1.2122776191522073, + "grad_norm": 2.6532979011535645, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.727562427520752, + "num_tokens": 280569321.0, + "step": 11039 + }, + { + "epoch": 1.212387436854821, + "grad_norm": 2.221529483795166, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7288631200790405, + "num_tokens": 280595118.0, + "step": 11040 + }, + { + "epoch": 1.2124972545574346, + "grad_norm": 2.3654472827911377, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7236967086791992, + "num_tokens": 280618292.0, + "step": 11041 + }, + { + "epoch": 1.2126070722600484, + "grad_norm": 2.2272069454193115, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7080436944961548, + "num_tokens": 280644999.0, + "step": 11042 + }, + { + "epoch": 1.212716889962662, + "grad_norm": 1.9823734760284424, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.730205774307251, + "num_tokens": 280676518.0, + "step": 11043 + }, + { + "epoch": 1.2128267076652757, + "grad_norm": 2.4618825912475586, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7101065516471863, + "num_tokens": 280699984.0, + "step": 11044 + }, + { + "epoch": 1.2129365253678892, + "grad_norm": 2.461968183517456, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7359910607337952, + "num_tokens": 280720305.0, + "step": 11045 + }, + { + "epoch": 1.213046343070503, + "grad_norm": 2.347820520401001, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7254201769828796, + "num_tokens": 280741960.0, + "step": 11046 + }, + { + "epoch": 1.2131561607731167, + "grad_norm": 2.0247743129730225, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7082406878471375, + "num_tokens": 280771405.0, + "step": 11047 + }, + { + "epoch": 1.2132659784757303, + "grad_norm": 2.188767194747925, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.6996573209762573, + "num_tokens": 280798106.0, + "step": 11048 + }, + { + "epoch": 1.213375796178344, + "grad_norm": 2.2850308418273926, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7287134528160095, + "num_tokens": 280822550.0, + "step": 11049 + }, + { + "epoch": 1.2134856138809575, + "grad_norm": 3.0209922790527344, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7271243333816528, + "num_tokens": 280837580.0, + "step": 11050 + }, + { + "epoch": 1.2135954315835713, + "grad_norm": 2.412947416305542, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.6965383291244507, + "num_tokens": 280861351.0, + "step": 11051 + }, + { + "epoch": 1.2137052492861848, + "grad_norm": 2.3365862369537354, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7154277563095093, + "num_tokens": 280886590.0, + "step": 11052 + }, + { + "epoch": 1.2138150669887986, + "grad_norm": 2.195330858230591, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6877748966217041, + "num_tokens": 280915018.0, + "step": 11053 + }, + { + "epoch": 1.2139248846914121, + "grad_norm": 1.9749832153320312, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7323148846626282, + "num_tokens": 280945632.0, + "step": 11054 + }, + { + "epoch": 1.2140347023940259, + "grad_norm": 2.0636229515075684, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7117288708686829, + "num_tokens": 280975099.0, + "step": 11055 + }, + { + "epoch": 1.2141445200966396, + "grad_norm": 2.376532793045044, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7210294008255005, + "num_tokens": 280997536.0, + "step": 11056 + }, + { + "epoch": 1.2142543377992532, + "grad_norm": 2.3244779109954834, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7275371551513672, + "num_tokens": 281022054.0, + "step": 11057 + }, + { + "epoch": 1.214364155501867, + "grad_norm": 2.345851182937622, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7042996883392334, + "num_tokens": 281045659.0, + "step": 11058 + }, + { + "epoch": 1.2144739732044805, + "grad_norm": 2.3732285499572754, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7092165946960449, + "num_tokens": 281069298.0, + "step": 11059 + }, + { + "epoch": 1.2145837909070942, + "grad_norm": 2.1045751571655273, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6947852373123169, + "num_tokens": 281097837.0, + "step": 11060 + }, + { + "epoch": 1.214693608609708, + "grad_norm": 2.4114179611206055, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.7223597764968872, + "num_tokens": 281121681.0, + "step": 11061 + }, + { + "epoch": 1.2148034263123215, + "grad_norm": 2.2944135665893555, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7123203277587891, + "num_tokens": 281146770.0, + "step": 11062 + }, + { + "epoch": 1.2149132440149353, + "grad_norm": 2.2653026580810547, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7275249361991882, + "num_tokens": 281170434.0, + "step": 11063 + }, + { + "epoch": 1.2150230617175488, + "grad_norm": 2.510838031768799, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7330858707427979, + "num_tokens": 281192364.0, + "step": 11064 + }, + { + "epoch": 1.2151328794201626, + "grad_norm": 2.204800844192505, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7260836362838745, + "num_tokens": 281218609.0, + "step": 11065 + }, + { + "epoch": 1.215242697122776, + "grad_norm": 2.3601882457733154, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7114530801773071, + "num_tokens": 281242399.0, + "step": 11066 + }, + { + "epoch": 1.2153525148253899, + "grad_norm": 2.465139627456665, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7119298577308655, + "num_tokens": 281264301.0, + "step": 11067 + }, + { + "epoch": 1.2154623325280034, + "grad_norm": 2.4737961292266846, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7359812259674072, + "num_tokens": 281287124.0, + "step": 11068 + }, + { + "epoch": 1.2155721502306172, + "grad_norm": 2.2312889099121094, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.7067681550979614, + "num_tokens": 281314771.0, + "step": 11069 + }, + { + "epoch": 1.215681967933231, + "grad_norm": 2.7610011100769043, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.72218918800354, + "num_tokens": 281333354.0, + "step": 11070 + }, + { + "epoch": 1.2157917856358444, + "grad_norm": 2.166522264480591, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7088215947151184, + "num_tokens": 281361352.0, + "step": 11071 + }, + { + "epoch": 1.2159016033384582, + "grad_norm": 2.094179153442383, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7160580158233643, + "num_tokens": 281390131.0, + "step": 11072 + }, + { + "epoch": 1.2160114210410717, + "grad_norm": 2.3336920738220215, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7256036996841431, + "num_tokens": 281414351.0, + "step": 11073 + }, + { + "epoch": 1.2161212387436855, + "grad_norm": 2.2138607501983643, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7168265581130981, + "num_tokens": 281441350.0, + "step": 11074 + }, + { + "epoch": 1.2162310564462993, + "grad_norm": 2.5598409175872803, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7370476722717285, + "num_tokens": 281463049.0, + "step": 11075 + }, + { + "epoch": 1.2163408741489128, + "grad_norm": 2.335400342941284, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7177723050117493, + "num_tokens": 281486874.0, + "step": 11076 + }, + { + "epoch": 1.2164506918515265, + "grad_norm": 2.5690436363220215, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7414817214012146, + "num_tokens": 281507851.0, + "step": 11077 + }, + { + "epoch": 1.21656050955414, + "grad_norm": 2.1277410984039307, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7328388690948486, + "num_tokens": 281534340.0, + "step": 11078 + }, + { + "epoch": 1.2166703272567538, + "grad_norm": 2.1660776138305664, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.734899640083313, + "num_tokens": 281559984.0, + "step": 11079 + }, + { + "epoch": 1.2167801449593674, + "grad_norm": 2.7062032222747803, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7220669984817505, + "num_tokens": 281581206.0, + "step": 11080 + }, + { + "epoch": 1.2168899626619811, + "grad_norm": 2.3846595287323, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7224908471107483, + "num_tokens": 281606610.0, + "step": 11081 + }, + { + "epoch": 1.2169997803645947, + "grad_norm": 2.551942825317383, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7419678568840027, + "num_tokens": 281626842.0, + "step": 11082 + }, + { + "epoch": 1.2171095980672084, + "grad_norm": 2.4400124549865723, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7108253836631775, + "num_tokens": 281651255.0, + "step": 11083 + }, + { + "epoch": 1.2172194157698222, + "grad_norm": 2.2000350952148438, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7102179527282715, + "num_tokens": 281677861.0, + "step": 11084 + }, + { + "epoch": 1.2173292334724357, + "grad_norm": 2.3158762454986572, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7086438536643982, + "num_tokens": 281703072.0, + "step": 11085 + }, + { + "epoch": 1.2174390511750495, + "grad_norm": 2.380051374435425, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7171153426170349, + "num_tokens": 281725885.0, + "step": 11086 + }, + { + "epoch": 1.217548868877663, + "grad_norm": 2.4458673000335693, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7193922996520996, + "num_tokens": 281748385.0, + "step": 11087 + }, + { + "epoch": 1.2176586865802768, + "grad_norm": 2.3126156330108643, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.738329291343689, + "num_tokens": 281773674.0, + "step": 11088 + }, + { + "epoch": 1.2177685042828905, + "grad_norm": 2.4898107051849365, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7343893647193909, + "num_tokens": 281793996.0, + "step": 11089 + }, + { + "epoch": 1.217878321985504, + "grad_norm": 2.4078550338745117, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7324306964874268, + "num_tokens": 281818204.0, + "step": 11090 + }, + { + "epoch": 1.2179881396881178, + "grad_norm": 2.495889902114868, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7216938734054565, + "num_tokens": 281838805.0, + "step": 11091 + }, + { + "epoch": 1.2180979573907313, + "grad_norm": 2.343629837036133, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7331526279449463, + "num_tokens": 281861006.0, + "step": 11092 + }, + { + "epoch": 1.218207775093345, + "grad_norm": 2.188445568084717, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7114758491516113, + "num_tokens": 281887433.0, + "step": 11093 + }, + { + "epoch": 1.2183175927959586, + "grad_norm": 1.8663551807403564, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7031390070915222, + "num_tokens": 281921823.0, + "step": 11094 + }, + { + "epoch": 1.2184274104985724, + "grad_norm": 2.5476417541503906, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7279582619667053, + "num_tokens": 281942047.0, + "step": 11095 + }, + { + "epoch": 1.218537228201186, + "grad_norm": 2.4296317100524902, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.709389865398407, + "num_tokens": 281965416.0, + "step": 11096 + }, + { + "epoch": 1.2186470459037997, + "grad_norm": 2.4083075523376465, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7083544135093689, + "num_tokens": 281989450.0, + "step": 11097 + }, + { + "epoch": 1.2187568636064134, + "grad_norm": 2.4435291290283203, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7129865288734436, + "num_tokens": 282014870.0, + "step": 11098 + }, + { + "epoch": 1.218866681309027, + "grad_norm": 2.4925765991210938, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7335221171379089, + "num_tokens": 282035192.0, + "step": 11099 + }, + { + "epoch": 1.2189764990116407, + "grad_norm": 2.5160930156707764, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7326580286026001, + "num_tokens": 282057374.0, + "step": 11100 + }, + { + "epoch": 1.2190863167142543, + "grad_norm": 2.4283816814422607, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7115645408630371, + "num_tokens": 282080676.0, + "step": 11101 + }, + { + "epoch": 1.219196134416868, + "grad_norm": 2.3132383823394775, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.721916675567627, + "num_tokens": 282105879.0, + "step": 11102 + }, + { + "epoch": 1.2193059521194818, + "grad_norm": 2.3591487407684326, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7003400325775146, + "num_tokens": 282131958.0, + "step": 11103 + }, + { + "epoch": 1.2194157698220953, + "grad_norm": 2.199739456176758, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7186630368232727, + "num_tokens": 282157708.0, + "step": 11104 + }, + { + "epoch": 1.219525587524709, + "grad_norm": 2.249760389328003, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.6916931867599487, + "num_tokens": 282181936.0, + "step": 11105 + }, + { + "epoch": 1.2196354052273226, + "grad_norm": 2.4614741802215576, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7022922039031982, + "num_tokens": 282204278.0, + "step": 11106 + }, + { + "epoch": 1.2197452229299364, + "grad_norm": 2.05279803276062, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7180138230323792, + "num_tokens": 282233361.0, + "step": 11107 + }, + { + "epoch": 1.21985504063255, + "grad_norm": 2.440429449081421, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7100448608398438, + "num_tokens": 282256981.0, + "step": 11108 + }, + { + "epoch": 1.2199648583351637, + "grad_norm": 2.123624563217163, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7097536325454712, + "num_tokens": 282287444.0, + "step": 11109 + }, + { + "epoch": 1.2200746760377772, + "grad_norm": 2.6087751388549805, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7232826948165894, + "num_tokens": 282307448.0, + "step": 11110 + }, + { + "epoch": 1.220184493740391, + "grad_norm": 2.426812171936035, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7246894836425781, + "num_tokens": 282329676.0, + "step": 11111 + }, + { + "epoch": 1.2202943114430047, + "grad_norm": 2.6273446083068848, + "learning_rate": 1e-06, + "loss": 0.9895, + "mean_token_accuracy": 0.6962255239486694, + "num_tokens": 282351255.0, + "step": 11112 + }, + { + "epoch": 1.2204041291456182, + "grad_norm": 2.2495312690734863, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7354648113250732, + "num_tokens": 282376207.0, + "step": 11113 + }, + { + "epoch": 1.220513946848232, + "grad_norm": 2.216359853744507, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7052455544471741, + "num_tokens": 282404704.0, + "step": 11114 + }, + { + "epoch": 1.2206237645508455, + "grad_norm": 2.1944656372070312, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7157400846481323, + "num_tokens": 282431722.0, + "step": 11115 + }, + { + "epoch": 1.2207335822534593, + "grad_norm": 2.5621907711029053, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7202860116958618, + "num_tokens": 282451385.0, + "step": 11116 + }, + { + "epoch": 1.2208433999560728, + "grad_norm": 2.2214794158935547, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7276434898376465, + "num_tokens": 282477473.0, + "step": 11117 + }, + { + "epoch": 1.2209532176586866, + "grad_norm": 2.5593996047973633, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7151278257369995, + "num_tokens": 282499656.0, + "step": 11118 + }, + { + "epoch": 1.2210630353613001, + "grad_norm": 2.37505841255188, + "learning_rate": 1e-06, + "loss": 0.8367, + "mean_token_accuracy": 0.7343088984489441, + "num_tokens": 282522861.0, + "step": 11119 + }, + { + "epoch": 1.2211728530639139, + "grad_norm": 2.461179494857788, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7078094482421875, + "num_tokens": 282545669.0, + "step": 11120 + }, + { + "epoch": 1.2212826707665276, + "grad_norm": 2.0356826782226562, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7212810516357422, + "num_tokens": 282576302.0, + "step": 11121 + }, + { + "epoch": 1.2213924884691412, + "grad_norm": 2.1733920574188232, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7027117013931274, + "num_tokens": 282603481.0, + "step": 11122 + }, + { + "epoch": 1.221502306171755, + "grad_norm": 2.3021206855773926, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7259721755981445, + "num_tokens": 282628986.0, + "step": 11123 + }, + { + "epoch": 1.2216121238743685, + "grad_norm": 2.2559750080108643, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7106919288635254, + "num_tokens": 282655171.0, + "step": 11124 + }, + { + "epoch": 1.2217219415769822, + "grad_norm": 2.3564376831054688, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.6982029676437378, + "num_tokens": 282679259.0, + "step": 11125 + }, + { + "epoch": 1.221831759279596, + "grad_norm": 2.107781171798706, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7488521337509155, + "num_tokens": 282707452.0, + "step": 11126 + }, + { + "epoch": 1.2219415769822095, + "grad_norm": 2.186675786972046, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6970425844192505, + "num_tokens": 282735872.0, + "step": 11127 + }, + { + "epoch": 1.2220513946848233, + "grad_norm": 2.342238187789917, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7150484919548035, + "num_tokens": 282760512.0, + "step": 11128 + }, + { + "epoch": 1.2221612123874368, + "grad_norm": 2.1463050842285156, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7068846225738525, + "num_tokens": 282789082.0, + "step": 11129 + }, + { + "epoch": 1.2222710300900506, + "grad_norm": 2.5889508724212646, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7135107517242432, + "num_tokens": 282810607.0, + "step": 11130 + }, + { + "epoch": 1.222380847792664, + "grad_norm": 2.5017099380493164, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7265941500663757, + "num_tokens": 282832458.0, + "step": 11131 + }, + { + "epoch": 1.2224906654952779, + "grad_norm": 2.2608819007873535, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6905864477157593, + "num_tokens": 282859328.0, + "step": 11132 + }, + { + "epoch": 1.2226004831978914, + "grad_norm": 2.2252681255340576, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.7349261045455933, + "num_tokens": 282884414.0, + "step": 11133 + }, + { + "epoch": 1.2227103009005051, + "grad_norm": 2.399974822998047, + "learning_rate": 1e-06, + "loss": 0.8546, + "mean_token_accuracy": 0.727263331413269, + "num_tokens": 282906766.0, + "step": 11134 + }, + { + "epoch": 1.222820118603119, + "grad_norm": 2.3786041736602783, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.730331301689148, + "num_tokens": 282930257.0, + "step": 11135 + }, + { + "epoch": 1.2229299363057324, + "grad_norm": 2.1490752696990967, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7032959461212158, + "num_tokens": 282959123.0, + "step": 11136 + }, + { + "epoch": 1.2230397540083462, + "grad_norm": 1.8627514839172363, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7200212478637695, + "num_tokens": 282992469.0, + "step": 11137 + }, + { + "epoch": 1.2231495717109597, + "grad_norm": 1.9003585577011108, + "learning_rate": 1e-06, + "loss": 1.0104, + "mean_token_accuracy": 0.6879894137382507, + "num_tokens": 283027091.0, + "step": 11138 + }, + { + "epoch": 1.2232593894135735, + "grad_norm": 2.3375349044799805, + "learning_rate": 1e-06, + "loss": 0.9811, + "mean_token_accuracy": 0.7015863656997681, + "num_tokens": 283050383.0, + "step": 11139 + }, + { + "epoch": 1.2233692071161872, + "grad_norm": 2.20493745803833, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.6990467309951782, + "num_tokens": 283076709.0, + "step": 11140 + }, + { + "epoch": 1.2234790248188008, + "grad_norm": 2.312462329864502, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7239187359809875, + "num_tokens": 283099568.0, + "step": 11141 + }, + { + "epoch": 1.2235888425214145, + "grad_norm": 2.144890069961548, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7023217678070068, + "num_tokens": 283126902.0, + "step": 11142 + }, + { + "epoch": 1.223698660224028, + "grad_norm": 2.4059815406799316, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7142813205718994, + "num_tokens": 283149679.0, + "step": 11143 + }, + { + "epoch": 1.2238084779266418, + "grad_norm": 2.259092092514038, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7289209365844727, + "num_tokens": 283175241.0, + "step": 11144 + }, + { + "epoch": 1.2239182956292554, + "grad_norm": 2.2896194458007812, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7174496054649353, + "num_tokens": 283200922.0, + "step": 11145 + }, + { + "epoch": 1.2240281133318691, + "grad_norm": 2.5184144973754883, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7239937782287598, + "num_tokens": 283224049.0, + "step": 11146 + }, + { + "epoch": 1.2241379310344827, + "grad_norm": 2.155910015106201, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7379502058029175, + "num_tokens": 283251047.0, + "step": 11147 + }, + { + "epoch": 1.2242477487370964, + "grad_norm": 2.254936933517456, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7460765242576599, + "num_tokens": 283274807.0, + "step": 11148 + }, + { + "epoch": 1.2243575664397102, + "grad_norm": 2.44107723236084, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7147188782691956, + "num_tokens": 283299400.0, + "step": 11149 + }, + { + "epoch": 1.2244673841423237, + "grad_norm": 2.346465587615967, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.720549464225769, + "num_tokens": 283325181.0, + "step": 11150 + }, + { + "epoch": 1.2245772018449375, + "grad_norm": 2.2730860710144043, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7080307006835938, + "num_tokens": 283351437.0, + "step": 11151 + }, + { + "epoch": 1.224687019547551, + "grad_norm": 2.0155112743377686, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.720183253288269, + "num_tokens": 283382629.0, + "step": 11152 + }, + { + "epoch": 1.2247968372501647, + "grad_norm": 2.284710168838501, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7271857261657715, + "num_tokens": 283406556.0, + "step": 11153 + }, + { + "epoch": 1.2249066549527785, + "grad_norm": 2.163473606109619, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7238510251045227, + "num_tokens": 283432938.0, + "step": 11154 + }, + { + "epoch": 1.225016472655392, + "grad_norm": 2.085097312927246, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7175735831260681, + "num_tokens": 283461976.0, + "step": 11155 + }, + { + "epoch": 1.2251262903580058, + "grad_norm": 2.3366196155548096, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.7056303024291992, + "num_tokens": 283484673.0, + "step": 11156 + }, + { + "epoch": 1.2252361080606193, + "grad_norm": 2.276136636734009, + "learning_rate": 1e-06, + "loss": 0.9935, + "mean_token_accuracy": 0.6991642713546753, + "num_tokens": 283510075.0, + "step": 11157 + }, + { + "epoch": 1.225345925763233, + "grad_norm": 2.256889820098877, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7198624610900879, + "num_tokens": 283536904.0, + "step": 11158 + }, + { + "epoch": 1.2254557434658466, + "grad_norm": 2.16987681388855, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7103731632232666, + "num_tokens": 283562384.0, + "step": 11159 + }, + { + "epoch": 1.2255655611684604, + "grad_norm": 2.3351502418518066, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7351241707801819, + "num_tokens": 283585648.0, + "step": 11160 + }, + { + "epoch": 1.225675378871074, + "grad_norm": 2.3025143146514893, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7331953048706055, + "num_tokens": 283607933.0, + "step": 11161 + }, + { + "epoch": 1.2257851965736877, + "grad_norm": 2.201155185699463, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7178229093551636, + "num_tokens": 283633167.0, + "step": 11162 + }, + { + "epoch": 1.2258950142763014, + "grad_norm": 2.3125741481781006, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7252169847488403, + "num_tokens": 283657077.0, + "step": 11163 + }, + { + "epoch": 1.226004831978915, + "grad_norm": 2.4781694412231445, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.73674476146698, + "num_tokens": 283677712.0, + "step": 11164 + }, + { + "epoch": 1.2261146496815287, + "grad_norm": 2.5272343158721924, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7110457420349121, + "num_tokens": 283698514.0, + "step": 11165 + }, + { + "epoch": 1.2262244673841423, + "grad_norm": 2.131560802459717, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7168826460838318, + "num_tokens": 283728214.0, + "step": 11166 + }, + { + "epoch": 1.226334285086756, + "grad_norm": 2.6382927894592285, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7241277694702148, + "num_tokens": 283750324.0, + "step": 11167 + }, + { + "epoch": 1.2264441027893696, + "grad_norm": 2.054992914199829, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7196887731552124, + "num_tokens": 283779192.0, + "step": 11168 + }, + { + "epoch": 1.2265539204919833, + "grad_norm": 2.037590980529785, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7051172256469727, + "num_tokens": 283808281.0, + "step": 11169 + }, + { + "epoch": 1.226663738194597, + "grad_norm": 2.201688766479492, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7094895243644714, + "num_tokens": 283834438.0, + "step": 11170 + }, + { + "epoch": 1.2267735558972106, + "grad_norm": 2.360799789428711, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7248536348342896, + "num_tokens": 283858838.0, + "step": 11171 + }, + { + "epoch": 1.2268833735998244, + "grad_norm": 2.238795757293701, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7181742191314697, + "num_tokens": 283886032.0, + "step": 11172 + }, + { + "epoch": 1.226993191302438, + "grad_norm": 2.0951621532440186, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.695281445980072, + "num_tokens": 283916125.0, + "step": 11173 + }, + { + "epoch": 1.2271030090050516, + "grad_norm": 2.2721996307373047, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7169529795646667, + "num_tokens": 283942427.0, + "step": 11174 + }, + { + "epoch": 1.2272128267076652, + "grad_norm": 2.3745126724243164, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7142702341079712, + "num_tokens": 283967708.0, + "step": 11175 + }, + { + "epoch": 1.227322644410279, + "grad_norm": 1.969014048576355, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7278532981872559, + "num_tokens": 283997946.0, + "step": 11176 + }, + { + "epoch": 1.2274324621128927, + "grad_norm": 2.212765693664551, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7052363157272339, + "num_tokens": 284025224.0, + "step": 11177 + }, + { + "epoch": 1.2275422798155062, + "grad_norm": 2.2370195388793945, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7054380178451538, + "num_tokens": 284054460.0, + "step": 11178 + }, + { + "epoch": 1.22765209751812, + "grad_norm": 2.1265501976013184, + "learning_rate": 1e-06, + "loss": 0.9823, + "mean_token_accuracy": 0.6981000304222107, + "num_tokens": 284085080.0, + "step": 11179 + }, + { + "epoch": 1.2277619152207335, + "grad_norm": 2.3026022911071777, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7273368835449219, + "num_tokens": 284110002.0, + "step": 11180 + }, + { + "epoch": 1.2278717329233473, + "grad_norm": 2.097902536392212, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6974771618843079, + "num_tokens": 284139513.0, + "step": 11181 + }, + { + "epoch": 1.2279815506259608, + "grad_norm": 2.4454638957977295, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.7178177237510681, + "num_tokens": 284162448.0, + "step": 11182 + }, + { + "epoch": 1.2280913683285746, + "grad_norm": 2.2903010845184326, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7135162353515625, + "num_tokens": 284187067.0, + "step": 11183 + }, + { + "epoch": 1.228201186031188, + "grad_norm": 2.3417792320251465, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.706614077091217, + "num_tokens": 284211847.0, + "step": 11184 + }, + { + "epoch": 1.2283110037338019, + "grad_norm": 2.6271910667419434, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7244142293930054, + "num_tokens": 284230822.0, + "step": 11185 + }, + { + "epoch": 1.2284208214364156, + "grad_norm": 2.0554661750793457, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7214102745056152, + "num_tokens": 284260859.0, + "step": 11186 + }, + { + "epoch": 1.2285306391390292, + "grad_norm": 2.3229339122772217, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7135053873062134, + "num_tokens": 284284967.0, + "step": 11187 + }, + { + "epoch": 1.228640456841643, + "grad_norm": 2.6407315731048584, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7015821933746338, + "num_tokens": 284306269.0, + "step": 11188 + }, + { + "epoch": 1.2287502745442564, + "grad_norm": 2.508514404296875, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7229181528091431, + "num_tokens": 284327327.0, + "step": 11189 + }, + { + "epoch": 1.2288600922468702, + "grad_norm": 2.2440719604492188, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7080795168876648, + "num_tokens": 284353512.0, + "step": 11190 + }, + { + "epoch": 1.228969909949484, + "grad_norm": 2.3860158920288086, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7143533229827881, + "num_tokens": 284378153.0, + "step": 11191 + }, + { + "epoch": 1.2290797276520975, + "grad_norm": 2.2837331295013428, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7181866765022278, + "num_tokens": 284403605.0, + "step": 11192 + }, + { + "epoch": 1.2291895453547113, + "grad_norm": 2.231306314468384, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6990330219268799, + "num_tokens": 284428843.0, + "step": 11193 + }, + { + "epoch": 1.2292993630573248, + "grad_norm": 2.154449224472046, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7339598536491394, + "num_tokens": 284454475.0, + "step": 11194 + }, + { + "epoch": 1.2294091807599385, + "grad_norm": 2.0481269359588623, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7167192697525024, + "num_tokens": 284481133.0, + "step": 11195 + }, + { + "epoch": 1.229518998462552, + "grad_norm": 2.277188539505005, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7055310606956482, + "num_tokens": 284506814.0, + "step": 11196 + }, + { + "epoch": 1.2296288161651658, + "grad_norm": 2.3606438636779785, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7409446239471436, + "num_tokens": 284529657.0, + "step": 11197 + }, + { + "epoch": 1.2297386338677794, + "grad_norm": 2.2408339977264404, + "learning_rate": 1e-06, + "loss": 0.9537, + "mean_token_accuracy": 0.7080407738685608, + "num_tokens": 284555339.0, + "step": 11198 + }, + { + "epoch": 1.2298484515703931, + "grad_norm": 2.5924196243286133, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7172160148620605, + "num_tokens": 284576169.0, + "step": 11199 + }, + { + "epoch": 1.2299582692730069, + "grad_norm": 2.311663866043091, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7148511409759521, + "num_tokens": 284600170.0, + "step": 11200 + }, + { + "epoch": 1.2300680869756204, + "grad_norm": 2.15750789642334, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7169848084449768, + "num_tokens": 284629123.0, + "step": 11201 + }, + { + "epoch": 1.2301779046782342, + "grad_norm": 2.305368185043335, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7048153877258301, + "num_tokens": 284652668.0, + "step": 11202 + }, + { + "epoch": 1.2302877223808477, + "grad_norm": 2.249152421951294, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7179129123687744, + "num_tokens": 284677819.0, + "step": 11203 + }, + { + "epoch": 1.2303975400834615, + "grad_norm": 2.3415966033935547, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7101914882659912, + "num_tokens": 284702046.0, + "step": 11204 + }, + { + "epoch": 1.2305073577860752, + "grad_norm": 1.9686501026153564, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.6989483833312988, + "num_tokens": 284736946.0, + "step": 11205 + }, + { + "epoch": 1.2306171754886888, + "grad_norm": 2.2843544483184814, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7320762276649475, + "num_tokens": 284761708.0, + "step": 11206 + }, + { + "epoch": 1.2307269931913025, + "grad_norm": 2.4088244438171387, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7008447647094727, + "num_tokens": 284785770.0, + "step": 11207 + }, + { + "epoch": 1.230836810893916, + "grad_norm": 2.020956039428711, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.707836389541626, + "num_tokens": 284816271.0, + "step": 11208 + }, + { + "epoch": 1.2309466285965298, + "grad_norm": 2.1395325660705566, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7363764047622681, + "num_tokens": 284842535.0, + "step": 11209 + }, + { + "epoch": 1.2310564462991433, + "grad_norm": 2.273871421813965, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7076383829116821, + "num_tokens": 284869210.0, + "step": 11210 + }, + { + "epoch": 1.231166264001757, + "grad_norm": 2.601186752319336, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7158705592155457, + "num_tokens": 284888187.0, + "step": 11211 + }, + { + "epoch": 1.2312760817043706, + "grad_norm": 2.2839126586914062, + "learning_rate": 1e-06, + "loss": 0.8172, + "mean_token_accuracy": 0.7408046126365662, + "num_tokens": 284911886.0, + "step": 11212 + }, + { + "epoch": 1.2313858994069844, + "grad_norm": 2.344942331314087, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7302545309066772, + "num_tokens": 284936112.0, + "step": 11213 + }, + { + "epoch": 1.2314957171095982, + "grad_norm": 2.1464264392852783, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7091110944747925, + "num_tokens": 284962370.0, + "step": 11214 + }, + { + "epoch": 1.2316055348122117, + "grad_norm": 2.2013750076293945, + "learning_rate": 1e-06, + "loss": 0.9847, + "mean_token_accuracy": 0.698262631893158, + "num_tokens": 284989041.0, + "step": 11215 + }, + { + "epoch": 1.2317153525148254, + "grad_norm": 2.35762095451355, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7081787586212158, + "num_tokens": 285012657.0, + "step": 11216 + }, + { + "epoch": 1.231825170217439, + "grad_norm": 2.0162570476531982, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7212297916412354, + "num_tokens": 285044697.0, + "step": 11217 + }, + { + "epoch": 1.2319349879200527, + "grad_norm": 2.3052849769592285, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.728249728679657, + "num_tokens": 285067297.0, + "step": 11218 + }, + { + "epoch": 1.2320448056226665, + "grad_norm": 2.520156145095825, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7281368970870972, + "num_tokens": 285088594.0, + "step": 11219 + }, + { + "epoch": 1.23215462332528, + "grad_norm": 2.048860788345337, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.730201780796051, + "num_tokens": 285117282.0, + "step": 11220 + }, + { + "epoch": 1.2322644410278938, + "grad_norm": 2.4407482147216797, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.729118824005127, + "num_tokens": 285139031.0, + "step": 11221 + }, + { + "epoch": 1.2323742587305073, + "grad_norm": 2.3818514347076416, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.7409954071044922, + "num_tokens": 285162837.0, + "step": 11222 + }, + { + "epoch": 1.232484076433121, + "grad_norm": 2.7407729625701904, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.735696017742157, + "num_tokens": 285181034.0, + "step": 11223 + }, + { + "epoch": 1.2325938941357346, + "grad_norm": 2.340296506881714, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7088589668273926, + "num_tokens": 285206180.0, + "step": 11224 + }, + { + "epoch": 1.2327037118383484, + "grad_norm": 2.065567970275879, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7183060646057129, + "num_tokens": 285235660.0, + "step": 11225 + }, + { + "epoch": 1.232813529540962, + "grad_norm": 2.1550920009613037, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7130007147789001, + "num_tokens": 285266363.0, + "step": 11226 + }, + { + "epoch": 1.2329233472435757, + "grad_norm": 2.3006350994110107, + "learning_rate": 1e-06, + "loss": 1.0293, + "mean_token_accuracy": 0.6940591335296631, + "num_tokens": 285292566.0, + "step": 11227 + }, + { + "epoch": 1.2330331649461894, + "grad_norm": 2.2183279991149902, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7196669578552246, + "num_tokens": 285317851.0, + "step": 11228 + }, + { + "epoch": 1.233142982648803, + "grad_norm": 2.21285343170166, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.7120146751403809, + "num_tokens": 285344009.0, + "step": 11229 + }, + { + "epoch": 1.2332528003514167, + "grad_norm": 2.6946394443511963, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7379563450813293, + "num_tokens": 285363108.0, + "step": 11230 + }, + { + "epoch": 1.2333626180540302, + "grad_norm": 2.5505616664886475, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.702018141746521, + "num_tokens": 285384196.0, + "step": 11231 + }, + { + "epoch": 1.233472435756644, + "grad_norm": 2.2555031776428223, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7191253900527954, + "num_tokens": 285409590.0, + "step": 11232 + }, + { + "epoch": 1.2335822534592575, + "grad_norm": 2.2307820320129395, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.708838701248169, + "num_tokens": 285434616.0, + "step": 11233 + }, + { + "epoch": 1.2336920711618713, + "grad_norm": 2.3072030544281006, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7338440418243408, + "num_tokens": 285460539.0, + "step": 11234 + }, + { + "epoch": 1.2338018888644848, + "grad_norm": 2.190311908721924, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7261753082275391, + "num_tokens": 285486687.0, + "step": 11235 + }, + { + "epoch": 1.2339117065670986, + "grad_norm": 2.5167016983032227, + "learning_rate": 1e-06, + "loss": 1.0079, + "mean_token_accuracy": 0.6962035894393921, + "num_tokens": 285510350.0, + "step": 11236 + }, + { + "epoch": 1.2340215242697123, + "grad_norm": 2.423328161239624, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7007598876953125, + "num_tokens": 285534853.0, + "step": 11237 + }, + { + "epoch": 1.2341313419723259, + "grad_norm": 2.1051716804504395, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7119781970977783, + "num_tokens": 285564939.0, + "step": 11238 + }, + { + "epoch": 1.2342411596749396, + "grad_norm": 2.3754448890686035, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7151890993118286, + "num_tokens": 285588189.0, + "step": 11239 + }, + { + "epoch": 1.2343509773775532, + "grad_norm": 2.758042812347412, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7248658537864685, + "num_tokens": 285605403.0, + "step": 11240 + }, + { + "epoch": 1.234460795080167, + "grad_norm": 2.409144639968872, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7050992846488953, + "num_tokens": 285630259.0, + "step": 11241 + }, + { + "epoch": 1.2345706127827807, + "grad_norm": 2.0795912742614746, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7238695621490479, + "num_tokens": 285657958.0, + "step": 11242 + }, + { + "epoch": 1.2346804304853942, + "grad_norm": 2.197209119796753, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.707987368106842, + "num_tokens": 285684954.0, + "step": 11243 + }, + { + "epoch": 1.234790248188008, + "grad_norm": 2.357065439224243, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.6971513032913208, + "num_tokens": 285708485.0, + "step": 11244 + }, + { + "epoch": 1.2349000658906215, + "grad_norm": 2.312199115753174, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7276668548583984, + "num_tokens": 285730832.0, + "step": 11245 + }, + { + "epoch": 1.2350098835932353, + "grad_norm": 2.3128573894500732, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7328938841819763, + "num_tokens": 285755071.0, + "step": 11246 + }, + { + "epoch": 1.2351197012958488, + "grad_norm": 2.0495030879974365, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7083660960197449, + "num_tokens": 285786153.0, + "step": 11247 + }, + { + "epoch": 1.2352295189984626, + "grad_norm": 2.105008363723755, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7033490538597107, + "num_tokens": 285813185.0, + "step": 11248 + }, + { + "epoch": 1.235339336701076, + "grad_norm": 2.3095996379852295, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.713196873664856, + "num_tokens": 285837012.0, + "step": 11249 + }, + { + "epoch": 1.2354491544036899, + "grad_norm": 2.267615556716919, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7219369411468506, + "num_tokens": 285862218.0, + "step": 11250 + }, + { + "epoch": 1.2355589721063036, + "grad_norm": 2.065887689590454, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7299244999885559, + "num_tokens": 285893522.0, + "step": 11251 + }, + { + "epoch": 1.2356687898089171, + "grad_norm": 2.4338879585266113, + "learning_rate": 1e-06, + "loss": 1.0081, + "mean_token_accuracy": 0.6957755088806152, + "num_tokens": 285916765.0, + "step": 11252 + }, + { + "epoch": 1.235778607511531, + "grad_norm": 2.045348882675171, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7065461874008179, + "num_tokens": 285946752.0, + "step": 11253 + }, + { + "epoch": 1.2358884252141444, + "grad_norm": 2.266263246536255, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7022899389266968, + "num_tokens": 285972466.0, + "step": 11254 + }, + { + "epoch": 1.2359982429167582, + "grad_norm": 2.394646406173706, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7081104516983032, + "num_tokens": 285997274.0, + "step": 11255 + }, + { + "epoch": 1.236108060619372, + "grad_norm": 2.699307680130005, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7303369045257568, + "num_tokens": 286015239.0, + "step": 11256 + }, + { + "epoch": 1.2362178783219855, + "grad_norm": 2.2597358226776123, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7108389139175415, + "num_tokens": 286039821.0, + "step": 11257 + }, + { + "epoch": 1.2363276960245992, + "grad_norm": 2.1434786319732666, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7110490202903748, + "num_tokens": 286068456.0, + "step": 11258 + }, + { + "epoch": 1.2364375137272128, + "grad_norm": 2.2351717948913574, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.71360182762146, + "num_tokens": 286094154.0, + "step": 11259 + }, + { + "epoch": 1.2365473314298265, + "grad_norm": 2.175522804260254, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7101677656173706, + "num_tokens": 286120167.0, + "step": 11260 + }, + { + "epoch": 1.23665714913244, + "grad_norm": 2.2509560585021973, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.703044593334198, + "num_tokens": 286146764.0, + "step": 11261 + }, + { + "epoch": 1.2367669668350538, + "grad_norm": 2.1202964782714844, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7157216668128967, + "num_tokens": 286176221.0, + "step": 11262 + }, + { + "epoch": 1.2368767845376674, + "grad_norm": 2.236208438873291, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.7091081142425537, + "num_tokens": 286204965.0, + "step": 11263 + }, + { + "epoch": 1.2369866022402811, + "grad_norm": 2.3102383613586426, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7338110208511353, + "num_tokens": 286229103.0, + "step": 11264 + }, + { + "epoch": 1.2370964199428949, + "grad_norm": 2.17543625831604, + "learning_rate": 1e-06, + "loss": 1.0419, + "mean_token_accuracy": 0.6897225975990295, + "num_tokens": 286256202.0, + "step": 11265 + }, + { + "epoch": 1.2372062376455084, + "grad_norm": 2.3486130237579346, + "learning_rate": 1e-06, + "loss": 0.7714, + "mean_token_accuracy": 0.7609190940856934, + "num_tokens": 286278831.0, + "step": 11266 + }, + { + "epoch": 1.2373160553481222, + "grad_norm": 2.379486083984375, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.714309811592102, + "num_tokens": 286301626.0, + "step": 11267 + }, + { + "epoch": 1.2374258730507357, + "grad_norm": 2.3197665214538574, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7242962121963501, + "num_tokens": 286324889.0, + "step": 11268 + }, + { + "epoch": 1.2375356907533495, + "grad_norm": 2.2763586044311523, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7219637036323547, + "num_tokens": 286350759.0, + "step": 11269 + }, + { + "epoch": 1.2376455084559632, + "grad_norm": 2.26263427734375, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7190165519714355, + "num_tokens": 286375320.0, + "step": 11270 + }, + { + "epoch": 1.2377553261585768, + "grad_norm": 2.1882576942443848, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7095722556114197, + "num_tokens": 286402231.0, + "step": 11271 + }, + { + "epoch": 1.2378651438611905, + "grad_norm": 2.38836932182312, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7260152101516724, + "num_tokens": 286426230.0, + "step": 11272 + }, + { + "epoch": 1.237974961563804, + "grad_norm": 2.2833828926086426, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7226704359054565, + "num_tokens": 286451277.0, + "step": 11273 + }, + { + "epoch": 1.2380847792664178, + "grad_norm": 2.4693853855133057, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7228249311447144, + "num_tokens": 286472597.0, + "step": 11274 + }, + { + "epoch": 1.2381945969690313, + "grad_norm": 2.3724725246429443, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7276545763015747, + "num_tokens": 286494365.0, + "step": 11275 + }, + { + "epoch": 1.238304414671645, + "grad_norm": 2.13686203956604, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7128344178199768, + "num_tokens": 286521646.0, + "step": 11276 + }, + { + "epoch": 1.2384142323742586, + "grad_norm": 2.2410731315612793, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7176510691642761, + "num_tokens": 286547709.0, + "step": 11277 + }, + { + "epoch": 1.2385240500768724, + "grad_norm": 2.672869920730591, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7052930593490601, + "num_tokens": 286567768.0, + "step": 11278 + }, + { + "epoch": 1.2386338677794861, + "grad_norm": 2.5981829166412354, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7449963688850403, + "num_tokens": 286587351.0, + "step": 11279 + }, + { + "epoch": 1.2387436854820997, + "grad_norm": 2.494499921798706, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7201492190361023, + "num_tokens": 286608706.0, + "step": 11280 + }, + { + "epoch": 1.2388535031847134, + "grad_norm": 2.0387210845947266, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7139395475387573, + "num_tokens": 286640202.0, + "step": 11281 + }, + { + "epoch": 1.238963320887327, + "grad_norm": 2.043304443359375, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7195943593978882, + "num_tokens": 286667581.0, + "step": 11282 + }, + { + "epoch": 1.2390731385899407, + "grad_norm": 2.487368106842041, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7108618021011353, + "num_tokens": 286688668.0, + "step": 11283 + }, + { + "epoch": 1.2391829562925545, + "grad_norm": 2.225132942199707, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7235153913497925, + "num_tokens": 286714118.0, + "step": 11284 + }, + { + "epoch": 1.239292773995168, + "grad_norm": 2.3590996265411377, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7075026035308838, + "num_tokens": 286739749.0, + "step": 11285 + }, + { + "epoch": 1.2394025916977818, + "grad_norm": 2.140643835067749, + "learning_rate": 1e-06, + "loss": 1.0422, + "mean_token_accuracy": 0.6778010129928589, + "num_tokens": 286770205.0, + "step": 11286 + }, + { + "epoch": 1.2395124094003953, + "grad_norm": 2.2521402835845947, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7069575786590576, + "num_tokens": 286795438.0, + "step": 11287 + }, + { + "epoch": 1.239622227103009, + "grad_norm": 2.501613140106201, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7268161773681641, + "num_tokens": 286816283.0, + "step": 11288 + }, + { + "epoch": 1.2397320448056226, + "grad_norm": 2.1798136234283447, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7137082815170288, + "num_tokens": 286843310.0, + "step": 11289 + }, + { + "epoch": 1.2398418625082364, + "grad_norm": 2.2475454807281494, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7262657880783081, + "num_tokens": 286872105.0, + "step": 11290 + }, + { + "epoch": 1.23995168021085, + "grad_norm": 2.0467166900634766, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7320895195007324, + "num_tokens": 286899522.0, + "step": 11291 + }, + { + "epoch": 1.2400614979134637, + "grad_norm": 2.381157398223877, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7281815409660339, + "num_tokens": 286922873.0, + "step": 11292 + }, + { + "epoch": 1.2401713156160774, + "grad_norm": 2.3005478382110596, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7267360091209412, + "num_tokens": 286947779.0, + "step": 11293 + }, + { + "epoch": 1.240281133318691, + "grad_norm": 2.3710954189300537, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7064189910888672, + "num_tokens": 286973633.0, + "step": 11294 + }, + { + "epoch": 1.2403909510213047, + "grad_norm": 2.2051117420196533, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6921451091766357, + "num_tokens": 287002115.0, + "step": 11295 + }, + { + "epoch": 1.2405007687239182, + "grad_norm": 2.4497904777526855, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.728463351726532, + "num_tokens": 287025327.0, + "step": 11296 + }, + { + "epoch": 1.240610586426532, + "grad_norm": 2.3380942344665527, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.747405469417572, + "num_tokens": 287050273.0, + "step": 11297 + }, + { + "epoch": 1.2407204041291455, + "grad_norm": 2.018193483352661, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7054382562637329, + "num_tokens": 287081700.0, + "step": 11298 + }, + { + "epoch": 1.2408302218317593, + "grad_norm": 2.2630631923675537, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.713922917842865, + "num_tokens": 287106200.0, + "step": 11299 + }, + { + "epoch": 1.2409400395343728, + "grad_norm": 2.2410218715667725, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7284142971038818, + "num_tokens": 287132512.0, + "step": 11300 + }, + { + "epoch": 1.2410498572369866, + "grad_norm": 2.6407277584075928, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7198748588562012, + "num_tokens": 287152867.0, + "step": 11301 + }, + { + "epoch": 1.2411596749396003, + "grad_norm": 2.310889720916748, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7286633849143982, + "num_tokens": 287175341.0, + "step": 11302 + }, + { + "epoch": 1.2412694926422139, + "grad_norm": 2.388672113418579, + "learning_rate": 1e-06, + "loss": 0.9993, + "mean_token_accuracy": 0.7053651809692383, + "num_tokens": 287202564.0, + "step": 11303 + }, + { + "epoch": 1.2413793103448276, + "grad_norm": 2.1988365650177, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6932895183563232, + "num_tokens": 287230804.0, + "step": 11304 + }, + { + "epoch": 1.2414891280474412, + "grad_norm": 2.371732711791992, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7181621789932251, + "num_tokens": 287253894.0, + "step": 11305 + }, + { + "epoch": 1.241598945750055, + "grad_norm": 2.4423716068267822, + "learning_rate": 1e-06, + "loss": 0.8226, + "mean_token_accuracy": 0.7373716831207275, + "num_tokens": 287275678.0, + "step": 11306 + }, + { + "epoch": 1.2417087634526687, + "grad_norm": 2.435786008834839, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7144594788551331, + "num_tokens": 287297143.0, + "step": 11307 + }, + { + "epoch": 1.2418185811552822, + "grad_norm": 2.325127601623535, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7331784963607788, + "num_tokens": 287320818.0, + "step": 11308 + }, + { + "epoch": 1.241928398857896, + "grad_norm": 2.1907408237457275, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7220048904418945, + "num_tokens": 287347397.0, + "step": 11309 + }, + { + "epoch": 1.2420382165605095, + "grad_norm": 2.1648378372192383, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7217660546302795, + "num_tokens": 287373290.0, + "step": 11310 + }, + { + "epoch": 1.2421480342631233, + "grad_norm": 2.5075905323028564, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6987748146057129, + "num_tokens": 287395679.0, + "step": 11311 + }, + { + "epoch": 1.2422578519657368, + "grad_norm": 2.3385848999023438, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7204992175102234, + "num_tokens": 287420598.0, + "step": 11312 + }, + { + "epoch": 1.2423676696683505, + "grad_norm": 2.214841842651367, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7317856550216675, + "num_tokens": 287448658.0, + "step": 11313 + }, + { + "epoch": 1.242477487370964, + "grad_norm": 1.9615185260772705, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7049059867858887, + "num_tokens": 287477029.0, + "step": 11314 + }, + { + "epoch": 1.2425873050735778, + "grad_norm": 2.363295555114746, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7290776968002319, + "num_tokens": 287499789.0, + "step": 11315 + }, + { + "epoch": 1.2426971227761916, + "grad_norm": 2.170315980911255, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7203847765922546, + "num_tokens": 287527205.0, + "step": 11316 + }, + { + "epoch": 1.2428069404788051, + "grad_norm": 2.259248733520508, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7421878576278687, + "num_tokens": 287550061.0, + "step": 11317 + }, + { + "epoch": 1.242916758181419, + "grad_norm": 2.3211147785186768, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.716515839099884, + "num_tokens": 287572902.0, + "step": 11318 + }, + { + "epoch": 1.2430265758840324, + "grad_norm": 2.2215077877044678, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7419248223304749, + "num_tokens": 287596529.0, + "step": 11319 + }, + { + "epoch": 1.2431363935866462, + "grad_norm": 2.004926919937134, + "learning_rate": 1e-06, + "loss": 0.9752, + "mean_token_accuracy": 0.6970836520195007, + "num_tokens": 287628000.0, + "step": 11320 + }, + { + "epoch": 1.24324621128926, + "grad_norm": 2.1795761585235596, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7231752872467041, + "num_tokens": 287653194.0, + "step": 11321 + }, + { + "epoch": 1.2433560289918735, + "grad_norm": 2.662850856781006, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.726406455039978, + "num_tokens": 287673138.0, + "step": 11322 + }, + { + "epoch": 1.2434658466944872, + "grad_norm": 2.06111741065979, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7242142558097839, + "num_tokens": 287703572.0, + "step": 11323 + }, + { + "epoch": 1.2435756643971008, + "grad_norm": 2.071648359298706, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7000726461410522, + "num_tokens": 287733431.0, + "step": 11324 + }, + { + "epoch": 1.2436854820997145, + "grad_norm": 2.3966612815856934, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7240833044052124, + "num_tokens": 287757006.0, + "step": 11325 + }, + { + "epoch": 1.243795299802328, + "grad_norm": 2.159512758255005, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7323182821273804, + "num_tokens": 287786176.0, + "step": 11326 + }, + { + "epoch": 1.2439051175049418, + "grad_norm": 2.2414984703063965, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7171421051025391, + "num_tokens": 287813311.0, + "step": 11327 + }, + { + "epoch": 1.2440149352075554, + "grad_norm": 2.2864606380462646, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7254364490509033, + "num_tokens": 287837157.0, + "step": 11328 + }, + { + "epoch": 1.244124752910169, + "grad_norm": 2.3095743656158447, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7154102921485901, + "num_tokens": 287861881.0, + "step": 11329 + }, + { + "epoch": 1.2442345706127829, + "grad_norm": 2.2474915981292725, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7282354235649109, + "num_tokens": 287888453.0, + "step": 11330 + }, + { + "epoch": 1.2443443883153964, + "grad_norm": 2.5932533740997314, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7428838014602661, + "num_tokens": 287909460.0, + "step": 11331 + }, + { + "epoch": 1.2444542060180102, + "grad_norm": 2.1422810554504395, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7161277532577515, + "num_tokens": 287937222.0, + "step": 11332 + }, + { + "epoch": 1.2445640237206237, + "grad_norm": 2.2139735221862793, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7247705459594727, + "num_tokens": 287963864.0, + "step": 11333 + }, + { + "epoch": 1.2446738414232374, + "grad_norm": 2.31211519241333, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7225875854492188, + "num_tokens": 287986238.0, + "step": 11334 + }, + { + "epoch": 1.2447836591258512, + "grad_norm": 2.2661890983581543, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7214933633804321, + "num_tokens": 288013282.0, + "step": 11335 + }, + { + "epoch": 1.2448934768284647, + "grad_norm": 2.2802014350891113, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.726888120174408, + "num_tokens": 288039520.0, + "step": 11336 + }, + { + "epoch": 1.2450032945310785, + "grad_norm": 2.187507390975952, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7285796403884888, + "num_tokens": 288066636.0, + "step": 11337 + }, + { + "epoch": 1.245113112233692, + "grad_norm": 2.281655788421631, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7026962041854858, + "num_tokens": 288092223.0, + "step": 11338 + }, + { + "epoch": 1.2452229299363058, + "grad_norm": 2.331726312637329, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7261035442352295, + "num_tokens": 288113465.0, + "step": 11339 + }, + { + "epoch": 1.2453327476389193, + "grad_norm": 2.375821113586426, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7274421453475952, + "num_tokens": 288137652.0, + "step": 11340 + }, + { + "epoch": 1.245442565341533, + "grad_norm": 2.211827516555786, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.705211341381073, + "num_tokens": 288162780.0, + "step": 11341 + }, + { + "epoch": 1.2455523830441466, + "grad_norm": 2.0094432830810547, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7002073526382446, + "num_tokens": 288192286.0, + "step": 11342 + }, + { + "epoch": 1.2456622007467604, + "grad_norm": 2.640861749649048, + "learning_rate": 1e-06, + "loss": 0.7953, + "mean_token_accuracy": 0.7480595707893372, + "num_tokens": 288212003.0, + "step": 11343 + }, + { + "epoch": 1.2457720184493741, + "grad_norm": 2.5839319229125977, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.6990779638290405, + "num_tokens": 288235193.0, + "step": 11344 + }, + { + "epoch": 1.2458818361519877, + "grad_norm": 2.248354434967041, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7135448455810547, + "num_tokens": 288260803.0, + "step": 11345 + }, + { + "epoch": 1.2459916538546014, + "grad_norm": 2.3753013610839844, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7137815952301025, + "num_tokens": 288283700.0, + "step": 11346 + }, + { + "epoch": 1.246101471557215, + "grad_norm": 2.396191120147705, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7385337948799133, + "num_tokens": 288307271.0, + "step": 11347 + }, + { + "epoch": 1.2462112892598287, + "grad_norm": 2.1499550342559814, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7238190770149231, + "num_tokens": 288334622.0, + "step": 11348 + }, + { + "epoch": 1.2463211069624422, + "grad_norm": 2.3255484104156494, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6951586008071899, + "num_tokens": 288360739.0, + "step": 11349 + }, + { + "epoch": 1.246430924665056, + "grad_norm": 2.0617032051086426, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7011659145355225, + "num_tokens": 288390754.0, + "step": 11350 + }, + { + "epoch": 1.2465407423676698, + "grad_norm": 2.1473846435546875, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.718595027923584, + "num_tokens": 288417596.0, + "step": 11351 + }, + { + "epoch": 1.2466505600702833, + "grad_norm": 2.2089972496032715, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7081723213195801, + "num_tokens": 288444055.0, + "step": 11352 + }, + { + "epoch": 1.246760377772897, + "grad_norm": 2.2450430393218994, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7160229086875916, + "num_tokens": 288469644.0, + "step": 11353 + }, + { + "epoch": 1.2468701954755106, + "grad_norm": 2.3322789669036865, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7164351940155029, + "num_tokens": 288493553.0, + "step": 11354 + }, + { + "epoch": 1.2469800131781243, + "grad_norm": 2.5266549587249756, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.73298180103302, + "num_tokens": 288513364.0, + "step": 11355 + }, + { + "epoch": 1.2470898308807379, + "grad_norm": 2.148578405380249, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7070885300636292, + "num_tokens": 288542365.0, + "step": 11356 + }, + { + "epoch": 1.2471996485833516, + "grad_norm": 1.9883415699005127, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6992098093032837, + "num_tokens": 288575697.0, + "step": 11357 + }, + { + "epoch": 1.2473094662859654, + "grad_norm": 2.210249185562134, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7222671508789062, + "num_tokens": 288602019.0, + "step": 11358 + }, + { + "epoch": 1.247419283988579, + "grad_norm": 2.053696393966675, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.720909595489502, + "num_tokens": 288629679.0, + "step": 11359 + }, + { + "epoch": 1.2475291016911927, + "grad_norm": 2.126204490661621, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7040148973464966, + "num_tokens": 288659529.0, + "step": 11360 + }, + { + "epoch": 1.2476389193938062, + "grad_norm": 2.435711145401001, + "learning_rate": 1e-06, + "loss": 0.8244, + "mean_token_accuracy": 0.7348672747612, + "num_tokens": 288681439.0, + "step": 11361 + }, + { + "epoch": 1.24774873709642, + "grad_norm": 2.072387218475342, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7107350826263428, + "num_tokens": 288709654.0, + "step": 11362 + }, + { + "epoch": 1.2478585547990335, + "grad_norm": 2.4516661167144775, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7341582775115967, + "num_tokens": 288731019.0, + "step": 11363 + }, + { + "epoch": 1.2479683725016473, + "grad_norm": 2.3565292358398438, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7283475399017334, + "num_tokens": 288754970.0, + "step": 11364 + }, + { + "epoch": 1.2480781902042608, + "grad_norm": 2.699644088745117, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7345068454742432, + "num_tokens": 288773034.0, + "step": 11365 + }, + { + "epoch": 1.2481880079068746, + "grad_norm": 2.2091243267059326, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7123698592185974, + "num_tokens": 288798041.0, + "step": 11366 + }, + { + "epoch": 1.2482978256094883, + "grad_norm": 2.1213064193725586, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7097465991973877, + "num_tokens": 288828219.0, + "step": 11367 + }, + { + "epoch": 1.2484076433121019, + "grad_norm": 2.130180597305298, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7332226634025574, + "num_tokens": 288855316.0, + "step": 11368 + }, + { + "epoch": 1.2485174610147156, + "grad_norm": 2.487278699874878, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.729316771030426, + "num_tokens": 288876431.0, + "step": 11369 + }, + { + "epoch": 1.2486272787173291, + "grad_norm": 2.3767507076263428, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.70713210105896, + "num_tokens": 288900583.0, + "step": 11370 + }, + { + "epoch": 1.248737096419943, + "grad_norm": 2.1140594482421875, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7098103761672974, + "num_tokens": 288930555.0, + "step": 11371 + }, + { + "epoch": 1.2488469141225567, + "grad_norm": 2.175846815109253, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7270734906196594, + "num_tokens": 288958832.0, + "step": 11372 + }, + { + "epoch": 1.2489567318251702, + "grad_norm": 2.288360834121704, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6981478333473206, + "num_tokens": 288983761.0, + "step": 11373 + }, + { + "epoch": 1.249066549527784, + "grad_norm": 2.3129541873931885, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7486230731010437, + "num_tokens": 289006940.0, + "step": 11374 + }, + { + "epoch": 1.2491763672303975, + "grad_norm": 2.4831912517547607, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.719929575920105, + "num_tokens": 289028180.0, + "step": 11375 + }, + { + "epoch": 1.2492861849330112, + "grad_norm": 2.2119359970092773, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.690315842628479, + "num_tokens": 289055647.0, + "step": 11376 + }, + { + "epoch": 1.2493960026356248, + "grad_norm": 2.4483349323272705, + "learning_rate": 1e-06, + "loss": 0.7991, + "mean_token_accuracy": 0.7481680512428284, + "num_tokens": 289076703.0, + "step": 11377 + }, + { + "epoch": 1.2495058203382385, + "grad_norm": 2.51481556892395, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7191709876060486, + "num_tokens": 289097469.0, + "step": 11378 + }, + { + "epoch": 1.249615638040852, + "grad_norm": 2.0836524963378906, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7129766941070557, + "num_tokens": 289124705.0, + "step": 11379 + }, + { + "epoch": 1.2497254557434658, + "grad_norm": 2.2001750469207764, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7309454679489136, + "num_tokens": 289149788.0, + "step": 11380 + }, + { + "epoch": 1.2498352734460796, + "grad_norm": 2.2573695182800293, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7087526321411133, + "num_tokens": 289175101.0, + "step": 11381 + }, + { + "epoch": 1.2499450911486931, + "grad_norm": 2.089962959289551, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7107247114181519, + "num_tokens": 289201736.0, + "step": 11382 + }, + { + "epoch": 1.2500549088513069, + "grad_norm": 2.1497251987457275, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7041682004928589, + "num_tokens": 289229884.0, + "step": 11383 + }, + { + "epoch": 1.2501647265539204, + "grad_norm": 2.0967066287994385, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.717221736907959, + "num_tokens": 289254991.0, + "step": 11384 + }, + { + "epoch": 1.2502745442565342, + "grad_norm": 2.3679516315460205, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7177449464797974, + "num_tokens": 289278191.0, + "step": 11385 + }, + { + "epoch": 1.250384361959148, + "grad_norm": 2.651672601699829, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7189087867736816, + "num_tokens": 289298954.0, + "step": 11386 + }, + { + "epoch": 1.2504941796617615, + "grad_norm": 2.385016918182373, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7170838713645935, + "num_tokens": 289320811.0, + "step": 11387 + }, + { + "epoch": 1.2506039973643752, + "grad_norm": 2.419926881790161, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7196941375732422, + "num_tokens": 289342787.0, + "step": 11388 + }, + { + "epoch": 1.2507138150669888, + "grad_norm": 2.3222815990448, + "learning_rate": 1e-06, + "loss": 0.8016, + "mean_token_accuracy": 0.7509133219718933, + "num_tokens": 289367142.0, + "step": 11389 + }, + { + "epoch": 1.2508236327696025, + "grad_norm": 1.9910435676574707, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.6848580837249756, + "num_tokens": 289399426.0, + "step": 11390 + }, + { + "epoch": 1.250933450472216, + "grad_norm": 2.146749973297119, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7077422142028809, + "num_tokens": 289429941.0, + "step": 11391 + }, + { + "epoch": 1.2510432681748298, + "grad_norm": 2.3015127182006836, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7288740873336792, + "num_tokens": 289453222.0, + "step": 11392 + }, + { + "epoch": 1.2511530858774433, + "grad_norm": 2.580836772918701, + "learning_rate": 1e-06, + "loss": 0.8075, + "mean_token_accuracy": 0.7430701851844788, + "num_tokens": 289471594.0, + "step": 11393 + }, + { + "epoch": 1.251262903580057, + "grad_norm": 2.2955048084259033, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7358936071395874, + "num_tokens": 289496823.0, + "step": 11394 + }, + { + "epoch": 1.2513727212826709, + "grad_norm": 2.506803512573242, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.733507513999939, + "num_tokens": 289517717.0, + "step": 11395 + }, + { + "epoch": 1.2514825389852844, + "grad_norm": 2.5182859897613525, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7138049602508545, + "num_tokens": 289541548.0, + "step": 11396 + }, + { + "epoch": 1.2515923566878981, + "grad_norm": 2.475926637649536, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7434450387954712, + "num_tokens": 289563563.0, + "step": 11397 + }, + { + "epoch": 1.2517021743905117, + "grad_norm": 2.1059951782226562, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7125215530395508, + "num_tokens": 289592622.0, + "step": 11398 + }, + { + "epoch": 1.2518119920931254, + "grad_norm": 2.253037214279175, + "learning_rate": 1e-06, + "loss": 0.8961, + "mean_token_accuracy": 0.7289379835128784, + "num_tokens": 289619346.0, + "step": 11399 + }, + { + "epoch": 1.2519218097957392, + "grad_norm": 2.4276320934295654, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7312377691268921, + "num_tokens": 289642713.0, + "step": 11400 + }, + { + "epoch": 1.2520316274983527, + "grad_norm": 2.2257769107818604, + "learning_rate": 1e-06, + "loss": 0.9677, + "mean_token_accuracy": 0.7076669335365295, + "num_tokens": 289669069.0, + "step": 11401 + }, + { + "epoch": 1.2521414452009663, + "grad_norm": 2.266773223876953, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.702150285243988, + "num_tokens": 289694592.0, + "step": 11402 + }, + { + "epoch": 1.25225126290358, + "grad_norm": 2.3692526817321777, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.706351101398468, + "num_tokens": 289719399.0, + "step": 11403 + }, + { + "epoch": 1.2523610806061938, + "grad_norm": 2.4898223876953125, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7298005223274231, + "num_tokens": 289740308.0, + "step": 11404 + }, + { + "epoch": 1.2524708983088073, + "grad_norm": 2.1775624752044678, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7206321954727173, + "num_tokens": 289766857.0, + "step": 11405 + }, + { + "epoch": 1.252580716011421, + "grad_norm": 2.307699203491211, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.733681857585907, + "num_tokens": 289791178.0, + "step": 11406 + }, + { + "epoch": 1.2526905337140346, + "grad_norm": 2.376934289932251, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7322416305541992, + "num_tokens": 289815383.0, + "step": 11407 + }, + { + "epoch": 1.2528003514166484, + "grad_norm": 2.1583478450775146, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7382701635360718, + "num_tokens": 289843253.0, + "step": 11408 + }, + { + "epoch": 1.2529101691192621, + "grad_norm": 2.4581708908081055, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7269204258918762, + "num_tokens": 289864048.0, + "step": 11409 + }, + { + "epoch": 1.2530199868218757, + "grad_norm": 2.352736234664917, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7246882915496826, + "num_tokens": 289888197.0, + "step": 11410 + }, + { + "epoch": 1.2531298045244894, + "grad_norm": 2.15429425239563, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7220774292945862, + "num_tokens": 289914297.0, + "step": 11411 + }, + { + "epoch": 1.253239622227103, + "grad_norm": 2.5202202796936035, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.6995989084243774, + "num_tokens": 289936125.0, + "step": 11412 + }, + { + "epoch": 1.2533494399297167, + "grad_norm": 2.259434223175049, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7319756746292114, + "num_tokens": 289961434.0, + "step": 11413 + }, + { + "epoch": 1.2534592576323305, + "grad_norm": 2.2408344745635986, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7198909521102905, + "num_tokens": 289987907.0, + "step": 11414 + }, + { + "epoch": 1.253569075334944, + "grad_norm": 2.109513521194458, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.7198628783226013, + "num_tokens": 290016403.0, + "step": 11415 + }, + { + "epoch": 1.2536788930375575, + "grad_norm": 2.0933969020843506, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7077118158340454, + "num_tokens": 290047011.0, + "step": 11416 + }, + { + "epoch": 1.2537887107401713, + "grad_norm": 2.071380138397217, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7159682512283325, + "num_tokens": 290075607.0, + "step": 11417 + }, + { + "epoch": 1.253898528442785, + "grad_norm": 2.187833786010742, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.727952778339386, + "num_tokens": 290101636.0, + "step": 11418 + }, + { + "epoch": 1.2540083461453986, + "grad_norm": 2.3143794536590576, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7220401763916016, + "num_tokens": 290126740.0, + "step": 11419 + }, + { + "epoch": 1.2541181638480123, + "grad_norm": 2.0959599018096924, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7176022529602051, + "num_tokens": 290154320.0, + "step": 11420 + }, + { + "epoch": 1.2542279815506259, + "grad_norm": 2.1215922832489014, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7201647162437439, + "num_tokens": 290183307.0, + "step": 11421 + }, + { + "epoch": 1.2543377992532396, + "grad_norm": 2.0661818981170654, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7131978273391724, + "num_tokens": 290213225.0, + "step": 11422 + }, + { + "epoch": 1.2544476169558534, + "grad_norm": 2.1662344932556152, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7222214937210083, + "num_tokens": 290239877.0, + "step": 11423 + }, + { + "epoch": 1.254557434658467, + "grad_norm": 2.1173791885375977, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.723979115486145, + "num_tokens": 290267305.0, + "step": 11424 + }, + { + "epoch": 1.2546672523610807, + "grad_norm": 2.3982203006744385, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7317907810211182, + "num_tokens": 290289315.0, + "step": 11425 + }, + { + "epoch": 1.2547770700636942, + "grad_norm": 2.529801845550537, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7004681825637817, + "num_tokens": 290311378.0, + "step": 11426 + }, + { + "epoch": 1.254886887766308, + "grad_norm": 2.176279067993164, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7117049098014832, + "num_tokens": 290338600.0, + "step": 11427 + }, + { + "epoch": 1.2549967054689217, + "grad_norm": 2.569288969039917, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.727520227432251, + "num_tokens": 290357892.0, + "step": 11428 + }, + { + "epoch": 1.2551065231715353, + "grad_norm": 2.2858660221099854, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7241411209106445, + "num_tokens": 290383806.0, + "step": 11429 + }, + { + "epoch": 1.2552163408741488, + "grad_norm": 2.056817054748535, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.6958963871002197, + "num_tokens": 290415230.0, + "step": 11430 + }, + { + "epoch": 1.2553261585767626, + "grad_norm": 2.636643171310425, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7268638014793396, + "num_tokens": 290433953.0, + "step": 11431 + }, + { + "epoch": 1.2554359762793763, + "grad_norm": 2.4589807987213135, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7341447472572327, + "num_tokens": 290456753.0, + "step": 11432 + }, + { + "epoch": 1.2555457939819898, + "grad_norm": 2.172269821166992, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.724346935749054, + "num_tokens": 290483856.0, + "step": 11433 + }, + { + "epoch": 1.2556556116846036, + "grad_norm": 2.170344352722168, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7177103757858276, + "num_tokens": 290509670.0, + "step": 11434 + }, + { + "epoch": 1.2557654293872171, + "grad_norm": 2.2290236949920654, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7270047664642334, + "num_tokens": 290534322.0, + "step": 11435 + }, + { + "epoch": 1.255875247089831, + "grad_norm": 1.987465500831604, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6925094127655029, + "num_tokens": 290568959.0, + "step": 11436 + }, + { + "epoch": 1.2559850647924446, + "grad_norm": 2.2233433723449707, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7288941740989685, + "num_tokens": 290594090.0, + "step": 11437 + }, + { + "epoch": 1.2560948824950582, + "grad_norm": 2.1877129077911377, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7092934846878052, + "num_tokens": 290623128.0, + "step": 11438 + }, + { + "epoch": 1.256204700197672, + "grad_norm": 2.1821584701538086, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.704853892326355, + "num_tokens": 290652890.0, + "step": 11439 + }, + { + "epoch": 1.2563145179002855, + "grad_norm": 2.574073553085327, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7077616453170776, + "num_tokens": 290673794.0, + "step": 11440 + }, + { + "epoch": 1.2564243356028992, + "grad_norm": 2.216813087463379, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7176100015640259, + "num_tokens": 290698146.0, + "step": 11441 + }, + { + "epoch": 1.2565341533055128, + "grad_norm": 2.5914862155914307, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7256540060043335, + "num_tokens": 290718617.0, + "step": 11442 + }, + { + "epoch": 1.2566439710081265, + "grad_norm": 2.1770424842834473, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7151216268539429, + "num_tokens": 290747715.0, + "step": 11443 + }, + { + "epoch": 1.25675378871074, + "grad_norm": 2.650756359100342, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7477416396141052, + "num_tokens": 290765618.0, + "step": 11444 + }, + { + "epoch": 1.2568636064133538, + "grad_norm": 2.225149393081665, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7331240773200989, + "num_tokens": 290789831.0, + "step": 11445 + }, + { + "epoch": 1.2569734241159676, + "grad_norm": 2.0432822704315186, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7346938848495483, + "num_tokens": 290818656.0, + "step": 11446 + }, + { + "epoch": 1.257083241818581, + "grad_norm": 2.11806058883667, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7225253582000732, + "num_tokens": 290848015.0, + "step": 11447 + }, + { + "epoch": 1.2571930595211949, + "grad_norm": 2.1478419303894043, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7013936638832092, + "num_tokens": 290876100.0, + "step": 11448 + }, + { + "epoch": 1.2573028772238084, + "grad_norm": 2.0845792293548584, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7240727543830872, + "num_tokens": 290903909.0, + "step": 11449 + }, + { + "epoch": 1.2574126949264222, + "grad_norm": 2.462456464767456, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7221797704696655, + "num_tokens": 290925187.0, + "step": 11450 + }, + { + "epoch": 1.257522512629036, + "grad_norm": 1.9615614414215088, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7127131819725037, + "num_tokens": 290956770.0, + "step": 11451 + }, + { + "epoch": 1.2576323303316495, + "grad_norm": 2.4597678184509277, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7321562767028809, + "num_tokens": 290978566.0, + "step": 11452 + }, + { + "epoch": 1.257742148034263, + "grad_norm": 2.3449809551239014, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7099851369857788, + "num_tokens": 291003393.0, + "step": 11453 + }, + { + "epoch": 1.2578519657368767, + "grad_norm": 2.810835599899292, + "learning_rate": 1e-06, + "loss": 0.8204, + "mean_token_accuracy": 0.7408133149147034, + "num_tokens": 291021127.0, + "step": 11454 + }, + { + "epoch": 1.2579617834394905, + "grad_norm": 2.105426788330078, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.718146562576294, + "num_tokens": 291048573.0, + "step": 11455 + }, + { + "epoch": 1.258071601142104, + "grad_norm": 2.2499709129333496, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7275634407997131, + "num_tokens": 291075243.0, + "step": 11456 + }, + { + "epoch": 1.2581814188447178, + "grad_norm": 2.337751626968384, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7218073606491089, + "num_tokens": 291100492.0, + "step": 11457 + }, + { + "epoch": 1.2582912365473313, + "grad_norm": 2.1090216636657715, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7296020984649658, + "num_tokens": 291129263.0, + "step": 11458 + }, + { + "epoch": 1.258401054249945, + "grad_norm": 1.932628870010376, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7348960638046265, + "num_tokens": 291161249.0, + "step": 11459 + }, + { + "epoch": 1.2585108719525588, + "grad_norm": 1.9985302686691284, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.707213282585144, + "num_tokens": 291193957.0, + "step": 11460 + }, + { + "epoch": 1.2586206896551724, + "grad_norm": 2.071147918701172, + "learning_rate": 1e-06, + "loss": 0.8459, + "mean_token_accuracy": 0.7346854209899902, + "num_tokens": 291222570.0, + "step": 11461 + }, + { + "epoch": 1.2587305073577861, + "grad_norm": 2.3349905014038086, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7066724896430969, + "num_tokens": 291245655.0, + "step": 11462 + }, + { + "epoch": 1.2588403250603997, + "grad_norm": 2.30967116355896, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7198632955551147, + "num_tokens": 291271356.0, + "step": 11463 + }, + { + "epoch": 1.2589501427630134, + "grad_norm": 1.9842286109924316, + "learning_rate": 1e-06, + "loss": 0.9928, + "mean_token_accuracy": 0.6928158402442932, + "num_tokens": 291304827.0, + "step": 11464 + }, + { + "epoch": 1.2590599604656272, + "grad_norm": 2.6461572647094727, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.738618016242981, + "num_tokens": 291323574.0, + "step": 11465 + }, + { + "epoch": 1.2591697781682407, + "grad_norm": 2.3026821613311768, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7354110479354858, + "num_tokens": 291347734.0, + "step": 11466 + }, + { + "epoch": 1.2592795958708543, + "grad_norm": 2.010136604309082, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.697898268699646, + "num_tokens": 291377721.0, + "step": 11467 + }, + { + "epoch": 1.259389413573468, + "grad_norm": 2.2801194190979004, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7408978343009949, + "num_tokens": 291399156.0, + "step": 11468 + }, + { + "epoch": 1.2594992312760818, + "grad_norm": 2.1754558086395264, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7171533107757568, + "num_tokens": 291424248.0, + "step": 11469 + }, + { + "epoch": 1.2596090489786953, + "grad_norm": 2.3012914657592773, + "learning_rate": 1e-06, + "loss": 1.0041, + "mean_token_accuracy": 0.6936522722244263, + "num_tokens": 291448777.0, + "step": 11470 + }, + { + "epoch": 1.259718866681309, + "grad_norm": 2.1978626251220703, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7215442061424255, + "num_tokens": 291476529.0, + "step": 11471 + }, + { + "epoch": 1.2598286843839226, + "grad_norm": 2.115278720855713, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7319157719612122, + "num_tokens": 291504385.0, + "step": 11472 + }, + { + "epoch": 1.2599385020865363, + "grad_norm": 2.102104425430298, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7097001075744629, + "num_tokens": 291531047.0, + "step": 11473 + }, + { + "epoch": 1.26004831978915, + "grad_norm": 2.062615394592285, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7067402601242065, + "num_tokens": 291559180.0, + "step": 11474 + }, + { + "epoch": 1.2601581374917636, + "grad_norm": 2.2033445835113525, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.728622317314148, + "num_tokens": 291585387.0, + "step": 11475 + }, + { + "epoch": 1.2602679551943774, + "grad_norm": 2.2157351970672607, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7008242011070251, + "num_tokens": 291613771.0, + "step": 11476 + }, + { + "epoch": 1.260377772896991, + "grad_norm": 2.4525270462036133, + "learning_rate": 1e-06, + "loss": 0.829, + "mean_token_accuracy": 0.742238461971283, + "num_tokens": 291635910.0, + "step": 11477 + }, + { + "epoch": 1.2604875905996047, + "grad_norm": 2.0869762897491455, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7257000803947449, + "num_tokens": 291664611.0, + "step": 11478 + }, + { + "epoch": 1.2605974083022184, + "grad_norm": 2.5657782554626465, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7080683708190918, + "num_tokens": 291686772.0, + "step": 11479 + }, + { + "epoch": 1.260707226004832, + "grad_norm": 2.4186456203460693, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7314763069152832, + "num_tokens": 291708885.0, + "step": 11480 + }, + { + "epoch": 1.2608170437074455, + "grad_norm": 2.6456761360168457, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7152305841445923, + "num_tokens": 291729968.0, + "step": 11481 + }, + { + "epoch": 1.2609268614100593, + "grad_norm": 2.1816864013671875, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7171011567115784, + "num_tokens": 291757510.0, + "step": 11482 + }, + { + "epoch": 1.261036679112673, + "grad_norm": 2.2383666038513184, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.7003628611564636, + "num_tokens": 291784320.0, + "step": 11483 + }, + { + "epoch": 1.2611464968152866, + "grad_norm": 2.1163389682769775, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7140249609947205, + "num_tokens": 291812414.0, + "step": 11484 + }, + { + "epoch": 1.2612563145179003, + "grad_norm": 2.238101005554199, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7065148949623108, + "num_tokens": 291838813.0, + "step": 11485 + }, + { + "epoch": 1.2613661322205139, + "grad_norm": 2.4216885566711426, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7303733825683594, + "num_tokens": 291861803.0, + "step": 11486 + }, + { + "epoch": 1.2614759499231276, + "grad_norm": 2.268589496612549, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7433308959007263, + "num_tokens": 291886274.0, + "step": 11487 + }, + { + "epoch": 1.2615857676257414, + "grad_norm": 2.109060525894165, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7181293964385986, + "num_tokens": 291918250.0, + "step": 11488 + }, + { + "epoch": 1.261695585328355, + "grad_norm": 2.337272882461548, + "learning_rate": 1e-06, + "loss": 0.785, + "mean_token_accuracy": 0.7459682822227478, + "num_tokens": 291938115.0, + "step": 11489 + }, + { + "epoch": 1.2618054030309687, + "grad_norm": 2.3543341159820557, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7171169519424438, + "num_tokens": 291961418.0, + "step": 11490 + }, + { + "epoch": 1.2619152207335822, + "grad_norm": 2.5271339416503906, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7238235473632812, + "num_tokens": 291981195.0, + "step": 11491 + }, + { + "epoch": 1.262025038436196, + "grad_norm": 2.3860228061676025, + "learning_rate": 1e-06, + "loss": 0.7285, + "mean_token_accuracy": 0.7628903388977051, + "num_tokens": 292001955.0, + "step": 11492 + }, + { + "epoch": 1.2621348561388097, + "grad_norm": 2.237809419631958, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7102567553520203, + "num_tokens": 292028678.0, + "step": 11493 + }, + { + "epoch": 1.2622446738414232, + "grad_norm": 2.195543050765991, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7026139497756958, + "num_tokens": 292057384.0, + "step": 11494 + }, + { + "epoch": 1.2623544915440368, + "grad_norm": 2.320894479751587, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7094469666481018, + "num_tokens": 292083514.0, + "step": 11495 + }, + { + "epoch": 1.2624643092466505, + "grad_norm": 2.3390464782714844, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7077065110206604, + "num_tokens": 292108762.0, + "step": 11496 + }, + { + "epoch": 1.2625741269492643, + "grad_norm": 2.2306201457977295, + "learning_rate": 1e-06, + "loss": 1.0027, + "mean_token_accuracy": 0.6894936561584473, + "num_tokens": 292134336.0, + "step": 11497 + }, + { + "epoch": 1.2626839446518778, + "grad_norm": 2.542769432067871, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7268476486206055, + "num_tokens": 292155050.0, + "step": 11498 + }, + { + "epoch": 1.2627937623544916, + "grad_norm": 2.3821299076080322, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7190995812416077, + "num_tokens": 292177786.0, + "step": 11499 + }, + { + "epoch": 1.2629035800571051, + "grad_norm": 2.3630471229553223, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7076067924499512, + "num_tokens": 292200991.0, + "step": 11500 + }, + { + "epoch": 1.2630133977597189, + "grad_norm": 2.369330406188965, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7273558378219604, + "num_tokens": 292225678.0, + "step": 11501 + }, + { + "epoch": 1.2631232154623326, + "grad_norm": 2.325246572494507, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7295103669166565, + "num_tokens": 292251093.0, + "step": 11502 + }, + { + "epoch": 1.2632330331649462, + "grad_norm": 2.5044362545013428, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7199857234954834, + "num_tokens": 292273348.0, + "step": 11503 + }, + { + "epoch": 1.26334285086756, + "grad_norm": 2.005774974822998, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7374575138092041, + "num_tokens": 292302717.0, + "step": 11504 + }, + { + "epoch": 1.2634526685701735, + "grad_norm": 2.2948925495147705, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7166658639907837, + "num_tokens": 292328055.0, + "step": 11505 + }, + { + "epoch": 1.2635624862727872, + "grad_norm": 2.4427435398101807, + "learning_rate": 1e-06, + "loss": 0.75, + "mean_token_accuracy": 0.7561639547348022, + "num_tokens": 292350250.0, + "step": 11506 + }, + { + "epoch": 1.2636723039754008, + "grad_norm": 2.4397225379943848, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7314132452011108, + "num_tokens": 292371596.0, + "step": 11507 + }, + { + "epoch": 1.2637821216780145, + "grad_norm": 2.156649589538574, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7106627821922302, + "num_tokens": 292398902.0, + "step": 11508 + }, + { + "epoch": 1.263891939380628, + "grad_norm": 2.3841331005096436, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7133357524871826, + "num_tokens": 292421315.0, + "step": 11509 + }, + { + "epoch": 1.2640017570832418, + "grad_norm": 2.1941869258880615, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7049169540405273, + "num_tokens": 292449459.0, + "step": 11510 + }, + { + "epoch": 1.2641115747858556, + "grad_norm": 2.7561347484588623, + "learning_rate": 1e-06, + "loss": 0.8271, + "mean_token_accuracy": 0.7400538921356201, + "num_tokens": 292467951.0, + "step": 11511 + }, + { + "epoch": 1.264221392488469, + "grad_norm": 2.7004408836364746, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7377588748931885, + "num_tokens": 292484941.0, + "step": 11512 + }, + { + "epoch": 1.2643312101910829, + "grad_norm": 1.926670789718628, + "learning_rate": 1e-06, + "loss": 1.0222, + "mean_token_accuracy": 0.6910624504089355, + "num_tokens": 292519864.0, + "step": 11513 + }, + { + "epoch": 1.2644410278936964, + "grad_norm": 2.400642156600952, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7226426005363464, + "num_tokens": 292544570.0, + "step": 11514 + }, + { + "epoch": 1.2645508455963101, + "grad_norm": 2.8608155250549316, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.727158784866333, + "num_tokens": 292561849.0, + "step": 11515 + }, + { + "epoch": 1.264660663298924, + "grad_norm": 2.4187676906585693, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7080684900283813, + "num_tokens": 292586139.0, + "step": 11516 + }, + { + "epoch": 1.2647704810015374, + "grad_norm": 2.06915545463562, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7329387664794922, + "num_tokens": 292613134.0, + "step": 11517 + }, + { + "epoch": 1.264880298704151, + "grad_norm": 2.385225296020508, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.700755774974823, + "num_tokens": 292636438.0, + "step": 11518 + }, + { + "epoch": 1.2649901164067647, + "grad_norm": 2.338404893875122, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.721075177192688, + "num_tokens": 292661244.0, + "step": 11519 + }, + { + "epoch": 1.2650999341093785, + "grad_norm": 2.3369343280792236, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.695518970489502, + "num_tokens": 292686967.0, + "step": 11520 + }, + { + "epoch": 1.265209751811992, + "grad_norm": 2.484374523162842, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7334973216056824, + "num_tokens": 292708423.0, + "step": 11521 + }, + { + "epoch": 1.2653195695146058, + "grad_norm": 2.220141887664795, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.708945631980896, + "num_tokens": 292734104.0, + "step": 11522 + }, + { + "epoch": 1.2654293872172193, + "grad_norm": 1.9950278997421265, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7117830514907837, + "num_tokens": 292765917.0, + "step": 11523 + }, + { + "epoch": 1.265539204919833, + "grad_norm": 2.4008727073669434, + "learning_rate": 1e-06, + "loss": 0.8192, + "mean_token_accuracy": 0.7366236448287964, + "num_tokens": 292785864.0, + "step": 11524 + }, + { + "epoch": 1.2656490226224468, + "grad_norm": 2.225142002105713, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7100569009780884, + "num_tokens": 292813007.0, + "step": 11525 + }, + { + "epoch": 1.2657588403250604, + "grad_norm": 2.256915330886841, + "learning_rate": 1e-06, + "loss": 0.8268, + "mean_token_accuracy": 0.7374032139778137, + "num_tokens": 292836142.0, + "step": 11526 + }, + { + "epoch": 1.2658686580276741, + "grad_norm": 2.2049102783203125, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6921727061271667, + "num_tokens": 292862602.0, + "step": 11527 + }, + { + "epoch": 1.2659784757302877, + "grad_norm": 2.489675998687744, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.7379960417747498, + "num_tokens": 292883739.0, + "step": 11528 + }, + { + "epoch": 1.2660882934329014, + "grad_norm": 2.575744152069092, + "learning_rate": 1e-06, + "loss": 0.7655, + "mean_token_accuracy": 0.7533589601516724, + "num_tokens": 292902110.0, + "step": 11529 + }, + { + "epoch": 1.2661981111355152, + "grad_norm": 2.4078776836395264, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.736272931098938, + "num_tokens": 292925022.0, + "step": 11530 + }, + { + "epoch": 1.2663079288381287, + "grad_norm": 2.2982728481292725, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7205379605293274, + "num_tokens": 292950074.0, + "step": 11531 + }, + { + "epoch": 1.2664177465407422, + "grad_norm": 2.003706455230713, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7112796306610107, + "num_tokens": 292980312.0, + "step": 11532 + }, + { + "epoch": 1.266527564243356, + "grad_norm": 2.3111839294433594, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7234058380126953, + "num_tokens": 293005438.0, + "step": 11533 + }, + { + "epoch": 1.2666373819459698, + "grad_norm": 2.2109360694885254, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7334592342376709, + "num_tokens": 293030737.0, + "step": 11534 + }, + { + "epoch": 1.2667471996485833, + "grad_norm": 2.139221429824829, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7112969160079956, + "num_tokens": 293058415.0, + "step": 11535 + }, + { + "epoch": 1.266857017351197, + "grad_norm": 2.106245994567871, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7376255989074707, + "num_tokens": 293089402.0, + "step": 11536 + }, + { + "epoch": 1.2669668350538106, + "grad_norm": 2.100905418395996, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.7014361023902893, + "num_tokens": 293119444.0, + "step": 11537 + }, + { + "epoch": 1.2670766527564243, + "grad_norm": 1.9888726472854614, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7181112766265869, + "num_tokens": 293151392.0, + "step": 11538 + }, + { + "epoch": 1.267186470459038, + "grad_norm": 2.2521450519561768, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.712012529373169, + "num_tokens": 293176113.0, + "step": 11539 + }, + { + "epoch": 1.2672962881616516, + "grad_norm": 2.426426649093628, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7247262597084045, + "num_tokens": 293197621.0, + "step": 11540 + }, + { + "epoch": 1.2674061058642654, + "grad_norm": 2.350292682647705, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.724054217338562, + "num_tokens": 293223241.0, + "step": 11541 + }, + { + "epoch": 1.267515923566879, + "grad_norm": 2.527672052383423, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7354092597961426, + "num_tokens": 293244600.0, + "step": 11542 + }, + { + "epoch": 1.2676257412694927, + "grad_norm": 2.1712357997894287, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7253737449645996, + "num_tokens": 293270965.0, + "step": 11543 + }, + { + "epoch": 1.2677355589721064, + "grad_norm": 2.268047571182251, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7234067916870117, + "num_tokens": 293296428.0, + "step": 11544 + }, + { + "epoch": 1.26784537667472, + "grad_norm": 2.109976291656494, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7208845615386963, + "num_tokens": 293322469.0, + "step": 11545 + }, + { + "epoch": 1.2679551943773335, + "grad_norm": 2.068459987640381, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7115695476531982, + "num_tokens": 293351943.0, + "step": 11546 + }, + { + "epoch": 1.2680650120799473, + "grad_norm": 2.1277520656585693, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7132117748260498, + "num_tokens": 293380177.0, + "step": 11547 + }, + { + "epoch": 1.268174829782561, + "grad_norm": 2.0639517307281494, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7047083377838135, + "num_tokens": 293413868.0, + "step": 11548 + }, + { + "epoch": 1.2682846474851746, + "grad_norm": 2.2371137142181396, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7207164764404297, + "num_tokens": 293438892.0, + "step": 11549 + }, + { + "epoch": 1.2683944651877883, + "grad_norm": 2.323660373687744, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7051548957824707, + "num_tokens": 293465399.0, + "step": 11550 + }, + { + "epoch": 1.2685042828904018, + "grad_norm": 2.1780712604522705, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7318587303161621, + "num_tokens": 293491210.0, + "step": 11551 + }, + { + "epoch": 1.2686141005930156, + "grad_norm": 2.409412384033203, + "learning_rate": 1e-06, + "loss": 0.7347, + "mean_token_accuracy": 0.7573104500770569, + "num_tokens": 293511630.0, + "step": 11552 + }, + { + "epoch": 1.2687239182956294, + "grad_norm": 2.3535242080688477, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7396059036254883, + "num_tokens": 293533653.0, + "step": 11553 + }, + { + "epoch": 1.268833735998243, + "grad_norm": 2.2196507453918457, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7270886898040771, + "num_tokens": 293559765.0, + "step": 11554 + }, + { + "epoch": 1.2689435537008567, + "grad_norm": 2.488792657852173, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7122195363044739, + "num_tokens": 293582111.0, + "step": 11555 + }, + { + "epoch": 1.2690533714034702, + "grad_norm": 2.0929741859436035, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7155490517616272, + "num_tokens": 293609285.0, + "step": 11556 + }, + { + "epoch": 1.269163189106084, + "grad_norm": 2.2515392303466797, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7089198231697083, + "num_tokens": 293635242.0, + "step": 11557 + }, + { + "epoch": 1.2692730068086975, + "grad_norm": 2.4622559547424316, + "learning_rate": 1e-06, + "loss": 0.8827, + "mean_token_accuracy": 0.7268531322479248, + "num_tokens": 293656228.0, + "step": 11558 + }, + { + "epoch": 1.2693828245113112, + "grad_norm": 2.359218120574951, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7165526747703552, + "num_tokens": 293682776.0, + "step": 11559 + }, + { + "epoch": 1.2694926422139248, + "grad_norm": 2.227396011352539, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7389895915985107, + "num_tokens": 293708735.0, + "step": 11560 + }, + { + "epoch": 1.2696024599165385, + "grad_norm": 2.488102674484253, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7026639580726624, + "num_tokens": 293730097.0, + "step": 11561 + }, + { + "epoch": 1.2697122776191523, + "grad_norm": 2.2612247467041016, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7150178551673889, + "num_tokens": 293755223.0, + "step": 11562 + }, + { + "epoch": 1.2698220953217658, + "grad_norm": 2.2539308071136475, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.729506254196167, + "num_tokens": 293782214.0, + "step": 11563 + }, + { + "epoch": 1.2699319130243796, + "grad_norm": 2.216996192932129, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7364568710327148, + "num_tokens": 293807813.0, + "step": 11564 + }, + { + "epoch": 1.2700417307269931, + "grad_norm": 2.3226428031921387, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7222919464111328, + "num_tokens": 293830209.0, + "step": 11565 + }, + { + "epoch": 1.2701515484296069, + "grad_norm": 2.115840435028076, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7130558490753174, + "num_tokens": 293859868.0, + "step": 11566 + }, + { + "epoch": 1.2702613661322206, + "grad_norm": 2.4682626724243164, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7233314514160156, + "num_tokens": 293883045.0, + "step": 11567 + }, + { + "epoch": 1.2703711838348342, + "grad_norm": 2.5257325172424316, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.736623227596283, + "num_tokens": 293902839.0, + "step": 11568 + }, + { + "epoch": 1.270481001537448, + "grad_norm": 2.3687150478363037, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7180162072181702, + "num_tokens": 293925184.0, + "step": 11569 + }, + { + "epoch": 1.2705908192400615, + "grad_norm": 2.0965590476989746, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.707086443901062, + "num_tokens": 293953083.0, + "step": 11570 + }, + { + "epoch": 1.2707006369426752, + "grad_norm": 2.170441150665283, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7322966456413269, + "num_tokens": 293978746.0, + "step": 11571 + }, + { + "epoch": 1.2708104546452887, + "grad_norm": 2.6062350273132324, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.725594162940979, + "num_tokens": 293999130.0, + "step": 11572 + }, + { + "epoch": 1.2709202723479025, + "grad_norm": 2.2683088779449463, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7367244958877563, + "num_tokens": 294024300.0, + "step": 11573 + }, + { + "epoch": 1.271030090050516, + "grad_norm": 2.4294326305389404, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7219743728637695, + "num_tokens": 294046438.0, + "step": 11574 + }, + { + "epoch": 1.2711399077531298, + "grad_norm": 2.40818190574646, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7136542797088623, + "num_tokens": 294068770.0, + "step": 11575 + }, + { + "epoch": 1.2712497254557436, + "grad_norm": 1.9629088640213013, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.6995011568069458, + "num_tokens": 294099053.0, + "step": 11576 + }, + { + "epoch": 1.271359543158357, + "grad_norm": 2.16109037399292, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7254972457885742, + "num_tokens": 294125879.0, + "step": 11577 + }, + { + "epoch": 1.2714693608609708, + "grad_norm": 2.105633497238159, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7419947981834412, + "num_tokens": 294152712.0, + "step": 11578 + }, + { + "epoch": 1.2715791785635844, + "grad_norm": 2.411494731903076, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7197351455688477, + "num_tokens": 294176096.0, + "step": 11579 + }, + { + "epoch": 1.2716889962661981, + "grad_norm": 2.2762112617492676, + "learning_rate": 1e-06, + "loss": 0.8148, + "mean_token_accuracy": 0.7379251718521118, + "num_tokens": 294200908.0, + "step": 11580 + }, + { + "epoch": 1.271798813968812, + "grad_norm": 2.2992022037506104, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7388433218002319, + "num_tokens": 294223531.0, + "step": 11581 + }, + { + "epoch": 1.2719086316714254, + "grad_norm": 2.112196922302246, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7208712697029114, + "num_tokens": 294250520.0, + "step": 11582 + }, + { + "epoch": 1.272018449374039, + "grad_norm": 2.2200443744659424, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7247816324234009, + "num_tokens": 294278255.0, + "step": 11583 + }, + { + "epoch": 1.2721282670766527, + "grad_norm": 2.2990188598632812, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7354441285133362, + "num_tokens": 294303188.0, + "step": 11584 + }, + { + "epoch": 1.2722380847792665, + "grad_norm": 1.9943228960037231, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.717869758605957, + "num_tokens": 294332141.0, + "step": 11585 + }, + { + "epoch": 1.27234790248188, + "grad_norm": 2.04107928276062, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7177743911743164, + "num_tokens": 294363105.0, + "step": 11586 + }, + { + "epoch": 1.2724577201844938, + "grad_norm": 2.1485047340393066, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7149895429611206, + "num_tokens": 294391214.0, + "step": 11587 + }, + { + "epoch": 1.2725675378871073, + "grad_norm": 2.180111885070801, + "learning_rate": 1e-06, + "loss": 1.0339, + "mean_token_accuracy": 0.6876241564750671, + "num_tokens": 294417567.0, + "step": 11588 + }, + { + "epoch": 1.272677355589721, + "grad_norm": 2.310819387435913, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7099355459213257, + "num_tokens": 294442732.0, + "step": 11589 + }, + { + "epoch": 1.2727871732923348, + "grad_norm": 2.121204137802124, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7356681823730469, + "num_tokens": 294469882.0, + "step": 11590 + }, + { + "epoch": 1.2728969909949484, + "grad_norm": 2.245492696762085, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7076349854469299, + "num_tokens": 294498323.0, + "step": 11591 + }, + { + "epoch": 1.273006808697562, + "grad_norm": 2.1345584392547607, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7178625464439392, + "num_tokens": 294526604.0, + "step": 11592 + }, + { + "epoch": 1.2731166264001756, + "grad_norm": 2.381073474884033, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7159254550933838, + "num_tokens": 294550180.0, + "step": 11593 + }, + { + "epoch": 1.2732264441027894, + "grad_norm": 2.574617624282837, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7252020239830017, + "num_tokens": 294571173.0, + "step": 11594 + }, + { + "epoch": 1.2733362618054032, + "grad_norm": 2.293201446533203, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7137622833251953, + "num_tokens": 294594613.0, + "step": 11595 + }, + { + "epoch": 1.2734460795080167, + "grad_norm": 2.420376777648926, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7129442691802979, + "num_tokens": 294617985.0, + "step": 11596 + }, + { + "epoch": 1.2735558972106302, + "grad_norm": 2.4615437984466553, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7133663892745972, + "num_tokens": 294640299.0, + "step": 11597 + }, + { + "epoch": 1.273665714913244, + "grad_norm": 2.0587377548217773, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7033361196517944, + "num_tokens": 294669835.0, + "step": 11598 + }, + { + "epoch": 1.2737755326158577, + "grad_norm": 2.9845972061157227, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7207511067390442, + "num_tokens": 294685203.0, + "step": 11599 + }, + { + "epoch": 1.2738853503184713, + "grad_norm": 2.432982921600342, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7129360437393188, + "num_tokens": 294707659.0, + "step": 11600 + }, + { + "epoch": 1.273995168021085, + "grad_norm": 2.3494131565093994, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7229107022285461, + "num_tokens": 294731231.0, + "step": 11601 + }, + { + "epoch": 1.2741049857236986, + "grad_norm": 2.6300930976867676, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7368451952934265, + "num_tokens": 294749989.0, + "step": 11602 + }, + { + "epoch": 1.2742148034263123, + "grad_norm": 2.297604560852051, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7329835891723633, + "num_tokens": 294775109.0, + "step": 11603 + }, + { + "epoch": 1.274324621128926, + "grad_norm": 2.455984592437744, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7121787667274475, + "num_tokens": 294797874.0, + "step": 11604 + }, + { + "epoch": 1.2744344388315396, + "grad_norm": 2.1556005477905273, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7141751050949097, + "num_tokens": 294826129.0, + "step": 11605 + }, + { + "epoch": 1.2745442565341534, + "grad_norm": 2.757596015930176, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7187439203262329, + "num_tokens": 294844645.0, + "step": 11606 + }, + { + "epoch": 1.274654074236767, + "grad_norm": 2.337318181991577, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7243376970291138, + "num_tokens": 294869924.0, + "step": 11607 + }, + { + "epoch": 1.2747638919393807, + "grad_norm": 2.236044406890869, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7056810259819031, + "num_tokens": 294896455.0, + "step": 11608 + }, + { + "epoch": 1.2748737096419944, + "grad_norm": 2.1919915676116943, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7151861786842346, + "num_tokens": 294922494.0, + "step": 11609 + }, + { + "epoch": 1.274983527344608, + "grad_norm": 2.106302499771118, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.7095131874084473, + "num_tokens": 294951964.0, + "step": 11610 + }, + { + "epoch": 1.2750933450472215, + "grad_norm": 2.2250442504882812, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7293746471405029, + "num_tokens": 294976784.0, + "step": 11611 + }, + { + "epoch": 1.2752031627498353, + "grad_norm": 1.9437226057052612, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7169504165649414, + "num_tokens": 295007872.0, + "step": 11612 + }, + { + "epoch": 1.275312980452449, + "grad_norm": 2.1641077995300293, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.692869246006012, + "num_tokens": 295033405.0, + "step": 11613 + }, + { + "epoch": 1.2754227981550625, + "grad_norm": 2.336512565612793, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7047768831253052, + "num_tokens": 295057419.0, + "step": 11614 + }, + { + "epoch": 1.2755326158576763, + "grad_norm": 2.1774847507476807, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7188000082969666, + "num_tokens": 295084600.0, + "step": 11615 + }, + { + "epoch": 1.2756424335602898, + "grad_norm": 2.3586504459381104, + "learning_rate": 1e-06, + "loss": 0.987, + "mean_token_accuracy": 0.7068363428115845, + "num_tokens": 295108725.0, + "step": 11616 + }, + { + "epoch": 1.2757522512629036, + "grad_norm": 2.3787574768066406, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.727626621723175, + "num_tokens": 295130243.0, + "step": 11617 + }, + { + "epoch": 1.2758620689655173, + "grad_norm": 2.3415110111236572, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7106317281723022, + "num_tokens": 295152906.0, + "step": 11618 + }, + { + "epoch": 1.2759718866681309, + "grad_norm": 2.2256875038146973, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6940476894378662, + "num_tokens": 295179214.0, + "step": 11619 + }, + { + "epoch": 1.2760817043707446, + "grad_norm": 2.758530855178833, + "learning_rate": 1e-06, + "loss": 0.7558, + "mean_token_accuracy": 0.7536953687667847, + "num_tokens": 295196599.0, + "step": 11620 + }, + { + "epoch": 1.2761915220733582, + "grad_norm": 2.1777141094207764, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7547082901000977, + "num_tokens": 295222143.0, + "step": 11621 + }, + { + "epoch": 1.276301339775972, + "grad_norm": 2.3714630603790283, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7345154285430908, + "num_tokens": 295245340.0, + "step": 11622 + }, + { + "epoch": 1.2764111574785855, + "grad_norm": 2.2512028217315674, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.7144310474395752, + "num_tokens": 295272306.0, + "step": 11623 + }, + { + "epoch": 1.2765209751811992, + "grad_norm": 2.28786563873291, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7242562770843506, + "num_tokens": 295297812.0, + "step": 11624 + }, + { + "epoch": 1.2766307928838128, + "grad_norm": 2.576531410217285, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7334425449371338, + "num_tokens": 295318596.0, + "step": 11625 + }, + { + "epoch": 1.2767406105864265, + "grad_norm": 2.194978952407837, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.7107948064804077, + "num_tokens": 295347363.0, + "step": 11626 + }, + { + "epoch": 1.2768504282890403, + "grad_norm": 2.3183372020721436, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.6995078325271606, + "num_tokens": 295370193.0, + "step": 11627 + }, + { + "epoch": 1.2769602459916538, + "grad_norm": 2.136686086654663, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6887795925140381, + "num_tokens": 295396918.0, + "step": 11628 + }, + { + "epoch": 1.2770700636942676, + "grad_norm": 2.269331455230713, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7181954383850098, + "num_tokens": 295422714.0, + "step": 11629 + }, + { + "epoch": 1.277179881396881, + "grad_norm": 2.304868221282959, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7171297073364258, + "num_tokens": 295447907.0, + "step": 11630 + }, + { + "epoch": 1.2772896990994949, + "grad_norm": 2.1972615718841553, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.727401614189148, + "num_tokens": 295473039.0, + "step": 11631 + }, + { + "epoch": 1.2773995168021086, + "grad_norm": 2.676774024963379, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7312713861465454, + "num_tokens": 295491231.0, + "step": 11632 + }, + { + "epoch": 1.2775093345047221, + "grad_norm": 2.2311818599700928, + "learning_rate": 1e-06, + "loss": 1.0363, + "mean_token_accuracy": 0.6856133937835693, + "num_tokens": 295519140.0, + "step": 11633 + }, + { + "epoch": 1.2776191522073357, + "grad_norm": 2.8602027893066406, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7421393394470215, + "num_tokens": 295535570.0, + "step": 11634 + }, + { + "epoch": 1.2777289699099494, + "grad_norm": 2.1449031829833984, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7229423522949219, + "num_tokens": 295560754.0, + "step": 11635 + }, + { + "epoch": 1.2778387876125632, + "grad_norm": 2.060664415359497, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7261773347854614, + "num_tokens": 295589539.0, + "step": 11636 + }, + { + "epoch": 1.2779486053151767, + "grad_norm": 2.3650975227355957, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7196274995803833, + "num_tokens": 295611808.0, + "step": 11637 + }, + { + "epoch": 1.2780584230177905, + "grad_norm": 2.246600866317749, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.6967263221740723, + "num_tokens": 295638683.0, + "step": 11638 + }, + { + "epoch": 1.278168240720404, + "grad_norm": 2.203493118286133, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7199306488037109, + "num_tokens": 295664663.0, + "step": 11639 + }, + { + "epoch": 1.2782780584230178, + "grad_norm": 2.087024688720703, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7145296335220337, + "num_tokens": 295690683.0, + "step": 11640 + }, + { + "epoch": 1.2783878761256315, + "grad_norm": 1.9850276708602905, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7232794165611267, + "num_tokens": 295721300.0, + "step": 11641 + }, + { + "epoch": 1.278497693828245, + "grad_norm": 2.210916519165039, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7050825357437134, + "num_tokens": 295747399.0, + "step": 11642 + }, + { + "epoch": 1.2786075115308588, + "grad_norm": 2.201371669769287, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7182037234306335, + "num_tokens": 295772915.0, + "step": 11643 + }, + { + "epoch": 1.2787173292334724, + "grad_norm": 2.293045997619629, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7071294784545898, + "num_tokens": 295798222.0, + "step": 11644 + }, + { + "epoch": 1.2788271469360861, + "grad_norm": 2.1978139877319336, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7331023216247559, + "num_tokens": 295822506.0, + "step": 11645 + }, + { + "epoch": 1.2789369646386999, + "grad_norm": 2.6944518089294434, + "learning_rate": 1e-06, + "loss": 0.8401, + "mean_token_accuracy": 0.7333444356918335, + "num_tokens": 295841303.0, + "step": 11646 + }, + { + "epoch": 1.2790467823413134, + "grad_norm": 2.1727054119110107, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7265368700027466, + "num_tokens": 295867623.0, + "step": 11647 + }, + { + "epoch": 1.279156600043927, + "grad_norm": 2.170701503753662, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7220237851142883, + "num_tokens": 295894416.0, + "step": 11648 + }, + { + "epoch": 1.2792664177465407, + "grad_norm": 2.7348997592926025, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7252786159515381, + "num_tokens": 295912419.0, + "step": 11649 + }, + { + "epoch": 1.2793762354491545, + "grad_norm": 2.3432178497314453, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7175389528274536, + "num_tokens": 295936999.0, + "step": 11650 + }, + { + "epoch": 1.279486053151768, + "grad_norm": 2.3067493438720703, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7241486310958862, + "num_tokens": 295960445.0, + "step": 11651 + }, + { + "epoch": 1.2795958708543818, + "grad_norm": 2.370969533920288, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7293694615364075, + "num_tokens": 295986990.0, + "step": 11652 + }, + { + "epoch": 1.2797056885569953, + "grad_norm": 2.5246849060058594, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7110145092010498, + "num_tokens": 296009346.0, + "step": 11653 + }, + { + "epoch": 1.279815506259609, + "grad_norm": 2.251288890838623, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7233285903930664, + "num_tokens": 296033187.0, + "step": 11654 + }, + { + "epoch": 1.2799253239622228, + "grad_norm": 2.2124056816101074, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.69635409116745, + "num_tokens": 296060492.0, + "step": 11655 + }, + { + "epoch": 1.2800351416648363, + "grad_norm": 2.0572264194488525, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7019480466842651, + "num_tokens": 296089065.0, + "step": 11656 + }, + { + "epoch": 1.28014495936745, + "grad_norm": 2.4943408966064453, + "learning_rate": 1e-06, + "loss": 0.8138, + "mean_token_accuracy": 0.7423503398895264, + "num_tokens": 296109582.0, + "step": 11657 + }, + { + "epoch": 1.2802547770700636, + "grad_norm": 2.1397013664245605, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.709302544593811, + "num_tokens": 296137922.0, + "step": 11658 + }, + { + "epoch": 1.2803645947726774, + "grad_norm": 2.2208664417266846, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7238945960998535, + "num_tokens": 296164187.0, + "step": 11659 + }, + { + "epoch": 1.2804744124752911, + "grad_norm": 2.47668194770813, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7237720489501953, + "num_tokens": 296184449.0, + "step": 11660 + }, + { + "epoch": 1.2805842301779047, + "grad_norm": 2.4665305614471436, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7221637964248657, + "num_tokens": 296207403.0, + "step": 11661 + }, + { + "epoch": 1.2806940478805182, + "grad_norm": 2.496121644973755, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7397596836090088, + "num_tokens": 296227825.0, + "step": 11662 + }, + { + "epoch": 1.280803865583132, + "grad_norm": 2.1861939430236816, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7018231153488159, + "num_tokens": 296254338.0, + "step": 11663 + }, + { + "epoch": 1.2809136832857457, + "grad_norm": 2.1240122318267822, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.6972987651824951, + "num_tokens": 296282918.0, + "step": 11664 + }, + { + "epoch": 1.2810235009883593, + "grad_norm": 2.3775875568389893, + "learning_rate": 1e-06, + "loss": 1.0063, + "mean_token_accuracy": 0.6909908652305603, + "num_tokens": 296308340.0, + "step": 11665 + }, + { + "epoch": 1.281133318690973, + "grad_norm": 2.0154895782470703, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7248489260673523, + "num_tokens": 296338594.0, + "step": 11666 + }, + { + "epoch": 1.2812431363935866, + "grad_norm": 2.2367947101593018, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7199960350990295, + "num_tokens": 296365896.0, + "step": 11667 + }, + { + "epoch": 1.2813529540962003, + "grad_norm": 2.092053174972534, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.703615128993988, + "num_tokens": 296396285.0, + "step": 11668 + }, + { + "epoch": 1.281462771798814, + "grad_norm": 2.2452080249786377, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7226701974868774, + "num_tokens": 296422272.0, + "step": 11669 + }, + { + "epoch": 1.2815725895014276, + "grad_norm": 2.453674077987671, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7319942712783813, + "num_tokens": 296445186.0, + "step": 11670 + }, + { + "epoch": 1.2816824072040414, + "grad_norm": 2.290125608444214, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7249465584754944, + "num_tokens": 296470768.0, + "step": 11671 + }, + { + "epoch": 1.281792224906655, + "grad_norm": 2.142554759979248, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7212631702423096, + "num_tokens": 296497576.0, + "step": 11672 + }, + { + "epoch": 1.2819020426092687, + "grad_norm": 2.6525518894195557, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7318897247314453, + "num_tokens": 296516476.0, + "step": 11673 + }, + { + "epoch": 1.2820118603118824, + "grad_norm": 2.742445707321167, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7443323731422424, + "num_tokens": 296534680.0, + "step": 11674 + }, + { + "epoch": 1.282121678014496, + "grad_norm": 2.4669642448425293, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7416979670524597, + "num_tokens": 296555985.0, + "step": 11675 + }, + { + "epoch": 1.2822314957171095, + "grad_norm": 2.197082042694092, + "learning_rate": 1e-06, + "loss": 0.7568, + "mean_token_accuracy": 0.7589349746704102, + "num_tokens": 296579112.0, + "step": 11676 + }, + { + "epoch": 1.2823413134197232, + "grad_norm": 2.3087620735168457, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7448654770851135, + "num_tokens": 296602549.0, + "step": 11677 + }, + { + "epoch": 1.282451131122337, + "grad_norm": 2.2772817611694336, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.7028248310089111, + "num_tokens": 296631981.0, + "step": 11678 + }, + { + "epoch": 1.2825609488249505, + "grad_norm": 2.3316943645477295, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7190001010894775, + "num_tokens": 296656742.0, + "step": 11679 + }, + { + "epoch": 1.2826707665275643, + "grad_norm": 2.135758638381958, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7178454399108887, + "num_tokens": 296684677.0, + "step": 11680 + }, + { + "epoch": 1.2827805842301778, + "grad_norm": 2.5496740341186523, + "learning_rate": 1e-06, + "loss": 0.8416, + "mean_token_accuracy": 0.7335817813873291, + "num_tokens": 296706174.0, + "step": 11681 + }, + { + "epoch": 1.2828904019327916, + "grad_norm": 2.158940553665161, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7201615571975708, + "num_tokens": 296734554.0, + "step": 11682 + }, + { + "epoch": 1.2830002196354053, + "grad_norm": 2.0586984157562256, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7041562795639038, + "num_tokens": 296763530.0, + "step": 11683 + }, + { + "epoch": 1.2831100373380189, + "grad_norm": 2.4381580352783203, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7272590398788452, + "num_tokens": 296786463.0, + "step": 11684 + }, + { + "epoch": 1.2832198550406326, + "grad_norm": 2.4689977169036865, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7465079426765442, + "num_tokens": 296807305.0, + "step": 11685 + }, + { + "epoch": 1.2833296727432462, + "grad_norm": 2.3790135383605957, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7142559289932251, + "num_tokens": 296832392.0, + "step": 11686 + }, + { + "epoch": 1.28343949044586, + "grad_norm": 2.3570399284362793, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7109069228172302, + "num_tokens": 296855942.0, + "step": 11687 + }, + { + "epoch": 1.2835493081484735, + "grad_norm": 2.410055637359619, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.716256320476532, + "num_tokens": 296879126.0, + "step": 11688 + }, + { + "epoch": 1.2836591258510872, + "grad_norm": 2.1709413528442383, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7079190611839294, + "num_tokens": 296905011.0, + "step": 11689 + }, + { + "epoch": 1.2837689435537007, + "grad_norm": 2.360201597213745, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7155314683914185, + "num_tokens": 296926979.0, + "step": 11690 + }, + { + "epoch": 1.2838787612563145, + "grad_norm": 2.1256089210510254, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7440375089645386, + "num_tokens": 296954750.0, + "step": 11691 + }, + { + "epoch": 1.2839885789589283, + "grad_norm": 2.6481597423553467, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7308409214019775, + "num_tokens": 296974995.0, + "step": 11692 + }, + { + "epoch": 1.2840983966615418, + "grad_norm": 1.9906643629074097, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7289108037948608, + "num_tokens": 297002454.0, + "step": 11693 + }, + { + "epoch": 1.2842082143641556, + "grad_norm": 2.284059762954712, + "learning_rate": 1e-06, + "loss": 0.942, + "mean_token_accuracy": 0.7119495868682861, + "num_tokens": 297027472.0, + "step": 11694 + }, + { + "epoch": 1.284318032066769, + "grad_norm": 2.6376569271087646, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7347314953804016, + "num_tokens": 297045157.0, + "step": 11695 + }, + { + "epoch": 1.2844278497693828, + "grad_norm": 2.349630117416382, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7283104062080383, + "num_tokens": 297070432.0, + "step": 11696 + }, + { + "epoch": 1.2845376674719966, + "grad_norm": 2.141918420791626, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7131566405296326, + "num_tokens": 297097944.0, + "step": 11697 + }, + { + "epoch": 1.2846474851746101, + "grad_norm": 2.4124412536621094, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7172689437866211, + "num_tokens": 297123009.0, + "step": 11698 + }, + { + "epoch": 1.2847573028772237, + "grad_norm": 2.2554385662078857, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7105209827423096, + "num_tokens": 297147413.0, + "step": 11699 + }, + { + "epoch": 1.2848671205798374, + "grad_norm": 2.183004379272461, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7031813263893127, + "num_tokens": 297178427.0, + "step": 11700 + }, + { + "epoch": 1.2849769382824512, + "grad_norm": 2.394869565963745, + "learning_rate": 1e-06, + "loss": 0.8565, + "mean_token_accuracy": 0.7260967493057251, + "num_tokens": 297202601.0, + "step": 11701 + }, + { + "epoch": 1.2850867559850647, + "grad_norm": 2.183546304702759, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7185704112052917, + "num_tokens": 297230074.0, + "step": 11702 + }, + { + "epoch": 1.2851965736876785, + "grad_norm": 2.054161310195923, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7293120622634888, + "num_tokens": 297260371.0, + "step": 11703 + }, + { + "epoch": 1.285306391390292, + "grad_norm": 2.075888156890869, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7006730437278748, + "num_tokens": 297288803.0, + "step": 11704 + }, + { + "epoch": 1.2854162090929058, + "grad_norm": 2.2794642448425293, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7364528179168701, + "num_tokens": 297311171.0, + "step": 11705 + }, + { + "epoch": 1.2855260267955195, + "grad_norm": 1.8942002058029175, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7309268116950989, + "num_tokens": 297343968.0, + "step": 11706 + }, + { + "epoch": 1.285635844498133, + "grad_norm": 2.4437007904052734, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7350839376449585, + "num_tokens": 297365802.0, + "step": 11707 + }, + { + "epoch": 1.2857456622007468, + "grad_norm": 2.2352592945098877, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7346720695495605, + "num_tokens": 297391841.0, + "step": 11708 + }, + { + "epoch": 1.2858554799033604, + "grad_norm": 2.12608003616333, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7264081239700317, + "num_tokens": 297418825.0, + "step": 11709 + }, + { + "epoch": 1.2859652976059741, + "grad_norm": 1.9686788320541382, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7178680300712585, + "num_tokens": 297452898.0, + "step": 11710 + }, + { + "epoch": 1.2860751153085879, + "grad_norm": 2.3136305809020996, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7196565866470337, + "num_tokens": 297479009.0, + "step": 11711 + }, + { + "epoch": 1.2861849330112014, + "grad_norm": 2.3689939975738525, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7141470909118652, + "num_tokens": 297503625.0, + "step": 11712 + }, + { + "epoch": 1.286294750713815, + "grad_norm": 2.3642630577087402, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7152661085128784, + "num_tokens": 297527406.0, + "step": 11713 + }, + { + "epoch": 1.2864045684164287, + "grad_norm": 2.2442684173583984, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7308803796768188, + "num_tokens": 297552559.0, + "step": 11714 + }, + { + "epoch": 1.2865143861190425, + "grad_norm": 2.138084888458252, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7015136480331421, + "num_tokens": 297581392.0, + "step": 11715 + }, + { + "epoch": 1.286624203821656, + "grad_norm": 2.5227928161621094, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7464981079101562, + "num_tokens": 297602334.0, + "step": 11716 + }, + { + "epoch": 1.2867340215242697, + "grad_norm": 2.437256097793579, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7277126312255859, + "num_tokens": 297623347.0, + "step": 11717 + }, + { + "epoch": 1.2868438392268833, + "grad_norm": 2.2720494270324707, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7014552354812622, + "num_tokens": 297648742.0, + "step": 11718 + }, + { + "epoch": 1.286953656929497, + "grad_norm": 2.0386197566986084, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7080494165420532, + "num_tokens": 297678178.0, + "step": 11719 + }, + { + "epoch": 1.2870634746321108, + "grad_norm": 2.3179874420166016, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7388780117034912, + "num_tokens": 297701356.0, + "step": 11720 + }, + { + "epoch": 1.2871732923347243, + "grad_norm": 2.0694406032562256, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7184836864471436, + "num_tokens": 297730844.0, + "step": 11721 + }, + { + "epoch": 1.287283110037338, + "grad_norm": 2.3398711681365967, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7146919369697571, + "num_tokens": 297755026.0, + "step": 11722 + }, + { + "epoch": 1.2873929277399516, + "grad_norm": 2.1852588653564453, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7469253540039062, + "num_tokens": 297781377.0, + "step": 11723 + }, + { + "epoch": 1.2875027454425654, + "grad_norm": 2.3618075847625732, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7109218239784241, + "num_tokens": 297804441.0, + "step": 11724 + }, + { + "epoch": 1.2876125631451791, + "grad_norm": 2.093055248260498, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7065913677215576, + "num_tokens": 297835198.0, + "step": 11725 + }, + { + "epoch": 1.2877223808477927, + "grad_norm": 2.235196828842163, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7227320671081543, + "num_tokens": 297862015.0, + "step": 11726 + }, + { + "epoch": 1.2878321985504062, + "grad_norm": 2.334301233291626, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7264848947525024, + "num_tokens": 297885928.0, + "step": 11727 + }, + { + "epoch": 1.28794201625302, + "grad_norm": 2.2367777824401855, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.7001520991325378, + "num_tokens": 297914546.0, + "step": 11728 + }, + { + "epoch": 1.2880518339556337, + "grad_norm": 2.3181560039520264, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7183427810668945, + "num_tokens": 297940271.0, + "step": 11729 + }, + { + "epoch": 1.2881616516582473, + "grad_norm": 2.540271520614624, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7220816612243652, + "num_tokens": 297961522.0, + "step": 11730 + }, + { + "epoch": 1.288271469360861, + "grad_norm": 2.3456156253814697, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7054753303527832, + "num_tokens": 297984530.0, + "step": 11731 + }, + { + "epoch": 1.2883812870634745, + "grad_norm": 2.1445395946502686, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7091760635375977, + "num_tokens": 298012988.0, + "step": 11732 + }, + { + "epoch": 1.2884911047660883, + "grad_norm": 2.2561919689178467, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.6987273693084717, + "num_tokens": 298039809.0, + "step": 11733 + }, + { + "epoch": 1.288600922468702, + "grad_norm": 2.325505495071411, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7218731045722961, + "num_tokens": 298064129.0, + "step": 11734 + }, + { + "epoch": 1.2887107401713156, + "grad_norm": 2.342301368713379, + "learning_rate": 1e-06, + "loss": 0.8547, + "mean_token_accuracy": 0.7342594861984253, + "num_tokens": 298088046.0, + "step": 11735 + }, + { + "epoch": 1.2888205578739294, + "grad_norm": 1.9193443059921265, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6952776908874512, + "num_tokens": 298123953.0, + "step": 11736 + }, + { + "epoch": 1.2889303755765429, + "grad_norm": 2.00492787361145, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7380146980285645, + "num_tokens": 298154087.0, + "step": 11737 + }, + { + "epoch": 1.2890401932791566, + "grad_norm": 2.4881479740142822, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7407082319259644, + "num_tokens": 298174849.0, + "step": 11738 + }, + { + "epoch": 1.2891500109817704, + "grad_norm": 2.2022228240966797, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7261443734169006, + "num_tokens": 298200900.0, + "step": 11739 + }, + { + "epoch": 1.289259828684384, + "grad_norm": 2.2152678966522217, + "learning_rate": 1e-06, + "loss": 0.9826, + "mean_token_accuracy": 0.694948673248291, + "num_tokens": 298228531.0, + "step": 11740 + }, + { + "epoch": 1.2893696463869975, + "grad_norm": 2.0707204341888428, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7254898548126221, + "num_tokens": 298260132.0, + "step": 11741 + }, + { + "epoch": 1.2894794640896112, + "grad_norm": 2.249450922012329, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.7028380632400513, + "num_tokens": 298286716.0, + "step": 11742 + }, + { + "epoch": 1.289589281792225, + "grad_norm": 2.036546230316162, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7180432081222534, + "num_tokens": 298317138.0, + "step": 11743 + }, + { + "epoch": 1.2896990994948385, + "grad_norm": 2.2248852252960205, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7422531843185425, + "num_tokens": 298342106.0, + "step": 11744 + }, + { + "epoch": 1.2898089171974523, + "grad_norm": 2.291236639022827, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7318893074989319, + "num_tokens": 298367948.0, + "step": 11745 + }, + { + "epoch": 1.2899187349000658, + "grad_norm": 2.4061520099639893, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.709689736366272, + "num_tokens": 298393742.0, + "step": 11746 + }, + { + "epoch": 1.2900285526026796, + "grad_norm": 2.4031972885131836, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7268093824386597, + "num_tokens": 298416749.0, + "step": 11747 + }, + { + "epoch": 1.2901383703052933, + "grad_norm": 2.086172103881836, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7117563486099243, + "num_tokens": 298448171.0, + "step": 11748 + }, + { + "epoch": 1.2902481880079069, + "grad_norm": 2.4197630882263184, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7142430543899536, + "num_tokens": 298470268.0, + "step": 11749 + }, + { + "epoch": 1.2903580057105206, + "grad_norm": 2.1493139266967773, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.7372674345970154, + "num_tokens": 298496607.0, + "step": 11750 + }, + { + "epoch": 1.2904678234131342, + "grad_norm": 2.306159734725952, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7072033286094666, + "num_tokens": 298523231.0, + "step": 11751 + }, + { + "epoch": 1.290577641115748, + "grad_norm": 2.1419193744659424, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7176221609115601, + "num_tokens": 298549755.0, + "step": 11752 + }, + { + "epoch": 1.2906874588183614, + "grad_norm": 2.3853721618652344, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.70964515209198, + "num_tokens": 298573041.0, + "step": 11753 + }, + { + "epoch": 1.2907972765209752, + "grad_norm": 2.040595054626465, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7015098333358765, + "num_tokens": 298602335.0, + "step": 11754 + }, + { + "epoch": 1.2909070942235887, + "grad_norm": 2.655722141265869, + "learning_rate": 1e-06, + "loss": 0.8252, + "mean_token_accuracy": 0.7374247312545776, + "num_tokens": 298621340.0, + "step": 11755 + }, + { + "epoch": 1.2910169119262025, + "grad_norm": 2.60298228263855, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7325437068939209, + "num_tokens": 298642629.0, + "step": 11756 + }, + { + "epoch": 1.2911267296288162, + "grad_norm": 2.280388116836548, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7126598954200745, + "num_tokens": 298669302.0, + "step": 11757 + }, + { + "epoch": 1.2912365473314298, + "grad_norm": 2.3247334957122803, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.698395848274231, + "num_tokens": 298693259.0, + "step": 11758 + }, + { + "epoch": 1.2913463650340435, + "grad_norm": 2.506260871887207, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7221481800079346, + "num_tokens": 298713457.0, + "step": 11759 + }, + { + "epoch": 1.291456182736657, + "grad_norm": 2.244739055633545, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7222153544425964, + "num_tokens": 298738020.0, + "step": 11760 + }, + { + "epoch": 1.2915660004392708, + "grad_norm": 2.23728609085083, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.7458720207214355, + "num_tokens": 298762972.0, + "step": 11761 + }, + { + "epoch": 1.2916758181418846, + "grad_norm": 2.556046962738037, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7363645434379578, + "num_tokens": 298783198.0, + "step": 11762 + }, + { + "epoch": 1.2917856358444981, + "grad_norm": 2.2992188930511475, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7244932651519775, + "num_tokens": 298806750.0, + "step": 11763 + }, + { + "epoch": 1.2918954535471117, + "grad_norm": 2.1797146797180176, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7265511751174927, + "num_tokens": 298833055.0, + "step": 11764 + }, + { + "epoch": 1.2920052712497254, + "grad_norm": 2.22524356842041, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7075369358062744, + "num_tokens": 298859429.0, + "step": 11765 + }, + { + "epoch": 1.2921150889523392, + "grad_norm": 2.2874698638916016, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.6967770457267761, + "num_tokens": 298885603.0, + "step": 11766 + }, + { + "epoch": 1.2922249066549527, + "grad_norm": 2.027592420578003, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.697119951248169, + "num_tokens": 298917730.0, + "step": 11767 + }, + { + "epoch": 1.2923347243575665, + "grad_norm": 2.6004724502563477, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7327877879142761, + "num_tokens": 298938702.0, + "step": 11768 + }, + { + "epoch": 1.29244454206018, + "grad_norm": 2.541818857192993, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7044272422790527, + "num_tokens": 298961372.0, + "step": 11769 + }, + { + "epoch": 1.2925543597627938, + "grad_norm": 2.4607393741607666, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7348887324333191, + "num_tokens": 298983083.0, + "step": 11770 + }, + { + "epoch": 1.2926641774654075, + "grad_norm": 2.1088528633117676, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7219005227088928, + "num_tokens": 299012408.0, + "step": 11771 + }, + { + "epoch": 1.292773995168021, + "grad_norm": 2.1160285472869873, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7313246726989746, + "num_tokens": 299038921.0, + "step": 11772 + }, + { + "epoch": 1.2928838128706348, + "grad_norm": 2.3605003356933594, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7392949461936951, + "num_tokens": 299062021.0, + "step": 11773 + }, + { + "epoch": 1.2929936305732483, + "grad_norm": 2.2770888805389404, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7000178694725037, + "num_tokens": 299087202.0, + "step": 11774 + }, + { + "epoch": 1.293103448275862, + "grad_norm": 2.473262071609497, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7036128044128418, + "num_tokens": 299110903.0, + "step": 11775 + }, + { + "epoch": 1.2932132659784759, + "grad_norm": 2.236794948577881, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.718135416507721, + "num_tokens": 299136948.0, + "step": 11776 + }, + { + "epoch": 1.2933230836810894, + "grad_norm": 2.2920751571655273, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7333371639251709, + "num_tokens": 299161038.0, + "step": 11777 + }, + { + "epoch": 1.293432901383703, + "grad_norm": 1.9439342021942139, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7317823171615601, + "num_tokens": 299192789.0, + "step": 11778 + }, + { + "epoch": 1.2935427190863167, + "grad_norm": 2.0804238319396973, + "learning_rate": 1e-06, + "loss": 0.7999, + "mean_token_accuracy": 0.7463746070861816, + "num_tokens": 299220111.0, + "step": 11779 + }, + { + "epoch": 1.2936525367889304, + "grad_norm": 2.3055007457733154, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7206012010574341, + "num_tokens": 299241953.0, + "step": 11780 + }, + { + "epoch": 1.293762354491544, + "grad_norm": 2.1839282512664795, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7354003190994263, + "num_tokens": 299266489.0, + "step": 11781 + }, + { + "epoch": 1.2938721721941577, + "grad_norm": 2.415644645690918, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7072576284408569, + "num_tokens": 299288466.0, + "step": 11782 + }, + { + "epoch": 1.2939819898967713, + "grad_norm": 2.600660562515259, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7468007206916809, + "num_tokens": 299306652.0, + "step": 11783 + }, + { + "epoch": 1.294091807599385, + "grad_norm": 2.27776837348938, + "learning_rate": 1e-06, + "loss": 0.9508, + "mean_token_accuracy": 0.7162525653839111, + "num_tokens": 299332185.0, + "step": 11784 + }, + { + "epoch": 1.2942016253019988, + "grad_norm": 2.3769617080688477, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7021594047546387, + "num_tokens": 299355810.0, + "step": 11785 + }, + { + "epoch": 1.2943114430046123, + "grad_norm": 2.2565081119537354, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7322428822517395, + "num_tokens": 299380295.0, + "step": 11786 + }, + { + "epoch": 1.294421260707226, + "grad_norm": 2.0307042598724365, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7110334038734436, + "num_tokens": 299409676.0, + "step": 11787 + }, + { + "epoch": 1.2945310784098396, + "grad_norm": 2.7546746730804443, + "learning_rate": 1e-06, + "loss": 0.8259, + "mean_token_accuracy": 0.737824559211731, + "num_tokens": 299428273.0, + "step": 11788 + }, + { + "epoch": 1.2946408961124534, + "grad_norm": 2.5186967849731445, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7409879565238953, + "num_tokens": 299449974.0, + "step": 11789 + }, + { + "epoch": 1.2947507138150671, + "grad_norm": 2.291095495223999, + "learning_rate": 1e-06, + "loss": 0.7553, + "mean_token_accuracy": 0.7628557682037354, + "num_tokens": 299472819.0, + "step": 11790 + }, + { + "epoch": 1.2948605315176807, + "grad_norm": 2.2386138439178467, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7253566980361938, + "num_tokens": 299500537.0, + "step": 11791 + }, + { + "epoch": 1.2949703492202942, + "grad_norm": 2.458237409591675, + "learning_rate": 1e-06, + "loss": 0.8131, + "mean_token_accuracy": 0.7436820268630981, + "num_tokens": 299521407.0, + "step": 11792 + }, + { + "epoch": 1.295080166922908, + "grad_norm": 2.14540433883667, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.690385103225708, + "num_tokens": 299549647.0, + "step": 11793 + }, + { + "epoch": 1.2951899846255217, + "grad_norm": 2.356712818145752, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7459667325019836, + "num_tokens": 299570374.0, + "step": 11794 + }, + { + "epoch": 1.2952998023281352, + "grad_norm": 2.119091510772705, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.712254524230957, + "num_tokens": 299597918.0, + "step": 11795 + }, + { + "epoch": 1.295409620030749, + "grad_norm": 2.326853036880493, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.705640435218811, + "num_tokens": 299621811.0, + "step": 11796 + }, + { + "epoch": 1.2955194377333625, + "grad_norm": 2.392932176589966, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7200006246566772, + "num_tokens": 299644024.0, + "step": 11797 + }, + { + "epoch": 1.2956292554359763, + "grad_norm": 2.271817922592163, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7172171473503113, + "num_tokens": 299667708.0, + "step": 11798 + }, + { + "epoch": 1.29573907313859, + "grad_norm": 2.3788397312164307, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7410439252853394, + "num_tokens": 299689244.0, + "step": 11799 + }, + { + "epoch": 1.2958488908412036, + "grad_norm": 2.4715943336486816, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.7378005981445312, + "num_tokens": 299709428.0, + "step": 11800 + }, + { + "epoch": 1.2959587085438173, + "grad_norm": 2.6266896724700928, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7317905426025391, + "num_tokens": 299728372.0, + "step": 11801 + }, + { + "epoch": 1.2960685262464309, + "grad_norm": 2.213139772415161, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7123522162437439, + "num_tokens": 299755450.0, + "step": 11802 + }, + { + "epoch": 1.2961783439490446, + "grad_norm": 2.2109289169311523, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.6865953207015991, + "num_tokens": 299780864.0, + "step": 11803 + }, + { + "epoch": 1.2962881616516582, + "grad_norm": 2.6327548027038574, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7116577625274658, + "num_tokens": 299800798.0, + "step": 11804 + }, + { + "epoch": 1.296397979354272, + "grad_norm": 2.5443613529205322, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7273874282836914, + "num_tokens": 299822586.0, + "step": 11805 + }, + { + "epoch": 1.2965077970568855, + "grad_norm": 2.1790058612823486, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7112985849380493, + "num_tokens": 299848724.0, + "step": 11806 + }, + { + "epoch": 1.2966176147594992, + "grad_norm": 2.034761667251587, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7118750810623169, + "num_tokens": 299879416.0, + "step": 11807 + }, + { + "epoch": 1.296727432462113, + "grad_norm": 1.9616414308547974, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.714746356010437, + "num_tokens": 299911468.0, + "step": 11808 + }, + { + "epoch": 1.2968372501647265, + "grad_norm": 2.3200910091400146, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7230385541915894, + "num_tokens": 299934759.0, + "step": 11809 + }, + { + "epoch": 1.2969470678673403, + "grad_norm": 2.4796905517578125, + "learning_rate": 1e-06, + "loss": 0.8587, + "mean_token_accuracy": 0.729041337966919, + "num_tokens": 299954400.0, + "step": 11810 + }, + { + "epoch": 1.2970568855699538, + "grad_norm": 2.2568345069885254, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.7436226606369019, + "num_tokens": 299977250.0, + "step": 11811 + }, + { + "epoch": 1.2971667032725676, + "grad_norm": 2.342268228530884, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7298685312271118, + "num_tokens": 299999128.0, + "step": 11812 + }, + { + "epoch": 1.2972765209751813, + "grad_norm": 2.2033047676086426, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7220237255096436, + "num_tokens": 300024338.0, + "step": 11813 + }, + { + "epoch": 1.2973863386777948, + "grad_norm": 2.434758424758911, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.6993110179901123, + "num_tokens": 300048802.0, + "step": 11814 + }, + { + "epoch": 1.2974961563804086, + "grad_norm": 2.194331169128418, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7349367141723633, + "num_tokens": 300074330.0, + "step": 11815 + }, + { + "epoch": 1.2976059740830221, + "grad_norm": 2.264604330062866, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7084035873413086, + "num_tokens": 300100536.0, + "step": 11816 + }, + { + "epoch": 1.297715791785636, + "grad_norm": 2.413262367248535, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7140552997589111, + "num_tokens": 300123282.0, + "step": 11817 + }, + { + "epoch": 1.2978256094882494, + "grad_norm": 2.4040026664733887, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7145462036132812, + "num_tokens": 300145277.0, + "step": 11818 + }, + { + "epoch": 1.2979354271908632, + "grad_norm": 2.2156598567962646, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7100772261619568, + "num_tokens": 300172408.0, + "step": 11819 + }, + { + "epoch": 1.2980452448934767, + "grad_norm": 2.12540864944458, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.713343620300293, + "num_tokens": 300202375.0, + "step": 11820 + }, + { + "epoch": 1.2981550625960905, + "grad_norm": 2.0177645683288574, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7462410926818848, + "num_tokens": 300231278.0, + "step": 11821 + }, + { + "epoch": 1.2982648802987042, + "grad_norm": 2.214362859725952, + "learning_rate": 1e-06, + "loss": 0.7709, + "mean_token_accuracy": 0.7508192658424377, + "num_tokens": 300255004.0, + "step": 11822 + }, + { + "epoch": 1.2983746980013178, + "grad_norm": 2.4828171730041504, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7236163020133972, + "num_tokens": 300278940.0, + "step": 11823 + }, + { + "epoch": 1.2984845157039315, + "grad_norm": 2.2808656692504883, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7401198744773865, + "num_tokens": 300302282.0, + "step": 11824 + }, + { + "epoch": 1.298594333406545, + "grad_norm": 2.2551398277282715, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7077720761299133, + "num_tokens": 300328213.0, + "step": 11825 + }, + { + "epoch": 1.2987041511091588, + "grad_norm": 2.041121006011963, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7105787992477417, + "num_tokens": 300358406.0, + "step": 11826 + }, + { + "epoch": 1.2988139688117726, + "grad_norm": 2.2889466285705566, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7282091379165649, + "num_tokens": 300384004.0, + "step": 11827 + }, + { + "epoch": 1.2989237865143861, + "grad_norm": 2.8319640159606934, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7337586879730225, + "num_tokens": 300401247.0, + "step": 11828 + }, + { + "epoch": 1.2990336042169996, + "grad_norm": 2.3895249366760254, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7230902910232544, + "num_tokens": 300424155.0, + "step": 11829 + }, + { + "epoch": 1.2991434219196134, + "grad_norm": 2.1722567081451416, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7013395428657532, + "num_tokens": 300454182.0, + "step": 11830 + }, + { + "epoch": 1.2992532396222272, + "grad_norm": 2.2133092880249023, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7304241061210632, + "num_tokens": 300480332.0, + "step": 11831 + }, + { + "epoch": 1.2993630573248407, + "grad_norm": 2.5400121212005615, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.6971352696418762, + "num_tokens": 300501206.0, + "step": 11832 + }, + { + "epoch": 1.2994728750274545, + "grad_norm": 2.1225051879882812, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7052869200706482, + "num_tokens": 300531529.0, + "step": 11833 + }, + { + "epoch": 1.299582692730068, + "grad_norm": 2.2099108695983887, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7295684814453125, + "num_tokens": 300557455.0, + "step": 11834 + }, + { + "epoch": 1.2996925104326817, + "grad_norm": 2.32201886177063, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.7013161778450012, + "num_tokens": 300584271.0, + "step": 11835 + }, + { + "epoch": 1.2998023281352955, + "grad_norm": 2.7478814125061035, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7427389621734619, + "num_tokens": 300601238.0, + "step": 11836 + }, + { + "epoch": 1.299912145837909, + "grad_norm": 2.4475467205047607, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7129337787628174, + "num_tokens": 300623058.0, + "step": 11837 + }, + { + "epoch": 1.3000219635405228, + "grad_norm": 2.6077656745910645, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7282711863517761, + "num_tokens": 300643946.0, + "step": 11838 + }, + { + "epoch": 1.3001317812431363, + "grad_norm": 2.1168086528778076, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7038906812667847, + "num_tokens": 300671798.0, + "step": 11839 + }, + { + "epoch": 1.30024159894575, + "grad_norm": 2.1394670009613037, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7031468152999878, + "num_tokens": 300698325.0, + "step": 11840 + }, + { + "epoch": 1.3003514166483638, + "grad_norm": 2.2535390853881836, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7091154456138611, + "num_tokens": 300724830.0, + "step": 11841 + }, + { + "epoch": 1.3004612343509774, + "grad_norm": 2.211963415145874, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7366341948509216, + "num_tokens": 300751109.0, + "step": 11842 + }, + { + "epoch": 1.300571052053591, + "grad_norm": 2.270768880844116, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7093881964683533, + "num_tokens": 300776151.0, + "step": 11843 + }, + { + "epoch": 1.3006808697562047, + "grad_norm": 2.3003735542297363, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7325520515441895, + "num_tokens": 300799762.0, + "step": 11844 + }, + { + "epoch": 1.3007906874588184, + "grad_norm": 2.276885509490967, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7300389409065247, + "num_tokens": 300825248.0, + "step": 11845 + }, + { + "epoch": 1.300900505161432, + "grad_norm": 2.6621835231781006, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7231706380844116, + "num_tokens": 300845271.0, + "step": 11846 + }, + { + "epoch": 1.3010103228640457, + "grad_norm": 2.0042948722839355, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7059383392333984, + "num_tokens": 300876930.0, + "step": 11847 + }, + { + "epoch": 1.3011201405666593, + "grad_norm": 2.274027109146118, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7330219149589539, + "num_tokens": 300900939.0, + "step": 11848 + }, + { + "epoch": 1.301229958269273, + "grad_norm": 2.4349355697631836, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7167664170265198, + "num_tokens": 300924205.0, + "step": 11849 + }, + { + "epoch": 1.3013397759718868, + "grad_norm": 2.1199123859405518, + "learning_rate": 1e-06, + "loss": 0.9048, + "mean_token_accuracy": 0.7173636555671692, + "num_tokens": 300953331.0, + "step": 11850 + }, + { + "epoch": 1.3014495936745003, + "grad_norm": 2.1932947635650635, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7201908826828003, + "num_tokens": 300978725.0, + "step": 11851 + }, + { + "epoch": 1.301559411377114, + "grad_norm": 2.4186413288116455, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.7081297636032104, + "num_tokens": 301000649.0, + "step": 11852 + }, + { + "epoch": 1.3016692290797276, + "grad_norm": 2.0738039016723633, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7327985763549805, + "num_tokens": 301028487.0, + "step": 11853 + }, + { + "epoch": 1.3017790467823414, + "grad_norm": 2.3911683559417725, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7089009881019592, + "num_tokens": 301052759.0, + "step": 11854 + }, + { + "epoch": 1.301888864484955, + "grad_norm": 2.3395326137542725, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7297523617744446, + "num_tokens": 301076211.0, + "step": 11855 + }, + { + "epoch": 1.3019986821875686, + "grad_norm": 2.5271151065826416, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7428718209266663, + "num_tokens": 301096188.0, + "step": 11856 + }, + { + "epoch": 1.3021084998901822, + "grad_norm": 2.186373710632324, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.711683988571167, + "num_tokens": 301123417.0, + "step": 11857 + }, + { + "epoch": 1.302218317592796, + "grad_norm": 2.7239460945129395, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7232378721237183, + "num_tokens": 301142071.0, + "step": 11858 + }, + { + "epoch": 1.3023281352954097, + "grad_norm": 2.2249562740325928, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7170811891555786, + "num_tokens": 301167042.0, + "step": 11859 + }, + { + "epoch": 1.3024379529980232, + "grad_norm": 2.484882354736328, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7249073386192322, + "num_tokens": 301188283.0, + "step": 11860 + }, + { + "epoch": 1.302547770700637, + "grad_norm": 2.6104114055633545, + "learning_rate": 1e-06, + "loss": 0.7857, + "mean_token_accuracy": 0.7496415972709656, + "num_tokens": 301206158.0, + "step": 11861 + }, + { + "epoch": 1.3026575884032505, + "grad_norm": 2.3066389560699463, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.6864352226257324, + "num_tokens": 301233605.0, + "step": 11862 + }, + { + "epoch": 1.3027674061058643, + "grad_norm": 2.3463375568389893, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7202393412590027, + "num_tokens": 301258020.0, + "step": 11863 + }, + { + "epoch": 1.302877223808478, + "grad_norm": 2.2166507244110107, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.712437629699707, + "num_tokens": 301285278.0, + "step": 11864 + }, + { + "epoch": 1.3029870415110916, + "grad_norm": 2.340592861175537, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7171443700790405, + "num_tokens": 301308920.0, + "step": 11865 + }, + { + "epoch": 1.3030968592137053, + "grad_norm": 2.0496163368225098, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7460775971412659, + "num_tokens": 301336641.0, + "step": 11866 + }, + { + "epoch": 1.3032066769163189, + "grad_norm": 2.1126725673675537, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7217499613761902, + "num_tokens": 301366378.0, + "step": 11867 + }, + { + "epoch": 1.3033164946189326, + "grad_norm": 2.316711664199829, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7320386171340942, + "num_tokens": 301388228.0, + "step": 11868 + }, + { + "epoch": 1.3034263123215462, + "grad_norm": 2.414071798324585, + "learning_rate": 1e-06, + "loss": 0.8383, + "mean_token_accuracy": 0.7367527484893799, + "num_tokens": 301411319.0, + "step": 11869 + }, + { + "epoch": 1.30353613002416, + "grad_norm": 2.250485420227051, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7248440384864807, + "num_tokens": 301435568.0, + "step": 11870 + }, + { + "epoch": 1.3036459477267734, + "grad_norm": 2.01991605758667, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7209044694900513, + "num_tokens": 301465598.0, + "step": 11871 + }, + { + "epoch": 1.3037557654293872, + "grad_norm": 2.1584529876708984, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7221922874450684, + "num_tokens": 301493093.0, + "step": 11872 + }, + { + "epoch": 1.303865583132001, + "grad_norm": 2.1114609241485596, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7247451543807983, + "num_tokens": 301522486.0, + "step": 11873 + }, + { + "epoch": 1.3039754008346145, + "grad_norm": 2.2087173461914062, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7235456705093384, + "num_tokens": 301548008.0, + "step": 11874 + }, + { + "epoch": 1.3040852185372283, + "grad_norm": 2.172755718231201, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7332245111465454, + "num_tokens": 301574154.0, + "step": 11875 + }, + { + "epoch": 1.3041950362398418, + "grad_norm": 2.1372344493865967, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7011674642562866, + "num_tokens": 301602576.0, + "step": 11876 + }, + { + "epoch": 1.3043048539424555, + "grad_norm": 2.233527660369873, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7206921577453613, + "num_tokens": 301629173.0, + "step": 11877 + }, + { + "epoch": 1.3044146716450693, + "grad_norm": 2.4099621772766113, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7269759178161621, + "num_tokens": 301652290.0, + "step": 11878 + }, + { + "epoch": 1.3045244893476828, + "grad_norm": 2.591472864151001, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7127405405044556, + "num_tokens": 301673509.0, + "step": 11879 + }, + { + "epoch": 1.3046343070502964, + "grad_norm": 2.1710855960845947, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7023751735687256, + "num_tokens": 301702916.0, + "step": 11880 + }, + { + "epoch": 1.3047441247529101, + "grad_norm": 2.344238758087158, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7372534275054932, + "num_tokens": 301727475.0, + "step": 11881 + }, + { + "epoch": 1.3048539424555239, + "grad_norm": 2.350905418395996, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.711017370223999, + "num_tokens": 301750269.0, + "step": 11882 + }, + { + "epoch": 1.3049637601581374, + "grad_norm": 2.4994399547576904, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7404553890228271, + "num_tokens": 301770559.0, + "step": 11883 + }, + { + "epoch": 1.3050735778607512, + "grad_norm": 2.399033546447754, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7218810319900513, + "num_tokens": 301793584.0, + "step": 11884 + }, + { + "epoch": 1.3051833955633647, + "grad_norm": 2.336796998977661, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7109004855155945, + "num_tokens": 301816918.0, + "step": 11885 + }, + { + "epoch": 1.3052932132659785, + "grad_norm": 2.3886208534240723, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7354727387428284, + "num_tokens": 301839644.0, + "step": 11886 + }, + { + "epoch": 1.3054030309685922, + "grad_norm": 2.145988941192627, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7000744342803955, + "num_tokens": 301868956.0, + "step": 11887 + }, + { + "epoch": 1.3055128486712058, + "grad_norm": 2.249040365219116, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7261977195739746, + "num_tokens": 301894897.0, + "step": 11888 + }, + { + "epoch": 1.3056226663738195, + "grad_norm": 1.9826898574829102, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7257887125015259, + "num_tokens": 301925448.0, + "step": 11889 + }, + { + "epoch": 1.305732484076433, + "grad_norm": 2.2801613807678223, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.7039564251899719, + "num_tokens": 301958648.0, + "step": 11890 + }, + { + "epoch": 1.3058423017790468, + "grad_norm": 2.77877140045166, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7480343580245972, + "num_tokens": 301977142.0, + "step": 11891 + }, + { + "epoch": 1.3059521194816606, + "grad_norm": 2.180058717727661, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7453663349151611, + "num_tokens": 302004262.0, + "step": 11892 + }, + { + "epoch": 1.306061937184274, + "grad_norm": 2.194795608520508, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7183046340942383, + "num_tokens": 302031881.0, + "step": 11893 + }, + { + "epoch": 1.3061717548868876, + "grad_norm": 2.047433376312256, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7294158935546875, + "num_tokens": 302062148.0, + "step": 11894 + }, + { + "epoch": 1.3062815725895014, + "grad_norm": 2.1839849948883057, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7277267575263977, + "num_tokens": 302088940.0, + "step": 11895 + }, + { + "epoch": 1.3063913902921152, + "grad_norm": 2.4476237297058105, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7230035066604614, + "num_tokens": 302109260.0, + "step": 11896 + }, + { + "epoch": 1.3065012079947287, + "grad_norm": 2.4690001010894775, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7384200096130371, + "num_tokens": 302131145.0, + "step": 11897 + }, + { + "epoch": 1.3066110256973424, + "grad_norm": 2.3723366260528564, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7354641556739807, + "num_tokens": 302157499.0, + "step": 11898 + }, + { + "epoch": 1.306720843399956, + "grad_norm": 2.267381191253662, + "learning_rate": 1e-06, + "loss": 0.9955, + "mean_token_accuracy": 0.698286771774292, + "num_tokens": 302181498.0, + "step": 11899 + }, + { + "epoch": 1.3068306611025697, + "grad_norm": 2.1134145259857178, + "learning_rate": 1e-06, + "loss": 0.997, + "mean_token_accuracy": 0.6954849362373352, + "num_tokens": 302209747.0, + "step": 11900 + }, + { + "epoch": 1.3069404788051835, + "grad_norm": 2.162200927734375, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7151279449462891, + "num_tokens": 302235239.0, + "step": 11901 + }, + { + "epoch": 1.307050296507797, + "grad_norm": 2.0415263175964355, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7061187028884888, + "num_tokens": 302265405.0, + "step": 11902 + }, + { + "epoch": 1.3071601142104108, + "grad_norm": 2.4252054691314697, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7242907285690308, + "num_tokens": 302290304.0, + "step": 11903 + }, + { + "epoch": 1.3072699319130243, + "grad_norm": 2.361980676651001, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7249764204025269, + "num_tokens": 302313962.0, + "step": 11904 + }, + { + "epoch": 1.307379749615638, + "grad_norm": 2.205016851425171, + "learning_rate": 1e-06, + "loss": 0.9848, + "mean_token_accuracy": 0.7014122605323792, + "num_tokens": 302343649.0, + "step": 11905 + }, + { + "epoch": 1.3074895673182518, + "grad_norm": 2.4375524520874023, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7261326909065247, + "num_tokens": 302363964.0, + "step": 11906 + }, + { + "epoch": 1.3075993850208654, + "grad_norm": 2.0679831504821777, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.6998483538627625, + "num_tokens": 302394406.0, + "step": 11907 + }, + { + "epoch": 1.307709202723479, + "grad_norm": 1.9992635250091553, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7071850895881653, + "num_tokens": 302424300.0, + "step": 11908 + }, + { + "epoch": 1.3078190204260927, + "grad_norm": 2.1540167331695557, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7080292701721191, + "num_tokens": 302452620.0, + "step": 11909 + }, + { + "epoch": 1.3079288381287064, + "grad_norm": 2.2319984436035156, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7067031264305115, + "num_tokens": 302478875.0, + "step": 11910 + }, + { + "epoch": 1.30803865583132, + "grad_norm": 2.471299886703491, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7285494208335876, + "num_tokens": 302501101.0, + "step": 11911 + }, + { + "epoch": 1.3081484735339337, + "grad_norm": 2.661367177963257, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7393805980682373, + "num_tokens": 302521024.0, + "step": 11912 + }, + { + "epoch": 1.3082582912365472, + "grad_norm": 2.183907985687256, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7069184184074402, + "num_tokens": 302549659.0, + "step": 11913 + }, + { + "epoch": 1.308368108939161, + "grad_norm": 2.52860164642334, + "learning_rate": 1e-06, + "loss": 0.7672, + "mean_token_accuracy": 0.7508410811424255, + "num_tokens": 302568699.0, + "step": 11914 + }, + { + "epoch": 1.3084779266417748, + "grad_norm": 1.9695639610290527, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7107580304145813, + "num_tokens": 302599900.0, + "step": 11915 + }, + { + "epoch": 1.3085877443443883, + "grad_norm": 2.331214427947998, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7048932909965515, + "num_tokens": 302624476.0, + "step": 11916 + }, + { + "epoch": 1.308697562047002, + "grad_norm": 2.3745837211608887, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7472133636474609, + "num_tokens": 302647848.0, + "step": 11917 + }, + { + "epoch": 1.3088073797496156, + "grad_norm": 1.945879578590393, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.6986618041992188, + "num_tokens": 302680270.0, + "step": 11918 + }, + { + "epoch": 1.3089171974522293, + "grad_norm": 2.287349224090576, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7225443124771118, + "num_tokens": 302705608.0, + "step": 11919 + }, + { + "epoch": 1.309027015154843, + "grad_norm": 2.2560386657714844, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7356605529785156, + "num_tokens": 302731326.0, + "step": 11920 + }, + { + "epoch": 1.3091368328574566, + "grad_norm": 2.108774185180664, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.706916093826294, + "num_tokens": 302759180.0, + "step": 11921 + }, + { + "epoch": 1.3092466505600702, + "grad_norm": 2.282392740249634, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7220404744148254, + "num_tokens": 302785192.0, + "step": 11922 + }, + { + "epoch": 1.309356468262684, + "grad_norm": 2.4613351821899414, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7282355427742004, + "num_tokens": 302806093.0, + "step": 11923 + }, + { + "epoch": 1.3094662859652977, + "grad_norm": 2.235327959060669, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7478011846542358, + "num_tokens": 302831836.0, + "step": 11924 + }, + { + "epoch": 1.3095761036679112, + "grad_norm": 2.4918346405029297, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7294715642929077, + "num_tokens": 302855537.0, + "step": 11925 + }, + { + "epoch": 1.309685921370525, + "grad_norm": 2.2101492881774902, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7115856409072876, + "num_tokens": 302880451.0, + "step": 11926 + }, + { + "epoch": 1.3097957390731385, + "grad_norm": 2.467830181121826, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7360320091247559, + "num_tokens": 302902961.0, + "step": 11927 + }, + { + "epoch": 1.3099055567757523, + "grad_norm": 2.1498053073883057, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.727321982383728, + "num_tokens": 302930438.0, + "step": 11928 + }, + { + "epoch": 1.310015374478366, + "grad_norm": 1.9601565599441528, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.687251091003418, + "num_tokens": 302963354.0, + "step": 11929 + }, + { + "epoch": 1.3101251921809796, + "grad_norm": 2.1656429767608643, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.6995441913604736, + "num_tokens": 302989506.0, + "step": 11930 + }, + { + "epoch": 1.3102350098835933, + "grad_norm": 2.154369354248047, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7139171957969666, + "num_tokens": 303020182.0, + "step": 11931 + }, + { + "epoch": 1.3103448275862069, + "grad_norm": 1.997469186782837, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7241067290306091, + "num_tokens": 303050698.0, + "step": 11932 + }, + { + "epoch": 1.3104546452888206, + "grad_norm": 2.1182353496551514, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7089486122131348, + "num_tokens": 303079655.0, + "step": 11933 + }, + { + "epoch": 1.3105644629914341, + "grad_norm": 2.0540664196014404, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7076712846755981, + "num_tokens": 303108133.0, + "step": 11934 + }, + { + "epoch": 1.310674280694048, + "grad_norm": 2.230545997619629, + "learning_rate": 1e-06, + "loss": 1.0453, + "mean_token_accuracy": 0.688041627407074, + "num_tokens": 303138216.0, + "step": 11935 + }, + { + "epoch": 1.3107840983966614, + "grad_norm": 2.3251869678497314, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.711173415184021, + "num_tokens": 303164648.0, + "step": 11936 + }, + { + "epoch": 1.3108939160992752, + "grad_norm": 2.376209020614624, + "learning_rate": 1e-06, + "loss": 0.9851, + "mean_token_accuracy": 0.6990672945976257, + "num_tokens": 303188772.0, + "step": 11937 + }, + { + "epoch": 1.311003733801889, + "grad_norm": 2.213550329208374, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7196694612503052, + "num_tokens": 303214145.0, + "step": 11938 + }, + { + "epoch": 1.3111135515045025, + "grad_norm": 2.401414155960083, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7261327505111694, + "num_tokens": 303237153.0, + "step": 11939 + }, + { + "epoch": 1.3112233692071162, + "grad_norm": 2.246046543121338, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7367745041847229, + "num_tokens": 303263323.0, + "step": 11940 + }, + { + "epoch": 1.3113331869097298, + "grad_norm": 2.084684371948242, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6952805519104004, + "num_tokens": 303291952.0, + "step": 11941 + }, + { + "epoch": 1.3114430046123435, + "grad_norm": 2.322948455810547, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7073581218719482, + "num_tokens": 303317470.0, + "step": 11942 + }, + { + "epoch": 1.3115528223149573, + "grad_norm": 2.4650473594665527, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7174564003944397, + "num_tokens": 303340859.0, + "step": 11943 + }, + { + "epoch": 1.3116626400175708, + "grad_norm": 2.2335169315338135, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6986122131347656, + "num_tokens": 303366919.0, + "step": 11944 + }, + { + "epoch": 1.3117724577201844, + "grad_norm": 2.2993955612182617, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7012584209442139, + "num_tokens": 303391464.0, + "step": 11945 + }, + { + "epoch": 1.3118822754227981, + "grad_norm": 2.275510311126709, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7000175714492798, + "num_tokens": 303417026.0, + "step": 11946 + }, + { + "epoch": 1.3119920931254119, + "grad_norm": 2.088318347930908, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7137653827667236, + "num_tokens": 303443138.0, + "step": 11947 + }, + { + "epoch": 1.3121019108280254, + "grad_norm": 2.095571994781494, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7039498090744019, + "num_tokens": 303470658.0, + "step": 11948 + }, + { + "epoch": 1.3122117285306392, + "grad_norm": 2.2054004669189453, + "learning_rate": 1e-06, + "loss": 1.0131, + "mean_token_accuracy": 0.6964969038963318, + "num_tokens": 303496222.0, + "step": 11949 + }, + { + "epoch": 1.3123215462332527, + "grad_norm": 2.2302913665771484, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7029336094856262, + "num_tokens": 303523720.0, + "step": 11950 + }, + { + "epoch": 1.3124313639358665, + "grad_norm": 2.2593512535095215, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.726997971534729, + "num_tokens": 303547783.0, + "step": 11951 + }, + { + "epoch": 1.3125411816384802, + "grad_norm": 2.313966989517212, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7102827429771423, + "num_tokens": 303571423.0, + "step": 11952 + }, + { + "epoch": 1.3126509993410937, + "grad_norm": 2.165828227996826, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7089836001396179, + "num_tokens": 303596828.0, + "step": 11953 + }, + { + "epoch": 1.3127608170437075, + "grad_norm": 2.2487757205963135, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.6989104747772217, + "num_tokens": 303624766.0, + "step": 11954 + }, + { + "epoch": 1.312870634746321, + "grad_norm": 2.305103063583374, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7200117111206055, + "num_tokens": 303648913.0, + "step": 11955 + }, + { + "epoch": 1.3129804524489348, + "grad_norm": 2.702524423599243, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7285361289978027, + "num_tokens": 303667885.0, + "step": 11956 + }, + { + "epoch": 1.3130902701515486, + "grad_norm": 2.448347568511963, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7273145914077759, + "num_tokens": 303689755.0, + "step": 11957 + }, + { + "epoch": 1.313200087854162, + "grad_norm": 2.4047372341156006, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.726611852645874, + "num_tokens": 303710114.0, + "step": 11958 + }, + { + "epoch": 1.3133099055567756, + "grad_norm": 2.1377696990966797, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.697796106338501, + "num_tokens": 303739931.0, + "step": 11959 + }, + { + "epoch": 1.3134197232593894, + "grad_norm": 2.585799217224121, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7209734916687012, + "num_tokens": 303760179.0, + "step": 11960 + }, + { + "epoch": 1.3135295409620031, + "grad_norm": 2.0730538368225098, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.705644965171814, + "num_tokens": 303790133.0, + "step": 11961 + }, + { + "epoch": 1.3136393586646167, + "grad_norm": 2.1143243312835693, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7106757164001465, + "num_tokens": 303821774.0, + "step": 11962 + }, + { + "epoch": 1.3137491763672304, + "grad_norm": 2.380380630493164, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7182984352111816, + "num_tokens": 303845682.0, + "step": 11963 + }, + { + "epoch": 1.313858994069844, + "grad_norm": 2.378220796585083, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7498235106468201, + "num_tokens": 303868734.0, + "step": 11964 + }, + { + "epoch": 1.3139688117724577, + "grad_norm": 2.3396453857421875, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.6998358964920044, + "num_tokens": 303894565.0, + "step": 11965 + }, + { + "epoch": 1.3140786294750715, + "grad_norm": 2.0763611793518066, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7036126852035522, + "num_tokens": 303922471.0, + "step": 11966 + }, + { + "epoch": 1.314188447177685, + "grad_norm": 2.2190685272216797, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7343907356262207, + "num_tokens": 303948081.0, + "step": 11967 + }, + { + "epoch": 1.3142982648802988, + "grad_norm": 2.0712778568267822, + "learning_rate": 1e-06, + "loss": 1.0443, + "mean_token_accuracy": 0.6861963272094727, + "num_tokens": 303977913.0, + "step": 11968 + }, + { + "epoch": 1.3144080825829123, + "grad_norm": 2.467172384262085, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7324707508087158, + "num_tokens": 303999160.0, + "step": 11969 + }, + { + "epoch": 1.314517900285526, + "grad_norm": 2.10634708404541, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7131043076515198, + "num_tokens": 304028200.0, + "step": 11970 + }, + { + "epoch": 1.3146277179881398, + "grad_norm": 2.4330263137817383, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7082697749137878, + "num_tokens": 304050898.0, + "step": 11971 + }, + { + "epoch": 1.3147375356907534, + "grad_norm": 2.4483792781829834, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7275513410568237, + "num_tokens": 304072451.0, + "step": 11972 + }, + { + "epoch": 1.314847353393367, + "grad_norm": 2.1955347061157227, + "learning_rate": 1e-06, + "loss": 0.9921, + "mean_token_accuracy": 0.6961817741394043, + "num_tokens": 304100716.0, + "step": 11973 + }, + { + "epoch": 1.3149571710959806, + "grad_norm": 2.541567087173462, + "learning_rate": 1e-06, + "loss": 0.906, + "mean_token_accuracy": 0.7158465385437012, + "num_tokens": 304121839.0, + "step": 11974 + }, + { + "epoch": 1.3150669887985944, + "grad_norm": 2.2433059215545654, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.714511513710022, + "num_tokens": 304146152.0, + "step": 11975 + }, + { + "epoch": 1.315176806501208, + "grad_norm": 2.1642932891845703, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7093260884284973, + "num_tokens": 304173333.0, + "step": 11976 + }, + { + "epoch": 1.3152866242038217, + "grad_norm": 2.229865550994873, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7301229238510132, + "num_tokens": 304196897.0, + "step": 11977 + }, + { + "epoch": 1.3153964419064352, + "grad_norm": 2.017193555831909, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7031123638153076, + "num_tokens": 304228217.0, + "step": 11978 + }, + { + "epoch": 1.315506259609049, + "grad_norm": 2.4445111751556396, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7129826545715332, + "num_tokens": 304250590.0, + "step": 11979 + }, + { + "epoch": 1.3156160773116627, + "grad_norm": 2.167588710784912, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7139236330986023, + "num_tokens": 304275741.0, + "step": 11980 + }, + { + "epoch": 1.3157258950142763, + "grad_norm": 1.8999630212783813, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.726377010345459, + "num_tokens": 304309258.0, + "step": 11981 + }, + { + "epoch": 1.31583571271689, + "grad_norm": 2.5397021770477295, + "learning_rate": 1e-06, + "loss": 0.7979, + "mean_token_accuracy": 0.7427936792373657, + "num_tokens": 304328470.0, + "step": 11982 + }, + { + "epoch": 1.3159455304195036, + "grad_norm": 2.2303686141967773, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7249687910079956, + "num_tokens": 304354873.0, + "step": 11983 + }, + { + "epoch": 1.3160553481221173, + "grad_norm": 2.2748818397521973, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7308865785598755, + "num_tokens": 304380881.0, + "step": 11984 + }, + { + "epoch": 1.3161651658247309, + "grad_norm": 2.0960190296173096, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7269148826599121, + "num_tokens": 304408674.0, + "step": 11985 + }, + { + "epoch": 1.3162749835273446, + "grad_norm": 2.257009506225586, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7048326730728149, + "num_tokens": 304436216.0, + "step": 11986 + }, + { + "epoch": 1.3163848012299582, + "grad_norm": 2.4772579669952393, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7338517904281616, + "num_tokens": 304456971.0, + "step": 11987 + }, + { + "epoch": 1.316494618932572, + "grad_norm": 2.3833870887756348, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7069990634918213, + "num_tokens": 304479336.0, + "step": 11988 + }, + { + "epoch": 1.3166044366351857, + "grad_norm": 2.1374340057373047, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7424819469451904, + "num_tokens": 304507803.0, + "step": 11989 + }, + { + "epoch": 1.3167142543377992, + "grad_norm": 1.993470549583435, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.6988320350646973, + "num_tokens": 304538235.0, + "step": 11990 + }, + { + "epoch": 1.316824072040413, + "grad_norm": 2.4226791858673096, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.732571542263031, + "num_tokens": 304561575.0, + "step": 11991 + }, + { + "epoch": 1.3169338897430265, + "grad_norm": 2.2829689979553223, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7037392258644104, + "num_tokens": 304587936.0, + "step": 11992 + }, + { + "epoch": 1.3170437074456403, + "grad_norm": 2.170168876647949, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.711074709892273, + "num_tokens": 304616929.0, + "step": 11993 + }, + { + "epoch": 1.317153525148254, + "grad_norm": 2.4068193435668945, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7329343557357788, + "num_tokens": 304640497.0, + "step": 11994 + }, + { + "epoch": 1.3172633428508675, + "grad_norm": 2.1027300357818604, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7263210415840149, + "num_tokens": 304668535.0, + "step": 11995 + }, + { + "epoch": 1.3173731605534813, + "grad_norm": 2.4263575077056885, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7225592136383057, + "num_tokens": 304690950.0, + "step": 11996 + }, + { + "epoch": 1.3174829782560948, + "grad_norm": 2.2085912227630615, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7507867813110352, + "num_tokens": 304717031.0, + "step": 11997 + }, + { + "epoch": 1.3175927959587086, + "grad_norm": 2.174804210662842, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6908271312713623, + "num_tokens": 304745678.0, + "step": 11998 + }, + { + "epoch": 1.3177026136613221, + "grad_norm": 2.1263885498046875, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.7067769765853882, + "num_tokens": 304774104.0, + "step": 11999 + }, + { + "epoch": 1.3178124313639359, + "grad_norm": 2.236837148666382, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7108433842658997, + "num_tokens": 304801548.0, + "step": 12000 + }, + { + "epoch": 1.3179222490665494, + "grad_norm": 2.592109441757202, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7277030348777771, + "num_tokens": 304820689.0, + "step": 12001 + }, + { + "epoch": 1.3180320667691632, + "grad_norm": 2.3097102642059326, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7114925384521484, + "num_tokens": 304846573.0, + "step": 12002 + }, + { + "epoch": 1.318141884471777, + "grad_norm": 2.1771202087402344, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7223703861236572, + "num_tokens": 304870919.0, + "step": 12003 + }, + { + "epoch": 1.3182517021743905, + "grad_norm": 2.2575454711914062, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7014818787574768, + "num_tokens": 304896343.0, + "step": 12004 + }, + { + "epoch": 1.3183615198770042, + "grad_norm": 2.364093065261841, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7243408560752869, + "num_tokens": 304921316.0, + "step": 12005 + }, + { + "epoch": 1.3184713375796178, + "grad_norm": 2.3799962997436523, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7082894444465637, + "num_tokens": 304946307.0, + "step": 12006 + }, + { + "epoch": 1.3185811552822315, + "grad_norm": 2.23667573928833, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.702908992767334, + "num_tokens": 304973618.0, + "step": 12007 + }, + { + "epoch": 1.3186909729848453, + "grad_norm": 2.362919569015503, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7272642850875854, + "num_tokens": 304997032.0, + "step": 12008 + }, + { + "epoch": 1.3188007906874588, + "grad_norm": 2.3453211784362793, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7287462949752808, + "num_tokens": 305021060.0, + "step": 12009 + }, + { + "epoch": 1.3189106083900723, + "grad_norm": 2.5176339149475098, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7254644632339478, + "num_tokens": 305045187.0, + "step": 12010 + }, + { + "epoch": 1.319020426092686, + "grad_norm": 2.5135912895202637, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7393056154251099, + "num_tokens": 305065171.0, + "step": 12011 + }, + { + "epoch": 1.3191302437952999, + "grad_norm": 2.483212471008301, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7230179309844971, + "num_tokens": 305088372.0, + "step": 12012 + }, + { + "epoch": 1.3192400614979134, + "grad_norm": 2.0911662578582764, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7175400257110596, + "num_tokens": 305115674.0, + "step": 12013 + }, + { + "epoch": 1.3193498792005272, + "grad_norm": 2.25001859664917, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7212626934051514, + "num_tokens": 305141952.0, + "step": 12014 + }, + { + "epoch": 1.3194596969031407, + "grad_norm": 2.456333875656128, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7346454858779907, + "num_tokens": 305163661.0, + "step": 12015 + }, + { + "epoch": 1.3195695146057544, + "grad_norm": 2.5058889389038086, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7360843420028687, + "num_tokens": 305185489.0, + "step": 12016 + }, + { + "epoch": 1.3196793323083682, + "grad_norm": 2.4092307090759277, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7352656722068787, + "num_tokens": 305208191.0, + "step": 12017 + }, + { + "epoch": 1.3197891500109817, + "grad_norm": 2.3768510818481445, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7248084545135498, + "num_tokens": 305231249.0, + "step": 12018 + }, + { + "epoch": 1.3198989677135955, + "grad_norm": 2.6466593742370605, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7358027696609497, + "num_tokens": 305251349.0, + "step": 12019 + }, + { + "epoch": 1.320008785416209, + "grad_norm": 2.0737173557281494, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7392700910568237, + "num_tokens": 305279332.0, + "step": 12020 + }, + { + "epoch": 1.3201186031188228, + "grad_norm": 2.3125200271606445, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7222009897232056, + "num_tokens": 305304677.0, + "step": 12021 + }, + { + "epoch": 1.3202284208214365, + "grad_norm": 2.1975159645080566, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7346339225769043, + "num_tokens": 305331221.0, + "step": 12022 + }, + { + "epoch": 1.32033823852405, + "grad_norm": 2.4162065982818604, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7320774793624878, + "num_tokens": 305352808.0, + "step": 12023 + }, + { + "epoch": 1.3204480562266636, + "grad_norm": 2.2716965675354004, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7048634886741638, + "num_tokens": 305379543.0, + "step": 12024 + }, + { + "epoch": 1.3205578739292774, + "grad_norm": 2.1940863132476807, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.7034143209457397, + "num_tokens": 305407220.0, + "step": 12025 + }, + { + "epoch": 1.3206676916318911, + "grad_norm": 2.1514892578125, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7272982001304626, + "num_tokens": 305433835.0, + "step": 12026 + }, + { + "epoch": 1.3207775093345047, + "grad_norm": 2.118988275527954, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.704672634601593, + "num_tokens": 305462479.0, + "step": 12027 + }, + { + "epoch": 1.3208873270371184, + "grad_norm": 2.1867048740386963, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7310940027236938, + "num_tokens": 305488622.0, + "step": 12028 + }, + { + "epoch": 1.320997144739732, + "grad_norm": 2.455911636352539, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7372313141822815, + "num_tokens": 305510091.0, + "step": 12029 + }, + { + "epoch": 1.3211069624423457, + "grad_norm": 2.4666733741760254, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.709011971950531, + "num_tokens": 305532468.0, + "step": 12030 + }, + { + "epoch": 1.3212167801449595, + "grad_norm": 2.170132875442505, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7195165753364563, + "num_tokens": 305558651.0, + "step": 12031 + }, + { + "epoch": 1.321326597847573, + "grad_norm": 2.5616681575775146, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7240337133407593, + "num_tokens": 305579262.0, + "step": 12032 + }, + { + "epoch": 1.3214364155501868, + "grad_norm": 2.215343713760376, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.732997715473175, + "num_tokens": 305604080.0, + "step": 12033 + }, + { + "epoch": 1.3215462332528003, + "grad_norm": 2.1451330184936523, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7130084037780762, + "num_tokens": 305631440.0, + "step": 12034 + }, + { + "epoch": 1.321656050955414, + "grad_norm": 2.3658039569854736, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7348929643630981, + "num_tokens": 305655747.0, + "step": 12035 + }, + { + "epoch": 1.3217658686580278, + "grad_norm": 2.1454412937164307, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7132657170295715, + "num_tokens": 305683109.0, + "step": 12036 + }, + { + "epoch": 1.3218756863606413, + "grad_norm": 2.2798731327056885, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.716425895690918, + "num_tokens": 305707178.0, + "step": 12037 + }, + { + "epoch": 1.3219855040632549, + "grad_norm": 2.2542550563812256, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7374815940856934, + "num_tokens": 305732391.0, + "step": 12038 + }, + { + "epoch": 1.3220953217658686, + "grad_norm": 2.14247989654541, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7060526013374329, + "num_tokens": 305761586.0, + "step": 12039 + }, + { + "epoch": 1.3222051394684824, + "grad_norm": 2.2511885166168213, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7096981406211853, + "num_tokens": 305785626.0, + "step": 12040 + }, + { + "epoch": 1.322314957171096, + "grad_norm": 2.080974817276001, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.7081567645072937, + "num_tokens": 305815452.0, + "step": 12041 + }, + { + "epoch": 1.3224247748737097, + "grad_norm": 2.0618691444396973, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7029159665107727, + "num_tokens": 305845155.0, + "step": 12042 + }, + { + "epoch": 1.3225345925763232, + "grad_norm": 2.3005194664001465, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7190895676612854, + "num_tokens": 305869497.0, + "step": 12043 + }, + { + "epoch": 1.322644410278937, + "grad_norm": 2.4059646129608154, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7334139943122864, + "num_tokens": 305892110.0, + "step": 12044 + }, + { + "epoch": 1.3227542279815507, + "grad_norm": 2.5034215450286865, + "learning_rate": 1e-06, + "loss": 0.8491, + "mean_token_accuracy": 0.7312003970146179, + "num_tokens": 305912416.0, + "step": 12045 + }, + { + "epoch": 1.3228640456841643, + "grad_norm": 2.1996874809265137, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7155005931854248, + "num_tokens": 305940422.0, + "step": 12046 + }, + { + "epoch": 1.322973863386778, + "grad_norm": 2.2878947257995605, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6972984671592712, + "num_tokens": 305968191.0, + "step": 12047 + }, + { + "epoch": 1.3230836810893916, + "grad_norm": 2.2600860595703125, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7340725660324097, + "num_tokens": 305992627.0, + "step": 12048 + }, + { + "epoch": 1.3231934987920053, + "grad_norm": 2.2172579765319824, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7183738350868225, + "num_tokens": 306019837.0, + "step": 12049 + }, + { + "epoch": 1.3233033164946189, + "grad_norm": 1.9807806015014648, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.736366868019104, + "num_tokens": 306049731.0, + "step": 12050 + }, + { + "epoch": 1.3234131341972326, + "grad_norm": 2.2448067665100098, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7233268022537231, + "num_tokens": 306076406.0, + "step": 12051 + }, + { + "epoch": 1.3235229518998461, + "grad_norm": 2.233550786972046, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.73134446144104, + "num_tokens": 306100960.0, + "step": 12052 + }, + { + "epoch": 1.32363276960246, + "grad_norm": 2.0995724201202393, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7227019667625427, + "num_tokens": 306130823.0, + "step": 12053 + }, + { + "epoch": 1.3237425873050737, + "grad_norm": 2.0932271480560303, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7168083190917969, + "num_tokens": 306161467.0, + "step": 12054 + }, + { + "epoch": 1.3238524050076872, + "grad_norm": 2.007063150405884, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7197837829589844, + "num_tokens": 306190497.0, + "step": 12055 + }, + { + "epoch": 1.323962222710301, + "grad_norm": 2.3310813903808594, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7293078899383545, + "num_tokens": 306212668.0, + "step": 12056 + }, + { + "epoch": 1.3240720404129145, + "grad_norm": 2.204949378967285, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7313067317008972, + "num_tokens": 306237751.0, + "step": 12057 + }, + { + "epoch": 1.3241818581155282, + "grad_norm": 2.4841225147247314, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7326458692550659, + "num_tokens": 306259418.0, + "step": 12058 + }, + { + "epoch": 1.324291675818142, + "grad_norm": 2.5395514965057373, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7253327369689941, + "num_tokens": 306280916.0, + "step": 12059 + }, + { + "epoch": 1.3244014935207555, + "grad_norm": 2.127243995666504, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7260584831237793, + "num_tokens": 306307790.0, + "step": 12060 + }, + { + "epoch": 1.324511311223369, + "grad_norm": 2.330327272415161, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7187488675117493, + "num_tokens": 306333047.0, + "step": 12061 + }, + { + "epoch": 1.3246211289259828, + "grad_norm": 2.573805332183838, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7290820479393005, + "num_tokens": 306353936.0, + "step": 12062 + }, + { + "epoch": 1.3247309466285966, + "grad_norm": 2.2249510288238525, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.6978588104248047, + "num_tokens": 306384022.0, + "step": 12063 + }, + { + "epoch": 1.3248407643312101, + "grad_norm": 2.4923038482666016, + "learning_rate": 1e-06, + "loss": 0.9886, + "mean_token_accuracy": 0.706965982913971, + "num_tokens": 306406864.0, + "step": 12064 + }, + { + "epoch": 1.3249505820338239, + "grad_norm": 1.8779741525650024, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7012990117073059, + "num_tokens": 306443295.0, + "step": 12065 + }, + { + "epoch": 1.3250603997364374, + "grad_norm": 2.419908046722412, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7140369415283203, + "num_tokens": 306464725.0, + "step": 12066 + }, + { + "epoch": 1.3251702174390512, + "grad_norm": 2.5291354656219482, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7196513414382935, + "num_tokens": 306485732.0, + "step": 12067 + }, + { + "epoch": 1.325280035141665, + "grad_norm": 2.235910415649414, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.726646900177002, + "num_tokens": 306511666.0, + "step": 12068 + }, + { + "epoch": 1.3253898528442785, + "grad_norm": 2.3262877464294434, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7031524181365967, + "num_tokens": 306537228.0, + "step": 12069 + }, + { + "epoch": 1.3254996705468922, + "grad_norm": 2.035909652709961, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7125727534294128, + "num_tokens": 306565331.0, + "step": 12070 + }, + { + "epoch": 1.3256094882495058, + "grad_norm": 2.3654415607452393, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7180235981941223, + "num_tokens": 306588091.0, + "step": 12071 + }, + { + "epoch": 1.3257193059521195, + "grad_norm": 2.2281713485717773, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7230910658836365, + "num_tokens": 306614516.0, + "step": 12072 + }, + { + "epoch": 1.3258291236547333, + "grad_norm": 2.0480542182922363, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7130724191665649, + "num_tokens": 306644825.0, + "step": 12073 + }, + { + "epoch": 1.3259389413573468, + "grad_norm": 2.3217978477478027, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7454264163970947, + "num_tokens": 306666919.0, + "step": 12074 + }, + { + "epoch": 1.3260487590599603, + "grad_norm": 2.4291467666625977, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7103153467178345, + "num_tokens": 306689680.0, + "step": 12075 + }, + { + "epoch": 1.326158576762574, + "grad_norm": 2.0567619800567627, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7394791841506958, + "num_tokens": 306718621.0, + "step": 12076 + }, + { + "epoch": 1.3262683944651878, + "grad_norm": 2.1554694175720215, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.706397294998169, + "num_tokens": 306748242.0, + "step": 12077 + }, + { + "epoch": 1.3263782121678014, + "grad_norm": 2.064146041870117, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7277297973632812, + "num_tokens": 306777235.0, + "step": 12078 + }, + { + "epoch": 1.3264880298704151, + "grad_norm": 2.1964752674102783, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.706486165523529, + "num_tokens": 306805174.0, + "step": 12079 + }, + { + "epoch": 1.3265978475730287, + "grad_norm": 2.455875873565674, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7104606032371521, + "num_tokens": 306826883.0, + "step": 12080 + }, + { + "epoch": 1.3267076652756424, + "grad_norm": 2.337059736251831, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7289122939109802, + "num_tokens": 306851741.0, + "step": 12081 + }, + { + "epoch": 1.3268174829782562, + "grad_norm": 2.2242074012756348, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7433207035064697, + "num_tokens": 306876023.0, + "step": 12082 + }, + { + "epoch": 1.3269273006808697, + "grad_norm": 2.4051058292388916, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7253093719482422, + "num_tokens": 306898507.0, + "step": 12083 + }, + { + "epoch": 1.3270371183834835, + "grad_norm": 2.5165698528289795, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7086544036865234, + "num_tokens": 306921003.0, + "step": 12084 + }, + { + "epoch": 1.327146936086097, + "grad_norm": 2.1713919639587402, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7164599299430847, + "num_tokens": 306948466.0, + "step": 12085 + }, + { + "epoch": 1.3272567537887108, + "grad_norm": 2.2363781929016113, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7303553223609924, + "num_tokens": 306972646.0, + "step": 12086 + }, + { + "epoch": 1.3273665714913245, + "grad_norm": 2.340615749359131, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7429773211479187, + "num_tokens": 306993988.0, + "step": 12087 + }, + { + "epoch": 1.327476389193938, + "grad_norm": 2.2926185131073, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7205480933189392, + "num_tokens": 307020297.0, + "step": 12088 + }, + { + "epoch": 1.3275862068965516, + "grad_norm": 2.2497081756591797, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7211462259292603, + "num_tokens": 307045751.0, + "step": 12089 + }, + { + "epoch": 1.3276960245991654, + "grad_norm": 2.3635785579681396, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7293550372123718, + "num_tokens": 307067531.0, + "step": 12090 + }, + { + "epoch": 1.3278058423017791, + "grad_norm": 2.1456353664398193, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7232580780982971, + "num_tokens": 307094287.0, + "step": 12091 + }, + { + "epoch": 1.3279156600043927, + "grad_norm": 2.3963639736175537, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7272865176200867, + "num_tokens": 307117473.0, + "step": 12092 + }, + { + "epoch": 1.3280254777070064, + "grad_norm": 2.0823161602020264, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.724833607673645, + "num_tokens": 307147748.0, + "step": 12093 + }, + { + "epoch": 1.32813529540962, + "grad_norm": 2.1604769229888916, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7206322550773621, + "num_tokens": 307174002.0, + "step": 12094 + }, + { + "epoch": 1.3282451131122337, + "grad_norm": 2.2988784313201904, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7066890001296997, + "num_tokens": 307201343.0, + "step": 12095 + }, + { + "epoch": 1.3283549308148475, + "grad_norm": 2.2022545337677, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7101132869720459, + "num_tokens": 307229449.0, + "step": 12096 + }, + { + "epoch": 1.328464748517461, + "grad_norm": 2.1195619106292725, + "learning_rate": 1e-06, + "loss": 0.996, + "mean_token_accuracy": 0.689863920211792, + "num_tokens": 307259362.0, + "step": 12097 + }, + { + "epoch": 1.3285745662200747, + "grad_norm": 2.4512779712677, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7231597900390625, + "num_tokens": 307282105.0, + "step": 12098 + }, + { + "epoch": 1.3286843839226883, + "grad_norm": 2.3057498931884766, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7373926043510437, + "num_tokens": 307305926.0, + "step": 12099 + }, + { + "epoch": 1.328794201625302, + "grad_norm": 2.5795726776123047, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7198784351348877, + "num_tokens": 307325530.0, + "step": 12100 + }, + { + "epoch": 1.3289040193279158, + "grad_norm": 2.281921863555908, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7109034657478333, + "num_tokens": 307350269.0, + "step": 12101 + }, + { + "epoch": 1.3290138370305293, + "grad_norm": 2.2280972003936768, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7145237922668457, + "num_tokens": 307375252.0, + "step": 12102 + }, + { + "epoch": 1.3291236547331429, + "grad_norm": 2.018795967102051, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7123774886131287, + "num_tokens": 307405618.0, + "step": 12103 + }, + { + "epoch": 1.3292334724357566, + "grad_norm": 2.580587863922119, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7032639980316162, + "num_tokens": 307427415.0, + "step": 12104 + }, + { + "epoch": 1.3293432901383704, + "grad_norm": 2.203355550765991, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.724697470664978, + "num_tokens": 307454644.0, + "step": 12105 + }, + { + "epoch": 1.329453107840984, + "grad_norm": 2.331352472305298, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.716637134552002, + "num_tokens": 307478963.0, + "step": 12106 + }, + { + "epoch": 1.3295629255435977, + "grad_norm": 2.063608407974243, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.718159556388855, + "num_tokens": 307508480.0, + "step": 12107 + }, + { + "epoch": 1.3296727432462112, + "grad_norm": 2.4049665927886963, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7177897691726685, + "num_tokens": 307532392.0, + "step": 12108 + }, + { + "epoch": 1.329782560948825, + "grad_norm": 2.2656455039978027, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7161552906036377, + "num_tokens": 307557035.0, + "step": 12109 + }, + { + "epoch": 1.3298923786514387, + "grad_norm": 2.59340500831604, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7270780205726624, + "num_tokens": 307576148.0, + "step": 12110 + }, + { + "epoch": 1.3300021963540523, + "grad_norm": 2.5266623497009277, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7224367260932922, + "num_tokens": 307596930.0, + "step": 12111 + }, + { + "epoch": 1.330112014056666, + "grad_norm": 2.3335349559783936, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7262991666793823, + "num_tokens": 307620210.0, + "step": 12112 + }, + { + "epoch": 1.3302218317592795, + "grad_norm": 2.4155313968658447, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7124500870704651, + "num_tokens": 307644210.0, + "step": 12113 + }, + { + "epoch": 1.3303316494618933, + "grad_norm": 2.5239570140838623, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7006519436836243, + "num_tokens": 307667481.0, + "step": 12114 + }, + { + "epoch": 1.3304414671645068, + "grad_norm": 2.26961612701416, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7263500690460205, + "num_tokens": 307691536.0, + "step": 12115 + }, + { + "epoch": 1.3305512848671206, + "grad_norm": 2.375711679458618, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.70457524061203, + "num_tokens": 307717290.0, + "step": 12116 + }, + { + "epoch": 1.3306611025697341, + "grad_norm": 2.2275032997131348, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7022361159324646, + "num_tokens": 307742348.0, + "step": 12117 + }, + { + "epoch": 1.330770920272348, + "grad_norm": 2.1829817295074463, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7156622409820557, + "num_tokens": 307769808.0, + "step": 12118 + }, + { + "epoch": 1.3308807379749616, + "grad_norm": 2.5773181915283203, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7374042868614197, + "num_tokens": 307788794.0, + "step": 12119 + }, + { + "epoch": 1.3309905556775752, + "grad_norm": 2.361576795578003, + "learning_rate": 1e-06, + "loss": 0.9837, + "mean_token_accuracy": 0.6965217590332031, + "num_tokens": 307814356.0, + "step": 12120 + }, + { + "epoch": 1.331100373380189, + "grad_norm": 2.028024196624756, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7080169320106506, + "num_tokens": 307844495.0, + "step": 12121 + }, + { + "epoch": 1.3312101910828025, + "grad_norm": 2.3650686740875244, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7186582684516907, + "num_tokens": 307867305.0, + "step": 12122 + }, + { + "epoch": 1.3313200087854162, + "grad_norm": 2.2643911838531494, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7286913394927979, + "num_tokens": 307892149.0, + "step": 12123 + }, + { + "epoch": 1.33142982648803, + "grad_norm": 2.068714141845703, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7157727479934692, + "num_tokens": 307920266.0, + "step": 12124 + }, + { + "epoch": 1.3315396441906435, + "grad_norm": 2.160008192062378, + "learning_rate": 1e-06, + "loss": 0.971, + "mean_token_accuracy": 0.7013695240020752, + "num_tokens": 307949299.0, + "step": 12125 + }, + { + "epoch": 1.331649461893257, + "grad_norm": 2.50295090675354, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7201136350631714, + "num_tokens": 307970813.0, + "step": 12126 + }, + { + "epoch": 1.3317592795958708, + "grad_norm": 1.841117024421692, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7066857814788818, + "num_tokens": 308003902.0, + "step": 12127 + }, + { + "epoch": 1.3318690972984846, + "grad_norm": 1.9515092372894287, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7217506766319275, + "num_tokens": 308035349.0, + "step": 12128 + }, + { + "epoch": 1.331978915001098, + "grad_norm": 2.125932216644287, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7146562337875366, + "num_tokens": 308062744.0, + "step": 12129 + }, + { + "epoch": 1.3320887327037119, + "grad_norm": 2.2339935302734375, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7144743800163269, + "num_tokens": 308089915.0, + "step": 12130 + }, + { + "epoch": 1.3321985504063254, + "grad_norm": 2.1334943771362305, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7140793800354004, + "num_tokens": 308118961.0, + "step": 12131 + }, + { + "epoch": 1.3323083681089392, + "grad_norm": 1.950900673866272, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7112035751342773, + "num_tokens": 308152737.0, + "step": 12132 + }, + { + "epoch": 1.332418185811553, + "grad_norm": 2.2135794162750244, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7075859308242798, + "num_tokens": 308180269.0, + "step": 12133 + }, + { + "epoch": 1.3325280035141664, + "grad_norm": 2.064290761947632, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.704290509223938, + "num_tokens": 308211241.0, + "step": 12134 + }, + { + "epoch": 1.3326378212167802, + "grad_norm": 2.3607587814331055, + "learning_rate": 1e-06, + "loss": 1.0353, + "mean_token_accuracy": 0.6930521726608276, + "num_tokens": 308236492.0, + "step": 12135 + }, + { + "epoch": 1.3327476389193937, + "grad_norm": 2.119199514389038, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.707814633846283, + "num_tokens": 308266977.0, + "step": 12136 + }, + { + "epoch": 1.3328574566220075, + "grad_norm": 2.1741037368774414, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7135038375854492, + "num_tokens": 308295611.0, + "step": 12137 + }, + { + "epoch": 1.3329672743246213, + "grad_norm": 2.3970842361450195, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7327112555503845, + "num_tokens": 308318493.0, + "step": 12138 + }, + { + "epoch": 1.3330770920272348, + "grad_norm": 2.449646234512329, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7226783633232117, + "num_tokens": 308340874.0, + "step": 12139 + }, + { + "epoch": 1.3331869097298483, + "grad_norm": 1.9950807094573975, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7391417026519775, + "num_tokens": 308371600.0, + "step": 12140 + }, + { + "epoch": 1.333296727432462, + "grad_norm": 2.0196213722229004, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7106742262840271, + "num_tokens": 308401203.0, + "step": 12141 + }, + { + "epoch": 1.3334065451350758, + "grad_norm": 2.121986150741577, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7256247997283936, + "num_tokens": 308428981.0, + "step": 12142 + }, + { + "epoch": 1.3335163628376894, + "grad_norm": 2.425858497619629, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7031345963478088, + "num_tokens": 308451441.0, + "step": 12143 + }, + { + "epoch": 1.3336261805403031, + "grad_norm": 2.2072489261627197, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7475432753562927, + "num_tokens": 308476131.0, + "step": 12144 + }, + { + "epoch": 1.3337359982429167, + "grad_norm": 2.1867027282714844, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7229331731796265, + "num_tokens": 308503657.0, + "step": 12145 + }, + { + "epoch": 1.3338458159455304, + "grad_norm": 2.1835319995880127, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.7047338485717773, + "num_tokens": 308530953.0, + "step": 12146 + }, + { + "epoch": 1.3339556336481442, + "grad_norm": 2.1320035457611084, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7121790647506714, + "num_tokens": 308558148.0, + "step": 12147 + }, + { + "epoch": 1.3340654513507577, + "grad_norm": 2.422513961791992, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.736572265625, + "num_tokens": 308580627.0, + "step": 12148 + }, + { + "epoch": 1.3341752690533715, + "grad_norm": 2.077320098876953, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6905813813209534, + "num_tokens": 308608738.0, + "step": 12149 + }, + { + "epoch": 1.334285086755985, + "grad_norm": 2.109466552734375, + "learning_rate": 1e-06, + "loss": 0.9833, + "mean_token_accuracy": 0.6928954124450684, + "num_tokens": 308637512.0, + "step": 12150 + }, + { + "epoch": 1.3343949044585988, + "grad_norm": 2.2376041412353516, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7249172925949097, + "num_tokens": 308662386.0, + "step": 12151 + }, + { + "epoch": 1.3345047221612125, + "grad_norm": 2.062563896179199, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.7084361910820007, + "num_tokens": 308690163.0, + "step": 12152 + }, + { + "epoch": 1.334614539863826, + "grad_norm": 2.536288022994995, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7404792308807373, + "num_tokens": 308710220.0, + "step": 12153 + }, + { + "epoch": 1.3347243575664396, + "grad_norm": 2.407545328140259, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7171295285224915, + "num_tokens": 308735251.0, + "step": 12154 + }, + { + "epoch": 1.3348341752690533, + "grad_norm": 2.6648623943328857, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7370718121528625, + "num_tokens": 308753931.0, + "step": 12155 + }, + { + "epoch": 1.334943992971667, + "grad_norm": 2.567078113555908, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7120280265808105, + "num_tokens": 308774654.0, + "step": 12156 + }, + { + "epoch": 1.3350538106742806, + "grad_norm": 2.1620514392852783, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7358101606369019, + "num_tokens": 308800321.0, + "step": 12157 + }, + { + "epoch": 1.3351636283768944, + "grad_norm": 2.3494222164154053, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.730602502822876, + "num_tokens": 308822178.0, + "step": 12158 + }, + { + "epoch": 1.335273446079508, + "grad_norm": 2.2657880783081055, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.692441999912262, + "num_tokens": 308847222.0, + "step": 12159 + }, + { + "epoch": 1.3353832637821217, + "grad_norm": 2.4863035678863525, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7056622505187988, + "num_tokens": 308869072.0, + "step": 12160 + }, + { + "epoch": 1.3354930814847354, + "grad_norm": 2.231494665145874, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7203216552734375, + "num_tokens": 308894340.0, + "step": 12161 + }, + { + "epoch": 1.335602899187349, + "grad_norm": 2.140122413635254, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7081806659698486, + "num_tokens": 308924113.0, + "step": 12162 + }, + { + "epoch": 1.3357127168899627, + "grad_norm": 2.247485637664795, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7176909446716309, + "num_tokens": 308950341.0, + "step": 12163 + }, + { + "epoch": 1.3358225345925763, + "grad_norm": 2.244375705718994, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7125300168991089, + "num_tokens": 308976738.0, + "step": 12164 + }, + { + "epoch": 1.33593235229519, + "grad_norm": 2.25447940826416, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7174350619316101, + "num_tokens": 309002533.0, + "step": 12165 + }, + { + "epoch": 1.3360421699978036, + "grad_norm": 1.9782917499542236, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7187545895576477, + "num_tokens": 309034764.0, + "step": 12166 + }, + { + "epoch": 1.3361519877004173, + "grad_norm": 1.967159390449524, + "learning_rate": 1e-06, + "loss": 0.9943, + "mean_token_accuracy": 0.6954915523529053, + "num_tokens": 309066532.0, + "step": 12167 + }, + { + "epoch": 1.3362618054030309, + "grad_norm": 2.014691114425659, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7128605842590332, + "num_tokens": 309097676.0, + "step": 12168 + }, + { + "epoch": 1.3363716231056446, + "grad_norm": 2.565211057662964, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7282395958900452, + "num_tokens": 309119048.0, + "step": 12169 + }, + { + "epoch": 1.3364814408082584, + "grad_norm": 2.06044602394104, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.6906452178955078, + "num_tokens": 309150834.0, + "step": 12170 + }, + { + "epoch": 1.336591258510872, + "grad_norm": 1.990610957145691, + "learning_rate": 1e-06, + "loss": 0.9893, + "mean_token_accuracy": 0.6977248191833496, + "num_tokens": 309182826.0, + "step": 12171 + }, + { + "epoch": 1.3367010762134857, + "grad_norm": 2.744537353515625, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7264407873153687, + "num_tokens": 309202508.0, + "step": 12172 + }, + { + "epoch": 1.3368108939160992, + "grad_norm": 1.9709436893463135, + "learning_rate": 1e-06, + "loss": 0.979, + "mean_token_accuracy": 0.6976362466812134, + "num_tokens": 309235285.0, + "step": 12173 + }, + { + "epoch": 1.336920711618713, + "grad_norm": 2.091538667678833, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7123327255249023, + "num_tokens": 309264454.0, + "step": 12174 + }, + { + "epoch": 1.3370305293213267, + "grad_norm": 2.6176114082336426, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7321000695228577, + "num_tokens": 309283426.0, + "step": 12175 + }, + { + "epoch": 1.3371403470239402, + "grad_norm": 2.2681734561920166, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7255991101264954, + "num_tokens": 309307312.0, + "step": 12176 + }, + { + "epoch": 1.337250164726554, + "grad_norm": 2.039386749267578, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7374399900436401, + "num_tokens": 309332479.0, + "step": 12177 + }, + { + "epoch": 1.3373599824291675, + "grad_norm": 2.406789779663086, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7279448509216309, + "num_tokens": 309357050.0, + "step": 12178 + }, + { + "epoch": 1.3374698001317813, + "grad_norm": 2.2181432247161865, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7129062414169312, + "num_tokens": 309381418.0, + "step": 12179 + }, + { + "epoch": 1.3375796178343948, + "grad_norm": 1.9066303968429565, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7057361006736755, + "num_tokens": 309415257.0, + "step": 12180 + }, + { + "epoch": 1.3376894355370086, + "grad_norm": 2.069852828979492, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7332384586334229, + "num_tokens": 309442854.0, + "step": 12181 + }, + { + "epoch": 1.3377992532396221, + "grad_norm": 2.359260082244873, + "learning_rate": 1e-06, + "loss": 0.9531, + "mean_token_accuracy": 0.7072737812995911, + "num_tokens": 309466963.0, + "step": 12182 + }, + { + "epoch": 1.3379090709422359, + "grad_norm": 2.4367010593414307, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7330013513565063, + "num_tokens": 309489204.0, + "step": 12183 + }, + { + "epoch": 1.3380188886448496, + "grad_norm": 2.258596658706665, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7145428657531738, + "num_tokens": 309512359.0, + "step": 12184 + }, + { + "epoch": 1.3381287063474632, + "grad_norm": 2.139955520629883, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7080825567245483, + "num_tokens": 309540355.0, + "step": 12185 + }, + { + "epoch": 1.338238524050077, + "grad_norm": 2.5056581497192383, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7194514274597168, + "num_tokens": 309561246.0, + "step": 12186 + }, + { + "epoch": 1.3383483417526905, + "grad_norm": 2.216862440109253, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.729576587677002, + "num_tokens": 309587313.0, + "step": 12187 + }, + { + "epoch": 1.3384581594553042, + "grad_norm": 2.0459201335906982, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7233697772026062, + "num_tokens": 309614324.0, + "step": 12188 + }, + { + "epoch": 1.338567977157918, + "grad_norm": 2.2044930458068848, + "learning_rate": 1e-06, + "loss": 0.8235, + "mean_token_accuracy": 0.7545726299285889, + "num_tokens": 309639357.0, + "step": 12189 + }, + { + "epoch": 1.3386777948605315, + "grad_norm": 2.220313787460327, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7189368009567261, + "num_tokens": 309665103.0, + "step": 12190 + }, + { + "epoch": 1.338787612563145, + "grad_norm": 2.5538039207458496, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7199342250823975, + "num_tokens": 309685128.0, + "step": 12191 + }, + { + "epoch": 1.3388974302657588, + "grad_norm": 2.180849075317383, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7207099795341492, + "num_tokens": 309711753.0, + "step": 12192 + }, + { + "epoch": 1.3390072479683726, + "grad_norm": 1.9427053928375244, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.726415753364563, + "num_tokens": 309746398.0, + "step": 12193 + }, + { + "epoch": 1.339117065670986, + "grad_norm": 2.409080982208252, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7254579067230225, + "num_tokens": 309768627.0, + "step": 12194 + }, + { + "epoch": 1.3392268833735999, + "grad_norm": 2.2617545127868652, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7225293517112732, + "num_tokens": 309794432.0, + "step": 12195 + }, + { + "epoch": 1.3393367010762134, + "grad_norm": 2.360076904296875, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7063246369361877, + "num_tokens": 309818130.0, + "step": 12196 + }, + { + "epoch": 1.3394465187788271, + "grad_norm": 2.375807046890259, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7320458889007568, + "num_tokens": 309842310.0, + "step": 12197 + }, + { + "epoch": 1.339556336481441, + "grad_norm": 2.6775381565093994, + "learning_rate": 1e-06, + "loss": 0.8426, + "mean_token_accuracy": 0.7360600233078003, + "num_tokens": 309860586.0, + "step": 12198 + }, + { + "epoch": 1.3396661541840544, + "grad_norm": 1.9377880096435547, + "learning_rate": 1e-06, + "loss": 1.0194, + "mean_token_accuracy": 0.6876787543296814, + "num_tokens": 309893650.0, + "step": 12199 + }, + { + "epoch": 1.3397759718866682, + "grad_norm": 2.2777915000915527, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7117947936058044, + "num_tokens": 309918839.0, + "step": 12200 + }, + { + "epoch": 1.3398857895892817, + "grad_norm": 2.1268441677093506, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7156625986099243, + "num_tokens": 309947219.0, + "step": 12201 + }, + { + "epoch": 1.3399956072918955, + "grad_norm": 2.0583930015563965, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.707718014717102, + "num_tokens": 309979379.0, + "step": 12202 + }, + { + "epoch": 1.3401054249945092, + "grad_norm": 2.645911693572998, + "learning_rate": 1e-06, + "loss": 0.8008, + "mean_token_accuracy": 0.7543298006057739, + "num_tokens": 309999113.0, + "step": 12203 + }, + { + "epoch": 1.3402152426971228, + "grad_norm": 2.3219239711761475, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7192103862762451, + "num_tokens": 310024473.0, + "step": 12204 + }, + { + "epoch": 1.3403250603997363, + "grad_norm": 2.3105852603912354, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7137787342071533, + "num_tokens": 310049391.0, + "step": 12205 + }, + { + "epoch": 1.34043487810235, + "grad_norm": 2.3875646591186523, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7177397012710571, + "num_tokens": 310072286.0, + "step": 12206 + }, + { + "epoch": 1.3405446958049638, + "grad_norm": 2.2586185932159424, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.7081232070922852, + "num_tokens": 310098681.0, + "step": 12207 + }, + { + "epoch": 1.3406545135075774, + "grad_norm": 2.230639934539795, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7430277466773987, + "num_tokens": 310125615.0, + "step": 12208 + }, + { + "epoch": 1.3407643312101911, + "grad_norm": 2.3920440673828125, + "learning_rate": 1e-06, + "loss": 0.8027, + "mean_token_accuracy": 0.7496273517608643, + "num_tokens": 310148196.0, + "step": 12209 + }, + { + "epoch": 1.3408741489128047, + "grad_norm": 2.4536361694335938, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.72319495677948, + "num_tokens": 310170602.0, + "step": 12210 + }, + { + "epoch": 1.3409839666154184, + "grad_norm": 2.0498814582824707, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7156411409378052, + "num_tokens": 310198132.0, + "step": 12211 + }, + { + "epoch": 1.3410937843180322, + "grad_norm": 2.3844542503356934, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7310127019882202, + "num_tokens": 310221334.0, + "step": 12212 + }, + { + "epoch": 1.3412036020206457, + "grad_norm": 2.3302786350250244, + "learning_rate": 1e-06, + "loss": 0.8384, + "mean_token_accuracy": 0.7359845042228699, + "num_tokens": 310244341.0, + "step": 12213 + }, + { + "epoch": 1.3413134197232595, + "grad_norm": 2.3142778873443604, + "learning_rate": 1e-06, + "loss": 0.8275, + "mean_token_accuracy": 0.7473862767219543, + "num_tokens": 310268533.0, + "step": 12214 + }, + { + "epoch": 1.341423237425873, + "grad_norm": 1.9739731550216675, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.6978864669799805, + "num_tokens": 310298754.0, + "step": 12215 + }, + { + "epoch": 1.3415330551284868, + "grad_norm": 2.240560293197632, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7163279056549072, + "num_tokens": 310323549.0, + "step": 12216 + }, + { + "epoch": 1.3416428728311005, + "grad_norm": 2.0999608039855957, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.7045698165893555, + "num_tokens": 310352494.0, + "step": 12217 + }, + { + "epoch": 1.341752690533714, + "grad_norm": 2.816314935684204, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7214877605438232, + "num_tokens": 310369858.0, + "step": 12218 + }, + { + "epoch": 1.3418625082363276, + "grad_norm": 2.111704111099243, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7241248488426208, + "num_tokens": 310397337.0, + "step": 12219 + }, + { + "epoch": 1.3419723259389413, + "grad_norm": 2.3962037563323975, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7083002924919128, + "num_tokens": 310420489.0, + "step": 12220 + }, + { + "epoch": 1.342082143641555, + "grad_norm": 2.4451944828033447, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7180658578872681, + "num_tokens": 310443123.0, + "step": 12221 + }, + { + "epoch": 1.3421919613441686, + "grad_norm": 1.9300565719604492, + "learning_rate": 1e-06, + "loss": 1.0248, + "mean_token_accuracy": 0.6921834945678711, + "num_tokens": 310475933.0, + "step": 12222 + }, + { + "epoch": 1.3423017790467824, + "grad_norm": 2.242077112197876, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.743308424949646, + "num_tokens": 310500190.0, + "step": 12223 + }, + { + "epoch": 1.342411596749396, + "grad_norm": 2.1312053203582764, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7299081683158875, + "num_tokens": 310525528.0, + "step": 12224 + }, + { + "epoch": 1.3425214144520097, + "grad_norm": 2.2045023441314697, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7405765056610107, + "num_tokens": 310551106.0, + "step": 12225 + }, + { + "epoch": 1.3426312321546234, + "grad_norm": 2.5389277935028076, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7146669030189514, + "num_tokens": 310574069.0, + "step": 12226 + }, + { + "epoch": 1.342741049857237, + "grad_norm": 2.3040833473205566, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7168499827384949, + "num_tokens": 310599166.0, + "step": 12227 + }, + { + "epoch": 1.3428508675598507, + "grad_norm": 2.699993371963501, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7147473692893982, + "num_tokens": 310617388.0, + "step": 12228 + }, + { + "epoch": 1.3429606852624643, + "grad_norm": 2.2001864910125732, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7232201099395752, + "num_tokens": 310646628.0, + "step": 12229 + }, + { + "epoch": 1.343070502965078, + "grad_norm": 2.3622336387634277, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7238269448280334, + "num_tokens": 310669806.0, + "step": 12230 + }, + { + "epoch": 1.3431803206676916, + "grad_norm": 2.364431619644165, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7272086143493652, + "num_tokens": 310693199.0, + "step": 12231 + }, + { + "epoch": 1.3432901383703053, + "grad_norm": 2.390808582305908, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7327135801315308, + "num_tokens": 310715426.0, + "step": 12232 + }, + { + "epoch": 1.3433999560729188, + "grad_norm": 2.1062896251678467, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7120402455329895, + "num_tokens": 310745454.0, + "step": 12233 + }, + { + "epoch": 1.3435097737755326, + "grad_norm": 2.0567996501922607, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7016865015029907, + "num_tokens": 310773756.0, + "step": 12234 + }, + { + "epoch": 1.3436195914781464, + "grad_norm": 1.9742696285247803, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7125779986381531, + "num_tokens": 310804195.0, + "step": 12235 + }, + { + "epoch": 1.34372940918076, + "grad_norm": 2.2146363258361816, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7112054824829102, + "num_tokens": 310830536.0, + "step": 12236 + }, + { + "epoch": 1.3438392268833736, + "grad_norm": 2.3636605739593506, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7296500205993652, + "num_tokens": 310853829.0, + "step": 12237 + }, + { + "epoch": 1.3439490445859872, + "grad_norm": 2.2727246284484863, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.6973042488098145, + "num_tokens": 310881487.0, + "step": 12238 + }, + { + "epoch": 1.344058862288601, + "grad_norm": 1.9400280714035034, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7110412120819092, + "num_tokens": 310915378.0, + "step": 12239 + }, + { + "epoch": 1.3441686799912147, + "grad_norm": 2.1615610122680664, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.718093752861023, + "num_tokens": 310943218.0, + "step": 12240 + }, + { + "epoch": 1.3442784976938282, + "grad_norm": 2.1936826705932617, + "learning_rate": 1e-06, + "loss": 1.0126, + "mean_token_accuracy": 0.6900357007980347, + "num_tokens": 310970175.0, + "step": 12241 + }, + { + "epoch": 1.3443883153964418, + "grad_norm": 2.2590603828430176, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7092961072921753, + "num_tokens": 310994766.0, + "step": 12242 + }, + { + "epoch": 1.3444981330990555, + "grad_norm": 2.6243693828582764, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7486695647239685, + "num_tokens": 311013566.0, + "step": 12243 + }, + { + "epoch": 1.3446079508016693, + "grad_norm": 1.9877387285232544, + "learning_rate": 1e-06, + "loss": 0.9867, + "mean_token_accuracy": 0.7011427879333496, + "num_tokens": 311046537.0, + "step": 12244 + }, + { + "epoch": 1.3447177685042828, + "grad_norm": 2.2216715812683105, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7149266004562378, + "num_tokens": 311073112.0, + "step": 12245 + }, + { + "epoch": 1.3448275862068966, + "grad_norm": 2.4304182529449463, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7375473976135254, + "num_tokens": 311093771.0, + "step": 12246 + }, + { + "epoch": 1.34493740390951, + "grad_norm": 2.3584585189819336, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.7098488807678223, + "num_tokens": 311118309.0, + "step": 12247 + }, + { + "epoch": 1.3450472216121239, + "grad_norm": 2.341620683670044, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7000594139099121, + "num_tokens": 311144034.0, + "step": 12248 + }, + { + "epoch": 1.3451570393147376, + "grad_norm": 2.1052327156066895, + "learning_rate": 1e-06, + "loss": 1.0349, + "mean_token_accuracy": 0.6865785717964172, + "num_tokens": 311174700.0, + "step": 12249 + }, + { + "epoch": 1.3452668570173512, + "grad_norm": 2.1257011890411377, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.704149067401886, + "num_tokens": 311205655.0, + "step": 12250 + }, + { + "epoch": 1.345376674719965, + "grad_norm": 2.64897084236145, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7372870445251465, + "num_tokens": 311226424.0, + "step": 12251 + }, + { + "epoch": 1.3454864924225785, + "grad_norm": 2.373689889907837, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7179803252220154, + "num_tokens": 311251245.0, + "step": 12252 + }, + { + "epoch": 1.3455963101251922, + "grad_norm": 2.145174026489258, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7218987941741943, + "num_tokens": 311277913.0, + "step": 12253 + }, + { + "epoch": 1.345706127827806, + "grad_norm": 2.3809092044830322, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7135158181190491, + "num_tokens": 311300091.0, + "step": 12254 + }, + { + "epoch": 1.3458159455304195, + "grad_norm": 2.4367592334747314, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7438669800758362, + "num_tokens": 311321029.0, + "step": 12255 + }, + { + "epoch": 1.345925763233033, + "grad_norm": 2.357279062271118, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7302766442298889, + "num_tokens": 311344278.0, + "step": 12256 + }, + { + "epoch": 1.3460355809356468, + "grad_norm": 2.34458589553833, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7159114480018616, + "num_tokens": 311367791.0, + "step": 12257 + }, + { + "epoch": 1.3461453986382605, + "grad_norm": 2.227762222290039, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7144423723220825, + "num_tokens": 311396443.0, + "step": 12258 + }, + { + "epoch": 1.346255216340874, + "grad_norm": 2.432312250137329, + "learning_rate": 1e-06, + "loss": 0.8121, + "mean_token_accuracy": 0.7436118125915527, + "num_tokens": 311417587.0, + "step": 12259 + }, + { + "epoch": 1.3463650340434878, + "grad_norm": 2.342914581298828, + "learning_rate": 1e-06, + "loss": 0.8133, + "mean_token_accuracy": 0.7384777069091797, + "num_tokens": 311441395.0, + "step": 12260 + }, + { + "epoch": 1.3464748517461014, + "grad_norm": 2.6218740940093994, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7281357049942017, + "num_tokens": 311462199.0, + "step": 12261 + }, + { + "epoch": 1.3465846694487151, + "grad_norm": 2.2762887477874756, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7215261459350586, + "num_tokens": 311487240.0, + "step": 12262 + }, + { + "epoch": 1.346694487151329, + "grad_norm": 2.031564950942993, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7074001431465149, + "num_tokens": 311517280.0, + "step": 12263 + }, + { + "epoch": 1.3468043048539424, + "grad_norm": 2.257016181945801, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7019668817520142, + "num_tokens": 311545405.0, + "step": 12264 + }, + { + "epoch": 1.3469141225565562, + "grad_norm": 2.144742727279663, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7272620797157288, + "num_tokens": 311573801.0, + "step": 12265 + }, + { + "epoch": 1.3470239402591697, + "grad_norm": 2.3234407901763916, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7163355350494385, + "num_tokens": 311597028.0, + "step": 12266 + }, + { + "epoch": 1.3471337579617835, + "grad_norm": 2.0602338314056396, + "learning_rate": 1e-06, + "loss": 0.998, + "mean_token_accuracy": 0.6939393877983093, + "num_tokens": 311626642.0, + "step": 12267 + }, + { + "epoch": 1.3472435756643972, + "grad_norm": 2.1883833408355713, + "learning_rate": 1e-06, + "loss": 1.0117, + "mean_token_accuracy": 0.6877624988555908, + "num_tokens": 311654497.0, + "step": 12268 + }, + { + "epoch": 1.3473533933670108, + "grad_norm": 2.528104782104492, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.6965242624282837, + "num_tokens": 311675126.0, + "step": 12269 + }, + { + "epoch": 1.3474632110696243, + "grad_norm": 2.212031364440918, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7011873722076416, + "num_tokens": 311701344.0, + "step": 12270 + }, + { + "epoch": 1.347573028772238, + "grad_norm": 1.962985873222351, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7346298098564148, + "num_tokens": 311733229.0, + "step": 12271 + }, + { + "epoch": 1.3476828464748518, + "grad_norm": 2.3402867317199707, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7197372913360596, + "num_tokens": 311758842.0, + "step": 12272 + }, + { + "epoch": 1.3477926641774653, + "grad_norm": 2.250455617904663, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.6992464065551758, + "num_tokens": 311784614.0, + "step": 12273 + }, + { + "epoch": 1.347902481880079, + "grad_norm": 2.3369510173797607, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7036452293395996, + "num_tokens": 311808657.0, + "step": 12274 + }, + { + "epoch": 1.3480122995826926, + "grad_norm": 2.093712568283081, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.708177924156189, + "num_tokens": 311838926.0, + "step": 12275 + }, + { + "epoch": 1.3481221172853064, + "grad_norm": 2.0017449855804443, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7289349436759949, + "num_tokens": 311867780.0, + "step": 12276 + }, + { + "epoch": 1.3482319349879202, + "grad_norm": 2.2924654483795166, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7210167646408081, + "num_tokens": 311892323.0, + "step": 12277 + }, + { + "epoch": 1.3483417526905337, + "grad_norm": 2.1716136932373047, + "learning_rate": 1e-06, + "loss": 0.9859, + "mean_token_accuracy": 0.7016975283622742, + "num_tokens": 311922236.0, + "step": 12278 + }, + { + "epoch": 1.3484515703931474, + "grad_norm": 1.978481650352478, + "learning_rate": 1e-06, + "loss": 1.0326, + "mean_token_accuracy": 0.6930117011070251, + "num_tokens": 311958471.0, + "step": 12279 + }, + { + "epoch": 1.348561388095761, + "grad_norm": 2.059800148010254, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.6982663869857788, + "num_tokens": 311989309.0, + "step": 12280 + }, + { + "epoch": 1.3486712057983747, + "grad_norm": 2.4698686599731445, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.726866602897644, + "num_tokens": 312009972.0, + "step": 12281 + }, + { + "epoch": 1.3487810235009885, + "grad_norm": 2.1968982219696045, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7196463942527771, + "num_tokens": 312035239.0, + "step": 12282 + }, + { + "epoch": 1.348890841203602, + "grad_norm": 2.0595102310180664, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7370287179946899, + "num_tokens": 312064177.0, + "step": 12283 + }, + { + "epoch": 1.3490006589062156, + "grad_norm": 2.3058040142059326, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7213197946548462, + "num_tokens": 312087390.0, + "step": 12284 + }, + { + "epoch": 1.3491104766088293, + "grad_norm": 2.107651710510254, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7034817934036255, + "num_tokens": 312115698.0, + "step": 12285 + }, + { + "epoch": 1.349220294311443, + "grad_norm": 2.3608362674713135, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7040966749191284, + "num_tokens": 312140778.0, + "step": 12286 + }, + { + "epoch": 1.3493301120140566, + "grad_norm": 2.407759428024292, + "learning_rate": 1e-06, + "loss": 0.8285, + "mean_token_accuracy": 0.7353762984275818, + "num_tokens": 312162915.0, + "step": 12287 + }, + { + "epoch": 1.3494399297166704, + "grad_norm": 2.431746006011963, + "learning_rate": 1e-06, + "loss": 0.8135, + "mean_token_accuracy": 0.739570140838623, + "num_tokens": 312183774.0, + "step": 12288 + }, + { + "epoch": 1.349549747419284, + "grad_norm": 2.22698974609375, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7382739186286926, + "num_tokens": 312207976.0, + "step": 12289 + }, + { + "epoch": 1.3496595651218977, + "grad_norm": 2.214097023010254, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7087911367416382, + "num_tokens": 312233840.0, + "step": 12290 + }, + { + "epoch": 1.3497693828245114, + "grad_norm": 2.3449089527130127, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7215560674667358, + "num_tokens": 312257610.0, + "step": 12291 + }, + { + "epoch": 1.349879200527125, + "grad_norm": 2.0876197814941406, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7229195833206177, + "num_tokens": 312285684.0, + "step": 12292 + }, + { + "epoch": 1.3499890182297387, + "grad_norm": 2.408447265625, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7122593522071838, + "num_tokens": 312309377.0, + "step": 12293 + }, + { + "epoch": 1.3500988359323522, + "grad_norm": 2.42586088180542, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7125140428543091, + "num_tokens": 312331382.0, + "step": 12294 + }, + { + "epoch": 1.350208653634966, + "grad_norm": 2.401017189025879, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.717898964881897, + "num_tokens": 312352666.0, + "step": 12295 + }, + { + "epoch": 1.3503184713375795, + "grad_norm": 2.242558479309082, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7302322387695312, + "num_tokens": 312379316.0, + "step": 12296 + }, + { + "epoch": 1.3504282890401933, + "grad_norm": 2.247758388519287, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7183953523635864, + "num_tokens": 312407357.0, + "step": 12297 + }, + { + "epoch": 1.3505381067428068, + "grad_norm": 1.9619839191436768, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7227376103401184, + "num_tokens": 312439943.0, + "step": 12298 + }, + { + "epoch": 1.3506479244454206, + "grad_norm": 2.1382758617401123, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7055398225784302, + "num_tokens": 312466136.0, + "step": 12299 + }, + { + "epoch": 1.3507577421480343, + "grad_norm": 2.2777493000030518, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7321794033050537, + "num_tokens": 312490225.0, + "step": 12300 + }, + { + "epoch": 1.3508675598506479, + "grad_norm": 2.117020606994629, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.6915327310562134, + "num_tokens": 312518815.0, + "step": 12301 + }, + { + "epoch": 1.3509773775532616, + "grad_norm": 2.563607931137085, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.724510908126831, + "num_tokens": 312538406.0, + "step": 12302 + }, + { + "epoch": 1.3510871952558752, + "grad_norm": 2.158085823059082, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7189537286758423, + "num_tokens": 312565605.0, + "step": 12303 + }, + { + "epoch": 1.351197012958489, + "grad_norm": 2.3338708877563477, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.71002197265625, + "num_tokens": 312590522.0, + "step": 12304 + }, + { + "epoch": 1.3513068306611027, + "grad_norm": 2.380962610244751, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.6990915536880493, + "num_tokens": 312614271.0, + "step": 12305 + }, + { + "epoch": 1.3514166483637162, + "grad_norm": 2.8186464309692383, + "learning_rate": 1e-06, + "loss": 0.7576, + "mean_token_accuracy": 0.753400981426239, + "num_tokens": 312631180.0, + "step": 12306 + }, + { + "epoch": 1.3515264660663298, + "grad_norm": 2.1595029830932617, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.687357485294342, + "num_tokens": 312657929.0, + "step": 12307 + }, + { + "epoch": 1.3516362837689435, + "grad_norm": 2.524038314819336, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7336570620536804, + "num_tokens": 312678102.0, + "step": 12308 + }, + { + "epoch": 1.3517461014715573, + "grad_norm": 2.2259089946746826, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.711178719997406, + "num_tokens": 312704465.0, + "step": 12309 + }, + { + "epoch": 1.3518559191741708, + "grad_norm": 2.5043365955352783, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7092733979225159, + "num_tokens": 312726187.0, + "step": 12310 + }, + { + "epoch": 1.3519657368767846, + "grad_norm": 2.3305716514587402, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7249301671981812, + "num_tokens": 312750858.0, + "step": 12311 + }, + { + "epoch": 1.352075554579398, + "grad_norm": 2.2927606105804443, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7220710515975952, + "num_tokens": 312774614.0, + "step": 12312 + }, + { + "epoch": 1.3521853722820119, + "grad_norm": 2.2215416431427, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7181119918823242, + "num_tokens": 312799915.0, + "step": 12313 + }, + { + "epoch": 1.3522951899846256, + "grad_norm": 2.078230381011963, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7298873662948608, + "num_tokens": 312828312.0, + "step": 12314 + }, + { + "epoch": 1.3524050076872391, + "grad_norm": 2.5377330780029297, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6959318518638611, + "num_tokens": 312853640.0, + "step": 12315 + }, + { + "epoch": 1.352514825389853, + "grad_norm": 2.256396770477295, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7360138297080994, + "num_tokens": 312881246.0, + "step": 12316 + }, + { + "epoch": 1.3526246430924664, + "grad_norm": 2.257624864578247, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7215350866317749, + "num_tokens": 312907288.0, + "step": 12317 + }, + { + "epoch": 1.3527344607950802, + "grad_norm": 2.3349955081939697, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7207044363021851, + "num_tokens": 312932248.0, + "step": 12318 + }, + { + "epoch": 1.352844278497694, + "grad_norm": 2.1956965923309326, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7108907699584961, + "num_tokens": 312957744.0, + "step": 12319 + }, + { + "epoch": 1.3529540962003075, + "grad_norm": 2.1574442386627197, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.706057071685791, + "num_tokens": 312985518.0, + "step": 12320 + }, + { + "epoch": 1.353063913902921, + "grad_norm": 2.298332452774048, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7236745357513428, + "num_tokens": 313009443.0, + "step": 12321 + }, + { + "epoch": 1.3531737316055348, + "grad_norm": 2.0693085193634033, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7218277454376221, + "num_tokens": 313039552.0, + "step": 12322 + }, + { + "epoch": 1.3532835493081485, + "grad_norm": 2.369666814804077, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7377889156341553, + "num_tokens": 313062337.0, + "step": 12323 + }, + { + "epoch": 1.353393367010762, + "grad_norm": 2.3211474418640137, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7306088209152222, + "num_tokens": 313085291.0, + "step": 12324 + }, + { + "epoch": 1.3535031847133758, + "grad_norm": 2.207186460494995, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7112017273902893, + "num_tokens": 313110398.0, + "step": 12325 + }, + { + "epoch": 1.3536130024159894, + "grad_norm": 2.493516206741333, + "learning_rate": 1e-06, + "loss": 0.8786, + "mean_token_accuracy": 0.7228540778160095, + "num_tokens": 313133470.0, + "step": 12326 + }, + { + "epoch": 1.3537228201186031, + "grad_norm": 2.2559728622436523, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7256021499633789, + "num_tokens": 313157484.0, + "step": 12327 + }, + { + "epoch": 1.3538326378212169, + "grad_norm": 2.307478904724121, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7153060436248779, + "num_tokens": 313181709.0, + "step": 12328 + }, + { + "epoch": 1.3539424555238304, + "grad_norm": 2.22933292388916, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7212875485420227, + "num_tokens": 313207848.0, + "step": 12329 + }, + { + "epoch": 1.3540522732264442, + "grad_norm": 2.216151714324951, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7299891114234924, + "num_tokens": 313233373.0, + "step": 12330 + }, + { + "epoch": 1.3541620909290577, + "grad_norm": 2.159966230392456, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7148922681808472, + "num_tokens": 313261478.0, + "step": 12331 + }, + { + "epoch": 1.3542719086316715, + "grad_norm": 2.381782054901123, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7191750407218933, + "num_tokens": 313285495.0, + "step": 12332 + }, + { + "epoch": 1.3543817263342852, + "grad_norm": 2.3651223182678223, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7381600141525269, + "num_tokens": 313307428.0, + "step": 12333 + }, + { + "epoch": 1.3544915440368988, + "grad_norm": 2.2059948444366455, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7188870310783386, + "num_tokens": 313336066.0, + "step": 12334 + }, + { + "epoch": 1.3546013617395123, + "grad_norm": 2.083225965499878, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.6983448266983032, + "num_tokens": 313366303.0, + "step": 12335 + }, + { + "epoch": 1.354711179442126, + "grad_norm": 2.771144151687622, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7345278263092041, + "num_tokens": 313384183.0, + "step": 12336 + }, + { + "epoch": 1.3548209971447398, + "grad_norm": 2.3006374835968018, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7092710137367249, + "num_tokens": 313410612.0, + "step": 12337 + }, + { + "epoch": 1.3549308148473533, + "grad_norm": 2.5398828983306885, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.719918429851532, + "num_tokens": 313431053.0, + "step": 12338 + }, + { + "epoch": 1.355040632549967, + "grad_norm": 2.105100154876709, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7265661954879761, + "num_tokens": 313460391.0, + "step": 12339 + }, + { + "epoch": 1.3551504502525806, + "grad_norm": 2.3951778411865234, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7127717137336731, + "num_tokens": 313485077.0, + "step": 12340 + }, + { + "epoch": 1.3552602679551944, + "grad_norm": 2.1578145027160645, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7023587226867676, + "num_tokens": 313511956.0, + "step": 12341 + }, + { + "epoch": 1.3553700856578081, + "grad_norm": 2.495192766189575, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.73476243019104, + "num_tokens": 313531354.0, + "step": 12342 + }, + { + "epoch": 1.3554799033604217, + "grad_norm": 2.206895351409912, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7328026294708252, + "num_tokens": 313557442.0, + "step": 12343 + }, + { + "epoch": 1.3555897210630354, + "grad_norm": 2.072277784347534, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7153351306915283, + "num_tokens": 313587506.0, + "step": 12344 + }, + { + "epoch": 1.355699538765649, + "grad_norm": 2.4137182235717773, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.732358455657959, + "num_tokens": 313609584.0, + "step": 12345 + }, + { + "epoch": 1.3558093564682627, + "grad_norm": 2.3991494178771973, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7239499092102051, + "num_tokens": 313632012.0, + "step": 12346 + }, + { + "epoch": 1.3559191741708763, + "grad_norm": 2.5805888175964355, + "learning_rate": 1e-06, + "loss": 0.8076, + "mean_token_accuracy": 0.7425934672355652, + "num_tokens": 313650692.0, + "step": 12347 + }, + { + "epoch": 1.35602899187349, + "grad_norm": 2.306220531463623, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.725462794303894, + "num_tokens": 313674517.0, + "step": 12348 + }, + { + "epoch": 1.3561388095761036, + "grad_norm": 2.247023344039917, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.716034471988678, + "num_tokens": 313699713.0, + "step": 12349 + }, + { + "epoch": 1.3562486272787173, + "grad_norm": 2.250797986984253, + "learning_rate": 1e-06, + "loss": 0.8605, + "mean_token_accuracy": 0.7330806255340576, + "num_tokens": 313726247.0, + "step": 12350 + }, + { + "epoch": 1.356358444981331, + "grad_norm": 2.1350059509277344, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7133865356445312, + "num_tokens": 313753574.0, + "step": 12351 + }, + { + "epoch": 1.3564682626839446, + "grad_norm": 2.397554874420166, + "learning_rate": 1e-06, + "loss": 0.7463, + "mean_token_accuracy": 0.7666419148445129, + "num_tokens": 313774080.0, + "step": 12352 + }, + { + "epoch": 1.3565780803865584, + "grad_norm": 2.092820167541504, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.6958034038543701, + "num_tokens": 313803363.0, + "step": 12353 + }, + { + "epoch": 1.356687898089172, + "grad_norm": 2.083725929260254, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7175679802894592, + "num_tokens": 313832102.0, + "step": 12354 + }, + { + "epoch": 1.3567977157917857, + "grad_norm": 2.189011812210083, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7315773963928223, + "num_tokens": 313858621.0, + "step": 12355 + }, + { + "epoch": 1.3569075334943994, + "grad_norm": 2.55059552192688, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7195068597793579, + "num_tokens": 313878849.0, + "step": 12356 + }, + { + "epoch": 1.357017351197013, + "grad_norm": 2.4479804039001465, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7227667570114136, + "num_tokens": 313902692.0, + "step": 12357 + }, + { + "epoch": 1.3571271688996267, + "grad_norm": 2.145967483520508, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.728868842124939, + "num_tokens": 313930756.0, + "step": 12358 + }, + { + "epoch": 1.3572369866022402, + "grad_norm": 2.584172248840332, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7367813587188721, + "num_tokens": 313951616.0, + "step": 12359 + }, + { + "epoch": 1.357346804304854, + "grad_norm": 2.068857431411743, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.6929728388786316, + "num_tokens": 313979683.0, + "step": 12360 + }, + { + "epoch": 1.3574566220074675, + "grad_norm": 2.345113754272461, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7172220349311829, + "num_tokens": 314003378.0, + "step": 12361 + }, + { + "epoch": 1.3575664397100813, + "grad_norm": 2.3895649909973145, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.7039463520050049, + "num_tokens": 314029008.0, + "step": 12362 + }, + { + "epoch": 1.3576762574126948, + "grad_norm": 2.425590753555298, + "learning_rate": 1e-06, + "loss": 0.8068, + "mean_token_accuracy": 0.7464656829833984, + "num_tokens": 314050761.0, + "step": 12363 + }, + { + "epoch": 1.3577860751153086, + "grad_norm": 2.3542282581329346, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.722655177116394, + "num_tokens": 314073501.0, + "step": 12364 + }, + { + "epoch": 1.3578958928179223, + "grad_norm": 2.192633628845215, + "learning_rate": 1e-06, + "loss": 0.7797, + "mean_token_accuracy": 0.7531712055206299, + "num_tokens": 314098617.0, + "step": 12365 + }, + { + "epoch": 1.3580057105205359, + "grad_norm": 2.60728120803833, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7081842422485352, + "num_tokens": 314119212.0, + "step": 12366 + }, + { + "epoch": 1.3581155282231496, + "grad_norm": 2.2663841247558594, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7209759950637817, + "num_tokens": 314146170.0, + "step": 12367 + }, + { + "epoch": 1.3582253459257632, + "grad_norm": 2.463681936264038, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.744046688079834, + "num_tokens": 314166584.0, + "step": 12368 + }, + { + "epoch": 1.358335163628377, + "grad_norm": 2.6091604232788086, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7066492438316345, + "num_tokens": 314188439.0, + "step": 12369 + }, + { + "epoch": 1.3584449813309907, + "grad_norm": 1.9534893035888672, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.691259503364563, + "num_tokens": 314222524.0, + "step": 12370 + }, + { + "epoch": 1.3585547990336042, + "grad_norm": 2.1079049110412598, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.714256227016449, + "num_tokens": 314251360.0, + "step": 12371 + }, + { + "epoch": 1.3586646167362177, + "grad_norm": 2.7764625549316406, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7161617279052734, + "num_tokens": 314271498.0, + "step": 12372 + }, + { + "epoch": 1.3587744344388315, + "grad_norm": 2.2575740814208984, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.6960370540618896, + "num_tokens": 314297741.0, + "step": 12373 + }, + { + "epoch": 1.3588842521414453, + "grad_norm": 2.170535087585449, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7003825902938843, + "num_tokens": 314322461.0, + "step": 12374 + }, + { + "epoch": 1.3589940698440588, + "grad_norm": 2.3246541023254395, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.7356852889060974, + "num_tokens": 314345661.0, + "step": 12375 + }, + { + "epoch": 1.3591038875466726, + "grad_norm": 2.2211880683898926, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7291440963745117, + "num_tokens": 314371957.0, + "step": 12376 + }, + { + "epoch": 1.359213705249286, + "grad_norm": 2.237210988998413, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7274087071418762, + "num_tokens": 314395435.0, + "step": 12377 + }, + { + "epoch": 1.3593235229518998, + "grad_norm": 2.431713581085205, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7290003895759583, + "num_tokens": 314418950.0, + "step": 12378 + }, + { + "epoch": 1.3594333406545136, + "grad_norm": 2.3115570545196533, + "learning_rate": 1e-06, + "loss": 0.8087, + "mean_token_accuracy": 0.7408193945884705, + "num_tokens": 314441714.0, + "step": 12379 + }, + { + "epoch": 1.3595431583571271, + "grad_norm": 2.5333542823791504, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7157223224639893, + "num_tokens": 314463069.0, + "step": 12380 + }, + { + "epoch": 1.359652976059741, + "grad_norm": 2.1519031524658203, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7241081595420837, + "num_tokens": 314490756.0, + "step": 12381 + }, + { + "epoch": 1.3597627937623544, + "grad_norm": 2.6219289302825928, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7170478105545044, + "num_tokens": 314510368.0, + "step": 12382 + }, + { + "epoch": 1.3598726114649682, + "grad_norm": 2.5427358150482178, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.700411319732666, + "num_tokens": 314534100.0, + "step": 12383 + }, + { + "epoch": 1.359982429167582, + "grad_norm": 2.4382500648498535, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7099279165267944, + "num_tokens": 314557605.0, + "step": 12384 + }, + { + "epoch": 1.3600922468701955, + "grad_norm": 2.37974214553833, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7059569358825684, + "num_tokens": 314582578.0, + "step": 12385 + }, + { + "epoch": 1.360202064572809, + "grad_norm": 2.1358532905578613, + "learning_rate": 1e-06, + "loss": 1.0273, + "mean_token_accuracy": 0.6929267644882202, + "num_tokens": 314613101.0, + "step": 12386 + }, + { + "epoch": 1.3603118822754228, + "grad_norm": 2.3482627868652344, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7312995791435242, + "num_tokens": 314636688.0, + "step": 12387 + }, + { + "epoch": 1.3604216999780365, + "grad_norm": 2.1012728214263916, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7109623551368713, + "num_tokens": 314666295.0, + "step": 12388 + }, + { + "epoch": 1.36053151768065, + "grad_norm": 2.345935344696045, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7375918626785278, + "num_tokens": 314689246.0, + "step": 12389 + }, + { + "epoch": 1.3606413353832638, + "grad_norm": 2.004101514816284, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7449486255645752, + "num_tokens": 314720470.0, + "step": 12390 + }, + { + "epoch": 1.3607511530858774, + "grad_norm": 2.3464696407318115, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7169138193130493, + "num_tokens": 314742878.0, + "step": 12391 + }, + { + "epoch": 1.360860970788491, + "grad_norm": 2.1893131732940674, + "learning_rate": 1e-06, + "loss": 1.0103, + "mean_token_accuracy": 0.696174144744873, + "num_tokens": 314769272.0, + "step": 12392 + }, + { + "epoch": 1.3609707884911049, + "grad_norm": 2.2385075092315674, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7037014365196228, + "num_tokens": 314795818.0, + "step": 12393 + }, + { + "epoch": 1.3610806061937184, + "grad_norm": 2.467305898666382, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7290619611740112, + "num_tokens": 314817861.0, + "step": 12394 + }, + { + "epoch": 1.3611904238963322, + "grad_norm": 2.2309534549713135, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7410017251968384, + "num_tokens": 314842927.0, + "step": 12395 + }, + { + "epoch": 1.3613002415989457, + "grad_norm": 2.3704967498779297, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.721405029296875, + "num_tokens": 314866861.0, + "step": 12396 + }, + { + "epoch": 1.3614100593015594, + "grad_norm": 2.135627031326294, + "learning_rate": 1e-06, + "loss": 0.9951, + "mean_token_accuracy": 0.6966555714607239, + "num_tokens": 314895654.0, + "step": 12397 + }, + { + "epoch": 1.3615198770041732, + "grad_norm": 2.0496647357940674, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7063160538673401, + "num_tokens": 314926276.0, + "step": 12398 + }, + { + "epoch": 1.3616296947067867, + "grad_norm": 2.1977508068084717, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6946302652359009, + "num_tokens": 314955839.0, + "step": 12399 + }, + { + "epoch": 1.3617395124094003, + "grad_norm": 2.1739747524261475, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.71384596824646, + "num_tokens": 314983408.0, + "step": 12400 + }, + { + "epoch": 1.361849330112014, + "grad_norm": 2.4803903102874756, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7078668475151062, + "num_tokens": 315005004.0, + "step": 12401 + }, + { + "epoch": 1.3619591478146278, + "grad_norm": 2.2699689865112305, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7256698608398438, + "num_tokens": 315028313.0, + "step": 12402 + }, + { + "epoch": 1.3620689655172413, + "grad_norm": 2.4447131156921387, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7046443223953247, + "num_tokens": 315050980.0, + "step": 12403 + }, + { + "epoch": 1.362178783219855, + "grad_norm": 2.249103307723999, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7203188538551331, + "num_tokens": 315077998.0, + "step": 12404 + }, + { + "epoch": 1.3622886009224686, + "grad_norm": 2.206617593765259, + "learning_rate": 1e-06, + "loss": 0.9996, + "mean_token_accuracy": 0.6954973936080933, + "num_tokens": 315106646.0, + "step": 12405 + }, + { + "epoch": 1.3623984186250824, + "grad_norm": 2.232314348220825, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7420648336410522, + "num_tokens": 315131167.0, + "step": 12406 + }, + { + "epoch": 1.3625082363276961, + "grad_norm": 2.4470512866973877, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7177286148071289, + "num_tokens": 315153477.0, + "step": 12407 + }, + { + "epoch": 1.3626180540303097, + "grad_norm": 2.4787213802337646, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7192432880401611, + "num_tokens": 315176253.0, + "step": 12408 + }, + { + "epoch": 1.3627278717329234, + "grad_norm": 2.1453750133514404, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7269089221954346, + "num_tokens": 315203579.0, + "step": 12409 + }, + { + "epoch": 1.362837689435537, + "grad_norm": 2.2914810180664062, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7232801914215088, + "num_tokens": 315231552.0, + "step": 12410 + }, + { + "epoch": 1.3629475071381507, + "grad_norm": 2.372192144393921, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7306851148605347, + "num_tokens": 315254789.0, + "step": 12411 + }, + { + "epoch": 1.3630573248407643, + "grad_norm": 2.1095519065856934, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7088750600814819, + "num_tokens": 315281725.0, + "step": 12412 + }, + { + "epoch": 1.363167142543378, + "grad_norm": 2.2487056255340576, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7287048101425171, + "num_tokens": 315305465.0, + "step": 12413 + }, + { + "epoch": 1.3632769602459915, + "grad_norm": 2.3525784015655518, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.733393669128418, + "num_tokens": 315327598.0, + "step": 12414 + }, + { + "epoch": 1.3633867779486053, + "grad_norm": 2.26188325881958, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7116822004318237, + "num_tokens": 315352645.0, + "step": 12415 + }, + { + "epoch": 1.363496595651219, + "grad_norm": 2.2366929054260254, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.7010641098022461, + "num_tokens": 315380430.0, + "step": 12416 + }, + { + "epoch": 1.3636064133538326, + "grad_norm": 2.2420291900634766, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7122819423675537, + "num_tokens": 315405156.0, + "step": 12417 + }, + { + "epoch": 1.3637162310564463, + "grad_norm": 1.9651719331741333, + "learning_rate": 1e-06, + "loss": 1.0167, + "mean_token_accuracy": 0.6814815998077393, + "num_tokens": 315439135.0, + "step": 12418 + }, + { + "epoch": 1.3638260487590599, + "grad_norm": 2.02941632270813, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7122341990470886, + "num_tokens": 315469969.0, + "step": 12419 + }, + { + "epoch": 1.3639358664616736, + "grad_norm": 2.2477941513061523, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7274290919303894, + "num_tokens": 315496334.0, + "step": 12420 + }, + { + "epoch": 1.3640456841642874, + "grad_norm": 2.1703450679779053, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.7225828766822815, + "num_tokens": 315523665.0, + "step": 12421 + }, + { + "epoch": 1.364155501866901, + "grad_norm": 2.030693531036377, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7023451924324036, + "num_tokens": 315554401.0, + "step": 12422 + }, + { + "epoch": 1.3642653195695145, + "grad_norm": 2.40525221824646, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7184467911720276, + "num_tokens": 315576801.0, + "step": 12423 + }, + { + "epoch": 1.3643751372721282, + "grad_norm": 2.1535212993621826, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7148506045341492, + "num_tokens": 315603864.0, + "step": 12424 + }, + { + "epoch": 1.364484954974742, + "grad_norm": 2.710489273071289, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7245470285415649, + "num_tokens": 315622175.0, + "step": 12425 + }, + { + "epoch": 1.3645947726773555, + "grad_norm": 2.3502516746520996, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.6981461644172668, + "num_tokens": 315649213.0, + "step": 12426 + }, + { + "epoch": 1.3647045903799693, + "grad_norm": 2.524190902709961, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7098756432533264, + "num_tokens": 315672396.0, + "step": 12427 + }, + { + "epoch": 1.3648144080825828, + "grad_norm": 2.198132038116455, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6946709156036377, + "num_tokens": 315702575.0, + "step": 12428 + }, + { + "epoch": 1.3649242257851966, + "grad_norm": 2.5208702087402344, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7281173467636108, + "num_tokens": 315724664.0, + "step": 12429 + }, + { + "epoch": 1.3650340434878103, + "grad_norm": 2.0309317111968994, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7446557283401489, + "num_tokens": 315752032.0, + "step": 12430 + }, + { + "epoch": 1.3651438611904239, + "grad_norm": 2.003300905227661, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7080715894699097, + "num_tokens": 315782455.0, + "step": 12431 + }, + { + "epoch": 1.3652536788930376, + "grad_norm": 2.276406764984131, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7116177678108215, + "num_tokens": 315806498.0, + "step": 12432 + }, + { + "epoch": 1.3653634965956511, + "grad_norm": 2.218005895614624, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7284351587295532, + "num_tokens": 315830682.0, + "step": 12433 + }, + { + "epoch": 1.365473314298265, + "grad_norm": 2.27492618560791, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7171609997749329, + "num_tokens": 315854640.0, + "step": 12434 + }, + { + "epoch": 1.3655831320008787, + "grad_norm": 2.273313522338867, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.7060985565185547, + "num_tokens": 315880119.0, + "step": 12435 + }, + { + "epoch": 1.3656929497034922, + "grad_norm": 2.2309415340423584, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7157333493232727, + "num_tokens": 315904281.0, + "step": 12436 + }, + { + "epoch": 1.3658027674061057, + "grad_norm": 2.237734079360962, + "learning_rate": 1e-06, + "loss": 1.0013, + "mean_token_accuracy": 0.6897264719009399, + "num_tokens": 315931587.0, + "step": 12437 + }, + { + "epoch": 1.3659125851087195, + "grad_norm": 2.384633779525757, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7251406908035278, + "num_tokens": 315954686.0, + "step": 12438 + }, + { + "epoch": 1.3660224028113332, + "grad_norm": 2.5255699157714844, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7228982448577881, + "num_tokens": 315976281.0, + "step": 12439 + }, + { + "epoch": 1.3661322205139468, + "grad_norm": 2.3575425148010254, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7098664045333862, + "num_tokens": 316001644.0, + "step": 12440 + }, + { + "epoch": 1.3662420382165605, + "grad_norm": 2.389275550842285, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7137141227722168, + "num_tokens": 316025860.0, + "step": 12441 + }, + { + "epoch": 1.366351855919174, + "grad_norm": 2.139547824859619, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.6983784437179565, + "num_tokens": 316054974.0, + "step": 12442 + }, + { + "epoch": 1.3664616736217878, + "grad_norm": 2.3394548892974854, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7368319034576416, + "num_tokens": 316077679.0, + "step": 12443 + }, + { + "epoch": 1.3665714913244016, + "grad_norm": 2.558971405029297, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7222371101379395, + "num_tokens": 316097599.0, + "step": 12444 + }, + { + "epoch": 1.3666813090270151, + "grad_norm": 2.359281063079834, + "learning_rate": 1e-06, + "loss": 0.8179, + "mean_token_accuracy": 0.7362759113311768, + "num_tokens": 316119633.0, + "step": 12445 + }, + { + "epoch": 1.3667911267296289, + "grad_norm": 2.2869246006011963, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7206709384918213, + "num_tokens": 316144474.0, + "step": 12446 + }, + { + "epoch": 1.3669009444322424, + "grad_norm": 2.7128243446350098, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7090849876403809, + "num_tokens": 316163947.0, + "step": 12447 + }, + { + "epoch": 1.3670107621348562, + "grad_norm": 2.5899126529693604, + "learning_rate": 1e-06, + "loss": 0.7735, + "mean_token_accuracy": 0.7473468780517578, + "num_tokens": 316183139.0, + "step": 12448 + }, + { + "epoch": 1.36712057983747, + "grad_norm": 2.294703960418701, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7311978936195374, + "num_tokens": 316206262.0, + "step": 12449 + }, + { + "epoch": 1.3672303975400835, + "grad_norm": 2.383469581604004, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7303376793861389, + "num_tokens": 316227674.0, + "step": 12450 + }, + { + "epoch": 1.367340215242697, + "grad_norm": 2.2620251178741455, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7266610860824585, + "num_tokens": 316252640.0, + "step": 12451 + }, + { + "epoch": 1.3674500329453108, + "grad_norm": 2.493844747543335, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7232331037521362, + "num_tokens": 316274524.0, + "step": 12452 + }, + { + "epoch": 1.3675598506479245, + "grad_norm": 2.1705687046051025, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7069069743156433, + "num_tokens": 316302134.0, + "step": 12453 + }, + { + "epoch": 1.367669668350538, + "grad_norm": 2.0462470054626465, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7241066694259644, + "num_tokens": 316331255.0, + "step": 12454 + }, + { + "epoch": 1.3677794860531518, + "grad_norm": 2.3915598392486572, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7129680514335632, + "num_tokens": 316354083.0, + "step": 12455 + }, + { + "epoch": 1.3678893037557653, + "grad_norm": 2.05277943611145, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.69640052318573, + "num_tokens": 316386665.0, + "step": 12456 + }, + { + "epoch": 1.367999121458379, + "grad_norm": 2.0658395290374756, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7134717702865601, + "num_tokens": 316416895.0, + "step": 12457 + }, + { + "epoch": 1.3681089391609929, + "grad_norm": 2.0326738357543945, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.7257720232009888, + "num_tokens": 316445773.0, + "step": 12458 + }, + { + "epoch": 1.3682187568636064, + "grad_norm": 2.3127784729003906, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7058997750282288, + "num_tokens": 316469106.0, + "step": 12459 + }, + { + "epoch": 1.3683285745662201, + "grad_norm": 2.379150390625, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7333090305328369, + "num_tokens": 316492700.0, + "step": 12460 + }, + { + "epoch": 1.3684383922688337, + "grad_norm": 2.159990072250366, + "learning_rate": 1e-06, + "loss": 1.0535, + "mean_token_accuracy": 0.6804914474487305, + "num_tokens": 316520627.0, + "step": 12461 + }, + { + "epoch": 1.3685482099714474, + "grad_norm": 2.391787052154541, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7138476371765137, + "num_tokens": 316543948.0, + "step": 12462 + }, + { + "epoch": 1.3686580276740612, + "grad_norm": 2.091681718826294, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7241948843002319, + "num_tokens": 316570608.0, + "step": 12463 + }, + { + "epoch": 1.3687678453766747, + "grad_norm": 2.1450908184051514, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7206714153289795, + "num_tokens": 316597560.0, + "step": 12464 + }, + { + "epoch": 1.3688776630792883, + "grad_norm": 2.3594751358032227, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7041606903076172, + "num_tokens": 316622588.0, + "step": 12465 + }, + { + "epoch": 1.368987480781902, + "grad_norm": 2.7090706825256348, + "learning_rate": 1e-06, + "loss": 0.8425, + "mean_token_accuracy": 0.7276772856712341, + "num_tokens": 316641305.0, + "step": 12466 + }, + { + "epoch": 1.3690972984845158, + "grad_norm": 2.249152660369873, + "learning_rate": 1e-06, + "loss": 0.8841, + "mean_token_accuracy": 0.7263592481613159, + "num_tokens": 316664495.0, + "step": 12467 + }, + { + "epoch": 1.3692071161871293, + "grad_norm": 2.3947062492370605, + "learning_rate": 1e-06, + "loss": 0.9827, + "mean_token_accuracy": 0.6980859041213989, + "num_tokens": 316689882.0, + "step": 12468 + }, + { + "epoch": 1.369316933889743, + "grad_norm": 2.3641085624694824, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7327840924263, + "num_tokens": 316713354.0, + "step": 12469 + }, + { + "epoch": 1.3694267515923566, + "grad_norm": 2.0086426734924316, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7142452597618103, + "num_tokens": 316747419.0, + "step": 12470 + }, + { + "epoch": 1.3695365692949704, + "grad_norm": 2.1525590419769287, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7145868539810181, + "num_tokens": 316774881.0, + "step": 12471 + }, + { + "epoch": 1.3696463869975841, + "grad_norm": 2.148171901702881, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7200228571891785, + "num_tokens": 316800051.0, + "step": 12472 + }, + { + "epoch": 1.3697562047001977, + "grad_norm": 2.189213275909424, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.722702145576477, + "num_tokens": 316827028.0, + "step": 12473 + }, + { + "epoch": 1.3698660224028114, + "grad_norm": 2.380654811859131, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.703904926776886, + "num_tokens": 316851726.0, + "step": 12474 + }, + { + "epoch": 1.369975840105425, + "grad_norm": 2.079986095428467, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7186504602432251, + "num_tokens": 316878561.0, + "step": 12475 + }, + { + "epoch": 1.3700856578080387, + "grad_norm": 2.411076307296753, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7300078868865967, + "num_tokens": 316901591.0, + "step": 12476 + }, + { + "epoch": 1.3701954755106522, + "grad_norm": 2.3067703247070312, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7173277139663696, + "num_tokens": 316926302.0, + "step": 12477 + }, + { + "epoch": 1.370305293213266, + "grad_norm": 2.042710065841675, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7359092235565186, + "num_tokens": 316954763.0, + "step": 12478 + }, + { + "epoch": 1.3704151109158795, + "grad_norm": 2.2957870960235596, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7174338698387146, + "num_tokens": 316979020.0, + "step": 12479 + }, + { + "epoch": 1.3705249286184933, + "grad_norm": 2.302809476852417, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7365075945854187, + "num_tokens": 317002157.0, + "step": 12480 + }, + { + "epoch": 1.370634746321107, + "grad_norm": 2.3034143447875977, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7224122881889343, + "num_tokens": 317024913.0, + "step": 12481 + }, + { + "epoch": 1.3707445640237206, + "grad_norm": 2.1852519512176514, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7139819264411926, + "num_tokens": 317052176.0, + "step": 12482 + }, + { + "epoch": 1.3708543817263343, + "grad_norm": 2.2470510005950928, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7181722521781921, + "num_tokens": 317079430.0, + "step": 12483 + }, + { + "epoch": 1.3709641994289479, + "grad_norm": 2.0842764377593994, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.732296347618103, + "num_tokens": 317108429.0, + "step": 12484 + }, + { + "epoch": 1.3710740171315616, + "grad_norm": 2.310307264328003, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7104519605636597, + "num_tokens": 317132555.0, + "step": 12485 + }, + { + "epoch": 1.3711838348341754, + "grad_norm": 2.2162230014801025, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.7422657012939453, + "num_tokens": 317158184.0, + "step": 12486 + }, + { + "epoch": 1.371293652536789, + "grad_norm": 2.1369550228118896, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.727731466293335, + "num_tokens": 317185041.0, + "step": 12487 + }, + { + "epoch": 1.3714034702394025, + "grad_norm": 2.242121696472168, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7125272750854492, + "num_tokens": 317212180.0, + "step": 12488 + }, + { + "epoch": 1.3715132879420162, + "grad_norm": 2.2746479511260986, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.732894778251648, + "num_tokens": 317236434.0, + "step": 12489 + }, + { + "epoch": 1.37162310564463, + "grad_norm": 2.500196933746338, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7199745178222656, + "num_tokens": 317258143.0, + "step": 12490 + }, + { + "epoch": 1.3717329233472435, + "grad_norm": 2.3362467288970947, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7286503911018372, + "num_tokens": 317281393.0, + "step": 12491 + }, + { + "epoch": 1.3718427410498573, + "grad_norm": 2.2103497982025146, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7406697869300842, + "num_tokens": 317306400.0, + "step": 12492 + }, + { + "epoch": 1.3719525587524708, + "grad_norm": 2.4474635124206543, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7276145219802856, + "num_tokens": 317329537.0, + "step": 12493 + }, + { + "epoch": 1.3720623764550846, + "grad_norm": 2.4344639778137207, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7057124376296997, + "num_tokens": 317352916.0, + "step": 12494 + }, + { + "epoch": 1.3721721941576983, + "grad_norm": 2.4223268032073975, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7388983964920044, + "num_tokens": 317374615.0, + "step": 12495 + }, + { + "epoch": 1.3722820118603118, + "grad_norm": 2.4586477279663086, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7192132472991943, + "num_tokens": 317398351.0, + "step": 12496 + }, + { + "epoch": 1.3723918295629256, + "grad_norm": 2.0849380493164062, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7156403064727783, + "num_tokens": 317428127.0, + "step": 12497 + }, + { + "epoch": 1.3725016472655391, + "grad_norm": 2.120872735977173, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.7264401912689209, + "num_tokens": 317453903.0, + "step": 12498 + }, + { + "epoch": 1.372611464968153, + "grad_norm": 2.2147414684295654, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7338513135910034, + "num_tokens": 317478380.0, + "step": 12499 + }, + { + "epoch": 1.3727212826707667, + "grad_norm": 2.160998821258545, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7190872430801392, + "num_tokens": 317507503.0, + "step": 12500 + }, + { + "epoch": 1.3728311003733802, + "grad_norm": 2.733534574508667, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7228674292564392, + "num_tokens": 317525907.0, + "step": 12501 + }, + { + "epoch": 1.3729409180759937, + "grad_norm": 2.456942558288574, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7175648212432861, + "num_tokens": 317546557.0, + "step": 12502 + }, + { + "epoch": 1.3730507357786075, + "grad_norm": 2.418729305267334, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.719603955745697, + "num_tokens": 317568410.0, + "step": 12503 + }, + { + "epoch": 1.3731605534812212, + "grad_norm": 2.08119797706604, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.712815523147583, + "num_tokens": 317598650.0, + "step": 12504 + }, + { + "epoch": 1.3732703711838348, + "grad_norm": 2.2366631031036377, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7074947357177734, + "num_tokens": 317626065.0, + "step": 12505 + }, + { + "epoch": 1.3733801888864485, + "grad_norm": 2.009181022644043, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7225756049156189, + "num_tokens": 317657642.0, + "step": 12506 + }, + { + "epoch": 1.373490006589062, + "grad_norm": 2.3894474506378174, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7153282761573792, + "num_tokens": 317680047.0, + "step": 12507 + }, + { + "epoch": 1.3735998242916758, + "grad_norm": 2.1157937049865723, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7134882211685181, + "num_tokens": 317706908.0, + "step": 12508 + }, + { + "epoch": 1.3737096419942896, + "grad_norm": 2.165409564971924, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7422833442687988, + "num_tokens": 317733088.0, + "step": 12509 + }, + { + "epoch": 1.3738194596969031, + "grad_norm": 2.396402597427368, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7103397846221924, + "num_tokens": 317756519.0, + "step": 12510 + }, + { + "epoch": 1.3739292773995169, + "grad_norm": 2.419903039932251, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7229261994361877, + "num_tokens": 317778775.0, + "step": 12511 + }, + { + "epoch": 1.3740390951021304, + "grad_norm": 2.08579421043396, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.7049000859260559, + "num_tokens": 317808527.0, + "step": 12512 + }, + { + "epoch": 1.3741489128047442, + "grad_norm": 2.252932548522949, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6853912472724915, + "num_tokens": 317835381.0, + "step": 12513 + }, + { + "epoch": 1.374258730507358, + "grad_norm": 2.066882371902466, + "learning_rate": 1e-06, + "loss": 1.0455, + "mean_token_accuracy": 0.683955192565918, + "num_tokens": 317866798.0, + "step": 12514 + }, + { + "epoch": 1.3743685482099715, + "grad_norm": 2.304354429244995, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7213916778564453, + "num_tokens": 317889349.0, + "step": 12515 + }, + { + "epoch": 1.374478365912585, + "grad_norm": 2.3235371112823486, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7190926671028137, + "num_tokens": 317913900.0, + "step": 12516 + }, + { + "epoch": 1.3745881836151987, + "grad_norm": 2.1043448448181152, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.6969455480575562, + "num_tokens": 317943197.0, + "step": 12517 + }, + { + "epoch": 1.3746980013178125, + "grad_norm": 2.528801679611206, + "learning_rate": 1e-06, + "loss": 0.9259, + "mean_token_accuracy": 0.7211984395980835, + "num_tokens": 317963874.0, + "step": 12518 + }, + { + "epoch": 1.374807819020426, + "grad_norm": 2.183382034301758, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7233837842941284, + "num_tokens": 317993112.0, + "step": 12519 + }, + { + "epoch": 1.3749176367230398, + "grad_norm": 2.0177481174468994, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7190617322921753, + "num_tokens": 318023425.0, + "step": 12520 + }, + { + "epoch": 1.3750274544256533, + "grad_norm": 2.325446605682373, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7337326407432556, + "num_tokens": 318046176.0, + "step": 12521 + }, + { + "epoch": 1.375137272128267, + "grad_norm": 2.3032126426696777, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7261698842048645, + "num_tokens": 318068753.0, + "step": 12522 + }, + { + "epoch": 1.3752470898308808, + "grad_norm": 2.1723251342773438, + "learning_rate": 1e-06, + "loss": 1.0174, + "mean_token_accuracy": 0.7024385929107666, + "num_tokens": 318095313.0, + "step": 12523 + }, + { + "epoch": 1.3753569075334944, + "grad_norm": 2.4511666297912598, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7225543260574341, + "num_tokens": 318118343.0, + "step": 12524 + }, + { + "epoch": 1.3754667252361081, + "grad_norm": 2.2215707302093506, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7440656423568726, + "num_tokens": 318143579.0, + "step": 12525 + }, + { + "epoch": 1.3755765429387217, + "grad_norm": 2.224026679992676, + "learning_rate": 1e-06, + "loss": 0.9642, + "mean_token_accuracy": 0.7039028406143188, + "num_tokens": 318169046.0, + "step": 12526 + }, + { + "epoch": 1.3756863606413354, + "grad_norm": 2.006283760070801, + "learning_rate": 1e-06, + "loss": 0.7828, + "mean_token_accuracy": 0.7532146573066711, + "num_tokens": 318197988.0, + "step": 12527 + }, + { + "epoch": 1.3757961783439492, + "grad_norm": 2.105832099914551, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7091473340988159, + "num_tokens": 318226396.0, + "step": 12528 + }, + { + "epoch": 1.3759059960465627, + "grad_norm": 2.462871551513672, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7145298719406128, + "num_tokens": 318249865.0, + "step": 12529 + }, + { + "epoch": 1.3760158137491763, + "grad_norm": 2.3581275939941406, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7367902994155884, + "num_tokens": 318272351.0, + "step": 12530 + }, + { + "epoch": 1.37612563145179, + "grad_norm": 2.2689459323883057, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7296688556671143, + "num_tokens": 318297986.0, + "step": 12531 + }, + { + "epoch": 1.3762354491544038, + "grad_norm": 2.343379497528076, + "learning_rate": 1e-06, + "loss": 0.8217, + "mean_token_accuracy": 0.7396315932273865, + "num_tokens": 318321539.0, + "step": 12532 + }, + { + "epoch": 1.3763452668570173, + "grad_norm": 2.0485188961029053, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7022533416748047, + "num_tokens": 318352707.0, + "step": 12533 + }, + { + "epoch": 1.376455084559631, + "grad_norm": 2.0984725952148438, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7277959585189819, + "num_tokens": 318380845.0, + "step": 12534 + }, + { + "epoch": 1.3765649022622446, + "grad_norm": 2.3224363327026367, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7347137928009033, + "num_tokens": 318403662.0, + "step": 12535 + }, + { + "epoch": 1.3766747199648584, + "grad_norm": 2.352839469909668, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7218540906906128, + "num_tokens": 318427168.0, + "step": 12536 + }, + { + "epoch": 1.376784537667472, + "grad_norm": 2.2616076469421387, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7109900712966919, + "num_tokens": 318453366.0, + "step": 12537 + }, + { + "epoch": 1.3768943553700856, + "grad_norm": 2.0599117279052734, + "learning_rate": 1e-06, + "loss": 0.9519, + "mean_token_accuracy": 0.7084013223648071, + "num_tokens": 318481647.0, + "step": 12538 + }, + { + "epoch": 1.3770041730726994, + "grad_norm": 2.2256460189819336, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.7049812078475952, + "num_tokens": 318508684.0, + "step": 12539 + }, + { + "epoch": 1.377113990775313, + "grad_norm": 2.1896040439605713, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.7050685286521912, + "num_tokens": 318536967.0, + "step": 12540 + }, + { + "epoch": 1.3772238084779267, + "grad_norm": 2.345963954925537, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7428666949272156, + "num_tokens": 318560470.0, + "step": 12541 + }, + { + "epoch": 1.3773336261805402, + "grad_norm": 2.1612977981567383, + "learning_rate": 1e-06, + "loss": 0.7493, + "mean_token_accuracy": 0.7686068415641785, + "num_tokens": 318585403.0, + "step": 12542 + }, + { + "epoch": 1.377443443883154, + "grad_norm": 2.2110610008239746, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7191022038459778, + "num_tokens": 318613439.0, + "step": 12543 + }, + { + "epoch": 1.3775532615857675, + "grad_norm": 2.3143675327301025, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7402688264846802, + "num_tokens": 318635939.0, + "step": 12544 + }, + { + "epoch": 1.3776630792883813, + "grad_norm": 2.449321746826172, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7068154215812683, + "num_tokens": 318658083.0, + "step": 12545 + }, + { + "epoch": 1.377772896990995, + "grad_norm": 2.2002813816070557, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7264147996902466, + "num_tokens": 318685087.0, + "step": 12546 + }, + { + "epoch": 1.3778827146936086, + "grad_norm": 2.0379199981689453, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.6982051134109497, + "num_tokens": 318716692.0, + "step": 12547 + }, + { + "epoch": 1.3779925323962223, + "grad_norm": 2.1914145946502686, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7536247968673706, + "num_tokens": 318741398.0, + "step": 12548 + }, + { + "epoch": 1.3781023500988359, + "grad_norm": 2.4093475341796875, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7228959798812866, + "num_tokens": 318761839.0, + "step": 12549 + }, + { + "epoch": 1.3782121678014496, + "grad_norm": 2.3737876415252686, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7310785055160522, + "num_tokens": 318785540.0, + "step": 12550 + }, + { + "epoch": 1.3783219855040634, + "grad_norm": 2.427717447280884, + "learning_rate": 1e-06, + "loss": 0.9661, + "mean_token_accuracy": 0.7087565064430237, + "num_tokens": 318810894.0, + "step": 12551 + }, + { + "epoch": 1.378431803206677, + "grad_norm": 2.3932104110717773, + "learning_rate": 1e-06, + "loss": 0.7925, + "mean_token_accuracy": 0.7524645924568176, + "num_tokens": 318834891.0, + "step": 12552 + }, + { + "epoch": 1.3785416209092904, + "grad_norm": 2.4962730407714844, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.7097278833389282, + "num_tokens": 318859167.0, + "step": 12553 + }, + { + "epoch": 1.3786514386119042, + "grad_norm": 2.278566598892212, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7495449781417847, + "num_tokens": 318881868.0, + "step": 12554 + }, + { + "epoch": 1.378761256314518, + "grad_norm": 2.8912394046783447, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7376531362533569, + "num_tokens": 318899589.0, + "step": 12555 + }, + { + "epoch": 1.3788710740171315, + "grad_norm": 2.455108880996704, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7173621654510498, + "num_tokens": 318922605.0, + "step": 12556 + }, + { + "epoch": 1.3789808917197452, + "grad_norm": 2.275183916091919, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7329383492469788, + "num_tokens": 318947311.0, + "step": 12557 + }, + { + "epoch": 1.3790907094223588, + "grad_norm": 2.3781845569610596, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7095292806625366, + "num_tokens": 318971560.0, + "step": 12558 + }, + { + "epoch": 1.3792005271249725, + "grad_norm": 2.2535934448242188, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7396198511123657, + "num_tokens": 318996293.0, + "step": 12559 + }, + { + "epoch": 1.3793103448275863, + "grad_norm": 2.3697097301483154, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7244164943695068, + "num_tokens": 319021096.0, + "step": 12560 + }, + { + "epoch": 1.3794201625301998, + "grad_norm": 2.467318058013916, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7326033115386963, + "num_tokens": 319042319.0, + "step": 12561 + }, + { + "epoch": 1.3795299802328136, + "grad_norm": 2.287842273712158, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7234508991241455, + "num_tokens": 319066753.0, + "step": 12562 + }, + { + "epoch": 1.3796397979354271, + "grad_norm": 2.1681556701660156, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.6994402408599854, + "num_tokens": 319096353.0, + "step": 12563 + }, + { + "epoch": 1.3797496156380409, + "grad_norm": 2.1009652614593506, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.701468825340271, + "num_tokens": 319126072.0, + "step": 12564 + }, + { + "epoch": 1.3798594333406546, + "grad_norm": 2.0275659561157227, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7515637278556824, + "num_tokens": 319155532.0, + "step": 12565 + }, + { + "epoch": 1.3799692510432682, + "grad_norm": 2.1008057594299316, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7101881504058838, + "num_tokens": 319184452.0, + "step": 12566 + }, + { + "epoch": 1.3800790687458817, + "grad_norm": 2.5374326705932617, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.732776939868927, + "num_tokens": 319205043.0, + "step": 12567 + }, + { + "epoch": 1.3801888864484955, + "grad_norm": 2.1833889484405518, + "learning_rate": 1e-06, + "loss": 0.978, + "mean_token_accuracy": 0.6979302167892456, + "num_tokens": 319232702.0, + "step": 12568 + }, + { + "epoch": 1.3802987041511092, + "grad_norm": 2.2322375774383545, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7054787874221802, + "num_tokens": 319257950.0, + "step": 12569 + }, + { + "epoch": 1.3804085218537228, + "grad_norm": 1.9529746770858765, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7400546669960022, + "num_tokens": 319290654.0, + "step": 12570 + }, + { + "epoch": 1.3805183395563365, + "grad_norm": 2.301577091217041, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7349967956542969, + "num_tokens": 319314252.0, + "step": 12571 + }, + { + "epoch": 1.38062815725895, + "grad_norm": 2.0943350791931152, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7263590693473816, + "num_tokens": 319342798.0, + "step": 12572 + }, + { + "epoch": 1.3807379749615638, + "grad_norm": 2.1585304737091064, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7178413271903992, + "num_tokens": 319370891.0, + "step": 12573 + }, + { + "epoch": 1.3808477926641776, + "grad_norm": 2.2946677207946777, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.718045175075531, + "num_tokens": 319396892.0, + "step": 12574 + }, + { + "epoch": 1.380957610366791, + "grad_norm": 2.2879316806793213, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7235018014907837, + "num_tokens": 319420581.0, + "step": 12575 + }, + { + "epoch": 1.3810674280694049, + "grad_norm": 2.257375717163086, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7232067584991455, + "num_tokens": 319446626.0, + "step": 12576 + }, + { + "epoch": 1.3811772457720184, + "grad_norm": 2.147911787033081, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7313507795333862, + "num_tokens": 319474623.0, + "step": 12577 + }, + { + "epoch": 1.3812870634746321, + "grad_norm": 2.3614351749420166, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7169640064239502, + "num_tokens": 319500175.0, + "step": 12578 + }, + { + "epoch": 1.381396881177246, + "grad_norm": 2.2675137519836426, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7051798701286316, + "num_tokens": 319526152.0, + "step": 12579 + }, + { + "epoch": 1.3815066988798594, + "grad_norm": 2.1575915813446045, + "learning_rate": 1e-06, + "loss": 0.988, + "mean_token_accuracy": 0.7056851387023926, + "num_tokens": 319555463.0, + "step": 12580 + }, + { + "epoch": 1.381616516582473, + "grad_norm": 2.519531726837158, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.736056923866272, + "num_tokens": 319576230.0, + "step": 12581 + }, + { + "epoch": 1.3817263342850867, + "grad_norm": 2.172130823135376, + "learning_rate": 1e-06, + "loss": 0.9696, + "mean_token_accuracy": 0.700352668762207, + "num_tokens": 319607102.0, + "step": 12582 + }, + { + "epoch": 1.3818361519877005, + "grad_norm": 2.1478779315948486, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7361498475074768, + "num_tokens": 319635947.0, + "step": 12583 + }, + { + "epoch": 1.381945969690314, + "grad_norm": 2.080437660217285, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7180485725402832, + "num_tokens": 319664064.0, + "step": 12584 + }, + { + "epoch": 1.3820557873929278, + "grad_norm": 2.377913475036621, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7183839082717896, + "num_tokens": 319687102.0, + "step": 12585 + }, + { + "epoch": 1.3821656050955413, + "grad_norm": 2.230257034301758, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7258297204971313, + "num_tokens": 319712615.0, + "step": 12586 + }, + { + "epoch": 1.382275422798155, + "grad_norm": 2.1816110610961914, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7073495388031006, + "num_tokens": 319740897.0, + "step": 12587 + }, + { + "epoch": 1.3823852405007688, + "grad_norm": 2.5058040618896484, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7126036882400513, + "num_tokens": 319764980.0, + "step": 12588 + }, + { + "epoch": 1.3824950582033824, + "grad_norm": 2.3779137134552, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7296044826507568, + "num_tokens": 319787815.0, + "step": 12589 + }, + { + "epoch": 1.3826048759059961, + "grad_norm": 2.082411050796509, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7201359272003174, + "num_tokens": 319815405.0, + "step": 12590 + }, + { + "epoch": 1.3827146936086097, + "grad_norm": 2.3275911808013916, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7061530351638794, + "num_tokens": 319840013.0, + "step": 12591 + }, + { + "epoch": 1.3828245113112234, + "grad_norm": 1.8955894708633423, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.71971595287323, + "num_tokens": 319869434.0, + "step": 12592 + }, + { + "epoch": 1.382934329013837, + "grad_norm": 2.2342898845672607, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7429884076118469, + "num_tokens": 319892962.0, + "step": 12593 + }, + { + "epoch": 1.3830441467164507, + "grad_norm": 2.675027370452881, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.716628909111023, + "num_tokens": 319912827.0, + "step": 12594 + }, + { + "epoch": 1.3831539644190642, + "grad_norm": 2.6442461013793945, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7222249507904053, + "num_tokens": 319932719.0, + "step": 12595 + }, + { + "epoch": 1.383263782121678, + "grad_norm": 2.604614496231079, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7355785369873047, + "num_tokens": 319951577.0, + "step": 12596 + }, + { + "epoch": 1.3833735998242918, + "grad_norm": 2.193913698196411, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7019491195678711, + "num_tokens": 319981037.0, + "step": 12597 + }, + { + "epoch": 1.3834834175269053, + "grad_norm": 2.217160701751709, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7238585352897644, + "num_tokens": 320006101.0, + "step": 12598 + }, + { + "epoch": 1.383593235229519, + "grad_norm": 2.215331554412842, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7073739767074585, + "num_tokens": 320031107.0, + "step": 12599 + }, + { + "epoch": 1.3837030529321326, + "grad_norm": 2.3446836471557617, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7186489105224609, + "num_tokens": 320054627.0, + "step": 12600 + }, + { + "epoch": 1.3838128706347463, + "grad_norm": 2.0418472290039062, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7351515889167786, + "num_tokens": 320085111.0, + "step": 12601 + }, + { + "epoch": 1.38392268833736, + "grad_norm": 2.4102423191070557, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.6988182067871094, + "num_tokens": 320107936.0, + "step": 12602 + }, + { + "epoch": 1.3840325060399736, + "grad_norm": 2.1346802711486816, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7041170597076416, + "num_tokens": 320135536.0, + "step": 12603 + }, + { + "epoch": 1.3841423237425874, + "grad_norm": 2.165875196456909, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7293565273284912, + "num_tokens": 320162115.0, + "step": 12604 + }, + { + "epoch": 1.384252141445201, + "grad_norm": 2.2031280994415283, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7498884201049805, + "num_tokens": 320186597.0, + "step": 12605 + }, + { + "epoch": 1.3843619591478147, + "grad_norm": 2.4911715984344482, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7341904640197754, + "num_tokens": 320206922.0, + "step": 12606 + }, + { + "epoch": 1.3844717768504282, + "grad_norm": 2.050564765930176, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7225337624549866, + "num_tokens": 320236257.0, + "step": 12607 + }, + { + "epoch": 1.384581594553042, + "grad_norm": 2.2690939903259277, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6935192346572876, + "num_tokens": 320262978.0, + "step": 12608 + }, + { + "epoch": 1.3846914122556555, + "grad_norm": 2.0559041500091553, + "learning_rate": 1e-06, + "loss": 0.9699, + "mean_token_accuracy": 0.706924319267273, + "num_tokens": 320295284.0, + "step": 12609 + }, + { + "epoch": 1.3848012299582693, + "grad_norm": 2.1081912517547607, + "learning_rate": 1e-06, + "loss": 1.0222, + "mean_token_accuracy": 0.6861092448234558, + "num_tokens": 320324955.0, + "step": 12610 + }, + { + "epoch": 1.384911047660883, + "grad_norm": 2.007568359375, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7156859636306763, + "num_tokens": 320353800.0, + "step": 12611 + }, + { + "epoch": 1.3850208653634966, + "grad_norm": 2.4102623462677, + "learning_rate": 1e-06, + "loss": 0.8006, + "mean_token_accuracy": 0.741277813911438, + "num_tokens": 320374676.0, + "step": 12612 + }, + { + "epoch": 1.3851306830661103, + "grad_norm": 2.198695182800293, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.706779956817627, + "num_tokens": 320403820.0, + "step": 12613 + }, + { + "epoch": 1.3852405007687238, + "grad_norm": 2.379220724105835, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.719306230545044, + "num_tokens": 320427357.0, + "step": 12614 + }, + { + "epoch": 1.3853503184713376, + "grad_norm": 2.057631731033325, + "learning_rate": 1e-06, + "loss": 0.9757, + "mean_token_accuracy": 0.6991757154464722, + "num_tokens": 320459359.0, + "step": 12615 + }, + { + "epoch": 1.3854601361739514, + "grad_norm": 2.2763514518737793, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7155055999755859, + "num_tokens": 320484967.0, + "step": 12616 + }, + { + "epoch": 1.385569953876565, + "grad_norm": 2.4398746490478516, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7037864327430725, + "num_tokens": 320508998.0, + "step": 12617 + }, + { + "epoch": 1.3856797715791784, + "grad_norm": 2.1449759006500244, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7095425724983215, + "num_tokens": 320537538.0, + "step": 12618 + }, + { + "epoch": 1.3857895892817922, + "grad_norm": 2.4650638103485107, + "learning_rate": 1e-06, + "loss": 0.8662, + "mean_token_accuracy": 0.7222638130187988, + "num_tokens": 320559892.0, + "step": 12619 + }, + { + "epoch": 1.385899406984406, + "grad_norm": 2.2580759525299072, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7145607471466064, + "num_tokens": 320584902.0, + "step": 12620 + }, + { + "epoch": 1.3860092246870195, + "grad_norm": 1.937159776687622, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7077133059501648, + "num_tokens": 320617548.0, + "step": 12621 + }, + { + "epoch": 1.3861190423896332, + "grad_norm": 2.4197824001312256, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7070293426513672, + "num_tokens": 320641227.0, + "step": 12622 + }, + { + "epoch": 1.3862288600922468, + "grad_norm": 2.213446617126465, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7305797934532166, + "num_tokens": 320668337.0, + "step": 12623 + }, + { + "epoch": 1.3863386777948605, + "grad_norm": 2.2458901405334473, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.7018497586250305, + "num_tokens": 320694933.0, + "step": 12624 + }, + { + "epoch": 1.3864484954974743, + "grad_norm": 2.167482614517212, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7115067839622498, + "num_tokens": 320722520.0, + "step": 12625 + }, + { + "epoch": 1.3865583132000878, + "grad_norm": 2.417775869369507, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.735680878162384, + "num_tokens": 320742849.0, + "step": 12626 + }, + { + "epoch": 1.3866681309027016, + "grad_norm": 2.1652140617370605, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.6969934701919556, + "num_tokens": 320772913.0, + "step": 12627 + }, + { + "epoch": 1.3867779486053151, + "grad_norm": 2.4594295024871826, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.7490009665489197, + "num_tokens": 320795157.0, + "step": 12628 + }, + { + "epoch": 1.3868877663079289, + "grad_norm": 2.252082109451294, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.735362708568573, + "num_tokens": 320819469.0, + "step": 12629 + }, + { + "epoch": 1.3869975840105426, + "grad_norm": 2.0870702266693115, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7329244017601013, + "num_tokens": 320845865.0, + "step": 12630 + }, + { + "epoch": 1.3871074017131562, + "grad_norm": 2.1081812381744385, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7268895506858826, + "num_tokens": 320873607.0, + "step": 12631 + }, + { + "epoch": 1.3872172194157697, + "grad_norm": 2.058011770248413, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7171267867088318, + "num_tokens": 320902835.0, + "step": 12632 + }, + { + "epoch": 1.3873270371183835, + "grad_norm": 2.1145057678222656, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7000749111175537, + "num_tokens": 320933895.0, + "step": 12633 + }, + { + "epoch": 1.3874368548209972, + "grad_norm": 2.2654659748077393, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7220057249069214, + "num_tokens": 320958207.0, + "step": 12634 + }, + { + "epoch": 1.3875466725236107, + "grad_norm": 2.4476563930511475, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7114184498786926, + "num_tokens": 320980919.0, + "step": 12635 + }, + { + "epoch": 1.3876564902262245, + "grad_norm": 2.2482621669769287, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.6826038956642151, + "num_tokens": 321008373.0, + "step": 12636 + }, + { + "epoch": 1.387766307928838, + "grad_norm": 2.6493117809295654, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.7194609045982361, + "num_tokens": 321028781.0, + "step": 12637 + }, + { + "epoch": 1.3878761256314518, + "grad_norm": 2.1697590351104736, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7121791243553162, + "num_tokens": 321058542.0, + "step": 12638 + }, + { + "epoch": 1.3879859433340656, + "grad_norm": 2.2400941848754883, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7282055020332336, + "num_tokens": 321082501.0, + "step": 12639 + }, + { + "epoch": 1.388095761036679, + "grad_norm": 2.130195379257202, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7169305086135864, + "num_tokens": 321110774.0, + "step": 12640 + }, + { + "epoch": 1.3882055787392928, + "grad_norm": 2.429875373840332, + "learning_rate": 1e-06, + "loss": 0.8041, + "mean_token_accuracy": 0.7468584775924683, + "num_tokens": 321131714.0, + "step": 12641 + }, + { + "epoch": 1.3883153964419064, + "grad_norm": 2.3895342350006104, + "learning_rate": 1e-06, + "loss": 0.8298, + "mean_token_accuracy": 0.7419508695602417, + "num_tokens": 321153670.0, + "step": 12642 + }, + { + "epoch": 1.3884252141445201, + "grad_norm": 2.3695790767669678, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7171207666397095, + "num_tokens": 321176272.0, + "step": 12643 + }, + { + "epoch": 1.388535031847134, + "grad_norm": 2.6866579055786133, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.722915530204773, + "num_tokens": 321196468.0, + "step": 12644 + }, + { + "epoch": 1.3886448495497474, + "grad_norm": 2.1512577533721924, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7371655702590942, + "num_tokens": 321224520.0, + "step": 12645 + }, + { + "epoch": 1.388754667252361, + "grad_norm": 2.262273073196411, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7119069695472717, + "num_tokens": 321250757.0, + "step": 12646 + }, + { + "epoch": 1.3888644849549747, + "grad_norm": 2.414978504180908, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7217051386833191, + "num_tokens": 321272708.0, + "step": 12647 + }, + { + "epoch": 1.3889743026575885, + "grad_norm": 2.341557264328003, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7232133150100708, + "num_tokens": 321296711.0, + "step": 12648 + }, + { + "epoch": 1.389084120360202, + "grad_norm": 2.6314570903778076, + "learning_rate": 1e-06, + "loss": 0.8062, + "mean_token_accuracy": 0.7430682182312012, + "num_tokens": 321316693.0, + "step": 12649 + }, + { + "epoch": 1.3891939380628158, + "grad_norm": 2.1876041889190674, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.6973500847816467, + "num_tokens": 321344990.0, + "step": 12650 + }, + { + "epoch": 1.3893037557654293, + "grad_norm": 2.0198283195495605, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7152333855628967, + "num_tokens": 321376874.0, + "step": 12651 + }, + { + "epoch": 1.389413573468043, + "grad_norm": 2.2332370281219482, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7166798114776611, + "num_tokens": 321405103.0, + "step": 12652 + }, + { + "epoch": 1.3895233911706568, + "grad_norm": 2.0444605350494385, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7046608924865723, + "num_tokens": 321435002.0, + "step": 12653 + }, + { + "epoch": 1.3896332088732704, + "grad_norm": 2.3216214179992676, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7157968878746033, + "num_tokens": 321460761.0, + "step": 12654 + }, + { + "epoch": 1.389743026575884, + "grad_norm": 2.000338315963745, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7043025493621826, + "num_tokens": 321492942.0, + "step": 12655 + }, + { + "epoch": 1.3898528442784976, + "grad_norm": 2.661900520324707, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7162827253341675, + "num_tokens": 321512598.0, + "step": 12656 + }, + { + "epoch": 1.3899626619811114, + "grad_norm": 2.0689971446990967, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7250765562057495, + "num_tokens": 321539716.0, + "step": 12657 + }, + { + "epoch": 1.390072479683725, + "grad_norm": 2.141756296157837, + "learning_rate": 1e-06, + "loss": 0.9592, + "mean_token_accuracy": 0.7076882123947144, + "num_tokens": 321570745.0, + "step": 12658 + }, + { + "epoch": 1.3901822973863387, + "grad_norm": 2.3494150638580322, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.722615122795105, + "num_tokens": 321593089.0, + "step": 12659 + }, + { + "epoch": 1.3902921150889522, + "grad_norm": 2.335810899734497, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7453182935714722, + "num_tokens": 321616187.0, + "step": 12660 + }, + { + "epoch": 1.390401932791566, + "grad_norm": 2.439563751220703, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7318865656852722, + "num_tokens": 321639596.0, + "step": 12661 + }, + { + "epoch": 1.3905117504941797, + "grad_norm": 2.2607710361480713, + "learning_rate": 1e-06, + "loss": 0.8094, + "mean_token_accuracy": 0.7421540021896362, + "num_tokens": 321662613.0, + "step": 12662 + }, + { + "epoch": 1.3906215681967933, + "grad_norm": 2.290846347808838, + "learning_rate": 1e-06, + "loss": 0.8358, + "mean_token_accuracy": 0.7346581816673279, + "num_tokens": 321686827.0, + "step": 12663 + }, + { + "epoch": 1.390731385899407, + "grad_norm": 2.068767786026001, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7105727195739746, + "num_tokens": 321716481.0, + "step": 12664 + }, + { + "epoch": 1.3908412036020206, + "grad_norm": 2.5097432136535645, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7139127254486084, + "num_tokens": 321738775.0, + "step": 12665 + }, + { + "epoch": 1.3909510213046343, + "grad_norm": 2.3445985317230225, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7251604795455933, + "num_tokens": 321761231.0, + "step": 12666 + }, + { + "epoch": 1.391060839007248, + "grad_norm": 2.1495437622070312, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7249855995178223, + "num_tokens": 321789761.0, + "step": 12667 + }, + { + "epoch": 1.3911706567098616, + "grad_norm": 2.168818712234497, + "learning_rate": 1e-06, + "loss": 0.7892, + "mean_token_accuracy": 0.7401121854782104, + "num_tokens": 321815117.0, + "step": 12668 + }, + { + "epoch": 1.3912804744124752, + "grad_norm": 2.275585889816284, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.6986823678016663, + "num_tokens": 321841306.0, + "step": 12669 + }, + { + "epoch": 1.391390292115089, + "grad_norm": 1.8289026021957397, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7019813656806946, + "num_tokens": 321874080.0, + "step": 12670 + }, + { + "epoch": 1.3915001098177027, + "grad_norm": 2.160590887069702, + "learning_rate": 1e-06, + "loss": 0.8232, + "mean_token_accuracy": 0.7289824485778809, + "num_tokens": 321900938.0, + "step": 12671 + }, + { + "epoch": 1.3916099275203162, + "grad_norm": 2.430528402328491, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7218170166015625, + "num_tokens": 321925453.0, + "step": 12672 + }, + { + "epoch": 1.39171974522293, + "grad_norm": 2.0738208293914795, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7143829464912415, + "num_tokens": 321952046.0, + "step": 12673 + }, + { + "epoch": 1.3918295629255435, + "grad_norm": 2.415987014770508, + "learning_rate": 1e-06, + "loss": 0.7829, + "mean_token_accuracy": 0.7436136603355408, + "num_tokens": 321972320.0, + "step": 12674 + }, + { + "epoch": 1.3919393806281573, + "grad_norm": 2.169522285461426, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7349389791488647, + "num_tokens": 321997370.0, + "step": 12675 + }, + { + "epoch": 1.392049198330771, + "grad_norm": 2.167257308959961, + "learning_rate": 1e-06, + "loss": 0.8108, + "mean_token_accuracy": 0.7384586334228516, + "num_tokens": 322023340.0, + "step": 12676 + }, + { + "epoch": 1.3921590160333845, + "grad_norm": 2.2726807594299316, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7229912877082825, + "num_tokens": 322048616.0, + "step": 12677 + }, + { + "epoch": 1.3922688337359983, + "grad_norm": 2.4612457752227783, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7131094336509705, + "num_tokens": 322070010.0, + "step": 12678 + }, + { + "epoch": 1.3923786514386118, + "grad_norm": 2.4286692142486572, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7140229940414429, + "num_tokens": 322093342.0, + "step": 12679 + }, + { + "epoch": 1.3924884691412256, + "grad_norm": 2.262366533279419, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.7400907278060913, + "num_tokens": 322116316.0, + "step": 12680 + }, + { + "epoch": 1.3925982868438394, + "grad_norm": 2.2133586406707764, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.730931282043457, + "num_tokens": 322142185.0, + "step": 12681 + }, + { + "epoch": 1.3927081045464529, + "grad_norm": 2.3205406665802, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7375301122665405, + "num_tokens": 322165527.0, + "step": 12682 + }, + { + "epoch": 1.3928179222490664, + "grad_norm": 2.41909122467041, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.736731231212616, + "num_tokens": 322186338.0, + "step": 12683 + }, + { + "epoch": 1.3929277399516802, + "grad_norm": 2.2376346588134766, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7256399989128113, + "num_tokens": 322211004.0, + "step": 12684 + }, + { + "epoch": 1.393037557654294, + "grad_norm": 2.3003063201904297, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7517536878585815, + "num_tokens": 322232989.0, + "step": 12685 + }, + { + "epoch": 1.3931473753569075, + "grad_norm": 2.072298526763916, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7066320180892944, + "num_tokens": 322263788.0, + "step": 12686 + }, + { + "epoch": 1.3932571930595212, + "grad_norm": 2.3080248832702637, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7259949445724487, + "num_tokens": 322290116.0, + "step": 12687 + }, + { + "epoch": 1.3933670107621348, + "grad_norm": 2.733851432800293, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7387441396713257, + "num_tokens": 322308691.0, + "step": 12688 + }, + { + "epoch": 1.3934768284647485, + "grad_norm": 2.1584811210632324, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7115957736968994, + "num_tokens": 322337489.0, + "step": 12689 + }, + { + "epoch": 1.3935866461673623, + "grad_norm": 2.4991517066955566, + "learning_rate": 1e-06, + "loss": 0.8419, + "mean_token_accuracy": 0.7330716848373413, + "num_tokens": 322358897.0, + "step": 12690 + }, + { + "epoch": 1.3936964638699758, + "grad_norm": 2.242067813873291, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7135456800460815, + "num_tokens": 322385557.0, + "step": 12691 + }, + { + "epoch": 1.3938062815725896, + "grad_norm": 2.3169400691986084, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7267122268676758, + "num_tokens": 322408166.0, + "step": 12692 + }, + { + "epoch": 1.393916099275203, + "grad_norm": 2.256180763244629, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7390541434288025, + "num_tokens": 322433375.0, + "step": 12693 + }, + { + "epoch": 1.3940259169778169, + "grad_norm": 2.123638391494751, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7321039438247681, + "num_tokens": 322460432.0, + "step": 12694 + }, + { + "epoch": 1.3941357346804306, + "grad_norm": 2.2346954345703125, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7244654893875122, + "num_tokens": 322485687.0, + "step": 12695 + }, + { + "epoch": 1.3942455523830442, + "grad_norm": 2.1738781929016113, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7199312448501587, + "num_tokens": 322512085.0, + "step": 12696 + }, + { + "epoch": 1.3943553700856577, + "grad_norm": 2.6244633197784424, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7444714307785034, + "num_tokens": 322532494.0, + "step": 12697 + }, + { + "epoch": 1.3944651877882714, + "grad_norm": 2.158811330795288, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.716796875, + "num_tokens": 322559456.0, + "step": 12698 + }, + { + "epoch": 1.3945750054908852, + "grad_norm": 2.7330899238586426, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.7250597476959229, + "num_tokens": 322578094.0, + "step": 12699 + }, + { + "epoch": 1.3946848231934987, + "grad_norm": 2.346050500869751, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7250404357910156, + "num_tokens": 322601286.0, + "step": 12700 + }, + { + "epoch": 1.3947946408961125, + "grad_norm": 2.723611354827881, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7366206645965576, + "num_tokens": 322620912.0, + "step": 12701 + }, + { + "epoch": 1.394904458598726, + "grad_norm": 2.8455944061279297, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.734870970249176, + "num_tokens": 322639668.0, + "step": 12702 + }, + { + "epoch": 1.3950142763013398, + "grad_norm": 1.991303563117981, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.6832894086837769, + "num_tokens": 322671788.0, + "step": 12703 + }, + { + "epoch": 1.3951240940039535, + "grad_norm": 2.1212611198425293, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.7071773409843445, + "num_tokens": 322701529.0, + "step": 12704 + }, + { + "epoch": 1.395233911706567, + "grad_norm": 2.2553088665008545, + "learning_rate": 1e-06, + "loss": 0.8034, + "mean_token_accuracy": 0.745902419090271, + "num_tokens": 322725693.0, + "step": 12705 + }, + { + "epoch": 1.3953437294091808, + "grad_norm": 2.106363534927368, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7282508015632629, + "num_tokens": 322751244.0, + "step": 12706 + }, + { + "epoch": 1.3954535471117944, + "grad_norm": 2.0847227573394775, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.6982370615005493, + "num_tokens": 322779696.0, + "step": 12707 + }, + { + "epoch": 1.3955633648144081, + "grad_norm": 2.406120538711548, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.719785213470459, + "num_tokens": 322805326.0, + "step": 12708 + }, + { + "epoch": 1.3956731825170219, + "grad_norm": 2.12618088722229, + "learning_rate": 1e-06, + "loss": 0.9596, + "mean_token_accuracy": 0.7028941512107849, + "num_tokens": 322833812.0, + "step": 12709 + }, + { + "epoch": 1.3957830002196354, + "grad_norm": 2.1862754821777344, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7322074174880981, + "num_tokens": 322860883.0, + "step": 12710 + }, + { + "epoch": 1.395892817922249, + "grad_norm": 2.3191840648651123, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7213301658630371, + "num_tokens": 322886289.0, + "step": 12711 + }, + { + "epoch": 1.3960026356248627, + "grad_norm": 2.1895992755889893, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.6955119967460632, + "num_tokens": 322915778.0, + "step": 12712 + }, + { + "epoch": 1.3961124533274765, + "grad_norm": 2.013514757156372, + "learning_rate": 1e-06, + "loss": 0.9572, + "mean_token_accuracy": 0.7039087414741516, + "num_tokens": 322945371.0, + "step": 12713 + }, + { + "epoch": 1.39622227103009, + "grad_norm": 2.063180446624756, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7153172492980957, + "num_tokens": 322975972.0, + "step": 12714 + }, + { + "epoch": 1.3963320887327038, + "grad_norm": 2.0476298332214355, + "learning_rate": 1e-06, + "loss": 0.961, + "mean_token_accuracy": 0.7021484375, + "num_tokens": 323007783.0, + "step": 12715 + }, + { + "epoch": 1.3964419064353173, + "grad_norm": 2.5663106441497803, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7408043146133423, + "num_tokens": 323027614.0, + "step": 12716 + }, + { + "epoch": 1.396551724137931, + "grad_norm": 2.334193229675293, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7170875072479248, + "num_tokens": 323053444.0, + "step": 12717 + }, + { + "epoch": 1.3966615418405448, + "grad_norm": 2.1496739387512207, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.6985129117965698, + "num_tokens": 323082411.0, + "step": 12718 + }, + { + "epoch": 1.3967713595431583, + "grad_norm": 2.1386735439300537, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7030940055847168, + "num_tokens": 323110420.0, + "step": 12719 + }, + { + "epoch": 1.396881177245772, + "grad_norm": 2.2068841457366943, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7163029313087463, + "num_tokens": 323136375.0, + "step": 12720 + }, + { + "epoch": 1.3969909949483856, + "grad_norm": 1.947275161743164, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7245463132858276, + "num_tokens": 323167715.0, + "step": 12721 + }, + { + "epoch": 1.3971008126509994, + "grad_norm": 2.3075039386749268, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7368597388267517, + "num_tokens": 323193330.0, + "step": 12722 + }, + { + "epoch": 1.397210630353613, + "grad_norm": 2.259826898574829, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7227333188056946, + "num_tokens": 323218648.0, + "step": 12723 + }, + { + "epoch": 1.3973204480562267, + "grad_norm": 2.0124433040618896, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7181192636489868, + "num_tokens": 323247857.0, + "step": 12724 + }, + { + "epoch": 1.3974302657588402, + "grad_norm": 2.140099287033081, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7152482271194458, + "num_tokens": 323275963.0, + "step": 12725 + }, + { + "epoch": 1.397540083461454, + "grad_norm": 2.1069884300231934, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7113247513771057, + "num_tokens": 323306383.0, + "step": 12726 + }, + { + "epoch": 1.3976499011640677, + "grad_norm": 2.2851502895355225, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7127820253372192, + "num_tokens": 323332718.0, + "step": 12727 + }, + { + "epoch": 1.3977597188666813, + "grad_norm": 2.277282953262329, + "learning_rate": 1e-06, + "loss": 1.0343, + "mean_token_accuracy": 0.6857134699821472, + "num_tokens": 323357623.0, + "step": 12728 + }, + { + "epoch": 1.397869536569295, + "grad_norm": 2.079660415649414, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.6983700394630432, + "num_tokens": 323386670.0, + "step": 12729 + }, + { + "epoch": 1.3979793542719086, + "grad_norm": 2.0766806602478027, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7069246768951416, + "num_tokens": 323417611.0, + "step": 12730 + }, + { + "epoch": 1.3980891719745223, + "grad_norm": 2.6574063301086426, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7311978340148926, + "num_tokens": 323436856.0, + "step": 12731 + }, + { + "epoch": 1.398198989677136, + "grad_norm": 2.465709686279297, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7425141930580139, + "num_tokens": 323457767.0, + "step": 12732 + }, + { + "epoch": 1.3983088073797496, + "grad_norm": 2.2256758213043213, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7157220244407654, + "num_tokens": 323483817.0, + "step": 12733 + }, + { + "epoch": 1.3984186250823631, + "grad_norm": 2.181145191192627, + "learning_rate": 1e-06, + "loss": 0.8387, + "mean_token_accuracy": 0.7391923666000366, + "num_tokens": 323509386.0, + "step": 12734 + }, + { + "epoch": 1.398528442784977, + "grad_norm": 2.4332265853881836, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7050364017486572, + "num_tokens": 323531187.0, + "step": 12735 + }, + { + "epoch": 1.3986382604875907, + "grad_norm": 2.173814535140991, + "learning_rate": 1e-06, + "loss": 0.9995, + "mean_token_accuracy": 0.6909288167953491, + "num_tokens": 323559015.0, + "step": 12736 + }, + { + "epoch": 1.3987480781902042, + "grad_norm": 2.2793242931365967, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7135929465293884, + "num_tokens": 323584311.0, + "step": 12737 + }, + { + "epoch": 1.398857895892818, + "grad_norm": 2.204413890838623, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7176843881607056, + "num_tokens": 323610253.0, + "step": 12738 + }, + { + "epoch": 1.3989677135954315, + "grad_norm": 2.091672658920288, + "learning_rate": 1e-06, + "loss": 0.9472, + "mean_token_accuracy": 0.7095491886138916, + "num_tokens": 323637927.0, + "step": 12739 + }, + { + "epoch": 1.3990775312980452, + "grad_norm": 2.462350845336914, + "learning_rate": 1e-06, + "loss": 0.8729, + "mean_token_accuracy": 0.7215738892555237, + "num_tokens": 323660293.0, + "step": 12740 + }, + { + "epoch": 1.399187349000659, + "grad_norm": 2.3533449172973633, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.709173321723938, + "num_tokens": 323684326.0, + "step": 12741 + }, + { + "epoch": 1.3992971667032725, + "grad_norm": 2.2351694107055664, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7042906284332275, + "num_tokens": 323709474.0, + "step": 12742 + }, + { + "epoch": 1.3994069844058863, + "grad_norm": 2.358639717102051, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7211829423904419, + "num_tokens": 323733530.0, + "step": 12743 + }, + { + "epoch": 1.3995168021084998, + "grad_norm": 2.3086163997650146, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7103726267814636, + "num_tokens": 323758790.0, + "step": 12744 + }, + { + "epoch": 1.3996266198111136, + "grad_norm": 2.3078880310058594, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7201104164123535, + "num_tokens": 323784081.0, + "step": 12745 + }, + { + "epoch": 1.3997364375137273, + "grad_norm": 2.4130055904388428, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7085842490196228, + "num_tokens": 323806783.0, + "step": 12746 + }, + { + "epoch": 1.3998462552163409, + "grad_norm": 2.13542103767395, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7122955322265625, + "num_tokens": 323835499.0, + "step": 12747 + }, + { + "epoch": 1.3999560729189544, + "grad_norm": 2.1285037994384766, + "learning_rate": 1e-06, + "loss": 1.0102, + "mean_token_accuracy": 0.685166597366333, + "num_tokens": 323863252.0, + "step": 12748 + }, + { + "epoch": 1.4000658906215682, + "grad_norm": 2.3202171325683594, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7246556282043457, + "num_tokens": 323887600.0, + "step": 12749 + }, + { + "epoch": 1.400175708324182, + "grad_norm": 2.33564829826355, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7221308350563049, + "num_tokens": 323910147.0, + "step": 12750 + }, + { + "epoch": 1.4002855260267955, + "grad_norm": 2.317725419998169, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7155147790908813, + "num_tokens": 323933736.0, + "step": 12751 + }, + { + "epoch": 1.4003953437294092, + "grad_norm": 2.2910349369049072, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7393838167190552, + "num_tokens": 323957236.0, + "step": 12752 + }, + { + "epoch": 1.4005051614320227, + "grad_norm": 2.0501394271850586, + "learning_rate": 1e-06, + "loss": 0.8356, + "mean_token_accuracy": 0.7326305508613586, + "num_tokens": 323985494.0, + "step": 12753 + }, + { + "epoch": 1.4006149791346365, + "grad_norm": 2.103408098220825, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.712389349937439, + "num_tokens": 324011098.0, + "step": 12754 + }, + { + "epoch": 1.4007247968372503, + "grad_norm": 2.2962048053741455, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7448235750198364, + "num_tokens": 324034067.0, + "step": 12755 + }, + { + "epoch": 1.4008346145398638, + "grad_norm": 2.1741819381713867, + "learning_rate": 1e-06, + "loss": 0.9925, + "mean_token_accuracy": 0.6980783343315125, + "num_tokens": 324062119.0, + "step": 12756 + }, + { + "epoch": 1.4009444322424776, + "grad_norm": 2.6998450756073, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7143681645393372, + "num_tokens": 324081417.0, + "step": 12757 + }, + { + "epoch": 1.401054249945091, + "grad_norm": 2.2396531105041504, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6968256831169128, + "num_tokens": 324108404.0, + "step": 12758 + }, + { + "epoch": 1.4011640676477048, + "grad_norm": 2.275616407394409, + "learning_rate": 1e-06, + "loss": 0.8857, + "mean_token_accuracy": 0.722964346408844, + "num_tokens": 324133139.0, + "step": 12759 + }, + { + "epoch": 1.4012738853503186, + "grad_norm": 2.18986439704895, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7475032806396484, + "num_tokens": 324158325.0, + "step": 12760 + }, + { + "epoch": 1.4013837030529321, + "grad_norm": 2.154350519180298, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7064245939254761, + "num_tokens": 324186788.0, + "step": 12761 + }, + { + "epoch": 1.4014935207555457, + "grad_norm": 2.3841419219970703, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7173702120780945, + "num_tokens": 324208566.0, + "step": 12762 + }, + { + "epoch": 1.4016033384581594, + "grad_norm": 2.2112205028533936, + "learning_rate": 1e-06, + "loss": 0.992, + "mean_token_accuracy": 0.7016392946243286, + "num_tokens": 324235281.0, + "step": 12763 + }, + { + "epoch": 1.4017131561607732, + "grad_norm": 2.456679582595825, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.711798369884491, + "num_tokens": 324257692.0, + "step": 12764 + }, + { + "epoch": 1.4018229738633867, + "grad_norm": 2.1567039489746094, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7225993871688843, + "num_tokens": 324286407.0, + "step": 12765 + }, + { + "epoch": 1.4019327915660005, + "grad_norm": 2.510967969894409, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7237464189529419, + "num_tokens": 324306911.0, + "step": 12766 + }, + { + "epoch": 1.402042609268614, + "grad_norm": 2.3092453479766846, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7437975406646729, + "num_tokens": 324328549.0, + "step": 12767 + }, + { + "epoch": 1.4021524269712278, + "grad_norm": 2.3450441360473633, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7237134575843811, + "num_tokens": 324351870.0, + "step": 12768 + }, + { + "epoch": 1.4022622446738415, + "grad_norm": 2.1485791206359863, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7220568060874939, + "num_tokens": 324377384.0, + "step": 12769 + }, + { + "epoch": 1.402372062376455, + "grad_norm": 2.2621219158172607, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7239990234375, + "num_tokens": 324401632.0, + "step": 12770 + }, + { + "epoch": 1.4024818800790688, + "grad_norm": 2.211200475692749, + "learning_rate": 1e-06, + "loss": 0.9862, + "mean_token_accuracy": 0.7019063830375671, + "num_tokens": 324428228.0, + "step": 12771 + }, + { + "epoch": 1.4025916977816824, + "grad_norm": 2.446697235107422, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7317577600479126, + "num_tokens": 324450494.0, + "step": 12772 + }, + { + "epoch": 1.4027015154842961, + "grad_norm": 2.375962018966675, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7327098846435547, + "num_tokens": 324474464.0, + "step": 12773 + }, + { + "epoch": 1.4028113331869096, + "grad_norm": 1.9999890327453613, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7339768409729004, + "num_tokens": 324503504.0, + "step": 12774 + }, + { + "epoch": 1.4029211508895234, + "grad_norm": 2.2082667350769043, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7305715084075928, + "num_tokens": 324528212.0, + "step": 12775 + }, + { + "epoch": 1.403030968592137, + "grad_norm": 2.2976293563842773, + "learning_rate": 1e-06, + "loss": 0.9388, + "mean_token_accuracy": 0.7101581692695618, + "num_tokens": 324553230.0, + "step": 12776 + }, + { + "epoch": 1.4031407862947507, + "grad_norm": 2.232992172241211, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7419082522392273, + "num_tokens": 324577279.0, + "step": 12777 + }, + { + "epoch": 1.4032506039973645, + "grad_norm": 2.1677372455596924, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7171186804771423, + "num_tokens": 324603871.0, + "step": 12778 + }, + { + "epoch": 1.403360421699978, + "grad_norm": 2.0189132690429688, + "learning_rate": 1e-06, + "loss": 0.9948, + "mean_token_accuracy": 0.7038968801498413, + "num_tokens": 324634774.0, + "step": 12779 + }, + { + "epoch": 1.4034702394025917, + "grad_norm": 1.9320024251937866, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6881519556045532, + "num_tokens": 324668107.0, + "step": 12780 + }, + { + "epoch": 1.4035800571052053, + "grad_norm": 1.9755303859710693, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7200908660888672, + "num_tokens": 324698694.0, + "step": 12781 + }, + { + "epoch": 1.403689874807819, + "grad_norm": 2.0200798511505127, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.6975656151771545, + "num_tokens": 324728571.0, + "step": 12782 + }, + { + "epoch": 1.4037996925104328, + "grad_norm": 2.271169900894165, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7138127088546753, + "num_tokens": 324753408.0, + "step": 12783 + }, + { + "epoch": 1.4039095102130463, + "grad_norm": 2.172410011291504, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7000859975814819, + "num_tokens": 324781694.0, + "step": 12784 + }, + { + "epoch": 1.40401932791566, + "grad_norm": 2.118666172027588, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.6994026303291321, + "num_tokens": 324809472.0, + "step": 12785 + }, + { + "epoch": 1.4041291456182736, + "grad_norm": 2.0103039741516113, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.6993745565414429, + "num_tokens": 324840704.0, + "step": 12786 + }, + { + "epoch": 1.4042389633208874, + "grad_norm": 2.432703733444214, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.722911536693573, + "num_tokens": 324862649.0, + "step": 12787 + }, + { + "epoch": 1.404348781023501, + "grad_norm": 2.410595178604126, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7379332780838013, + "num_tokens": 324884661.0, + "step": 12788 + }, + { + "epoch": 1.4044585987261147, + "grad_norm": 2.0718624591827393, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7189410328865051, + "num_tokens": 324911956.0, + "step": 12789 + }, + { + "epoch": 1.4045684164287282, + "grad_norm": 2.1297264099121094, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7292463779449463, + "num_tokens": 324941181.0, + "step": 12790 + }, + { + "epoch": 1.404678234131342, + "grad_norm": 2.3153321743011475, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.713608980178833, + "num_tokens": 324964674.0, + "step": 12791 + }, + { + "epoch": 1.4047880518339557, + "grad_norm": 2.1075356006622314, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7128525972366333, + "num_tokens": 324993079.0, + "step": 12792 + }, + { + "epoch": 1.4048978695365693, + "grad_norm": 2.48343825340271, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7295976877212524, + "num_tokens": 325014762.0, + "step": 12793 + }, + { + "epoch": 1.405007687239183, + "grad_norm": 2.338399887084961, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.705925703048706, + "num_tokens": 325038217.0, + "step": 12794 + }, + { + "epoch": 1.4051175049417965, + "grad_norm": 2.2636947631835938, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7119534015655518, + "num_tokens": 325063992.0, + "step": 12795 + }, + { + "epoch": 1.4052273226444103, + "grad_norm": 2.2483537197113037, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.722339928150177, + "num_tokens": 325089512.0, + "step": 12796 + }, + { + "epoch": 1.405337140347024, + "grad_norm": 2.5415451526641846, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7289806008338928, + "num_tokens": 325109642.0, + "step": 12797 + }, + { + "epoch": 1.4054469580496376, + "grad_norm": 2.5080158710479736, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7202537655830383, + "num_tokens": 325130644.0, + "step": 12798 + }, + { + "epoch": 1.4055567757522511, + "grad_norm": 2.228512763977051, + "learning_rate": 1e-06, + "loss": 0.9861, + "mean_token_accuracy": 0.698592483997345, + "num_tokens": 325159295.0, + "step": 12799 + }, + { + "epoch": 1.4056665934548649, + "grad_norm": 2.459036111831665, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7267460823059082, + "num_tokens": 325181164.0, + "step": 12800 + }, + { + "epoch": 1.4057764111574786, + "grad_norm": 2.311492919921875, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7151843309402466, + "num_tokens": 325206693.0, + "step": 12801 + }, + { + "epoch": 1.4058862288600922, + "grad_norm": 2.1705896854400635, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7074689865112305, + "num_tokens": 325235040.0, + "step": 12802 + }, + { + "epoch": 1.405996046562706, + "grad_norm": 2.46917724609375, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.713466465473175, + "num_tokens": 325257283.0, + "step": 12803 + }, + { + "epoch": 1.4061058642653195, + "grad_norm": 2.169318199157715, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.714954137802124, + "num_tokens": 325285209.0, + "step": 12804 + }, + { + "epoch": 1.4062156819679332, + "grad_norm": 1.9889185428619385, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.7239525318145752, + "num_tokens": 325313297.0, + "step": 12805 + }, + { + "epoch": 1.406325499670547, + "grad_norm": 2.468545436859131, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7317303419113159, + "num_tokens": 325334914.0, + "step": 12806 + }, + { + "epoch": 1.4064353173731605, + "grad_norm": 2.5936696529388428, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7371869087219238, + "num_tokens": 325354935.0, + "step": 12807 + }, + { + "epoch": 1.4065451350757743, + "grad_norm": 2.0856077671051025, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7190738320350647, + "num_tokens": 325381817.0, + "step": 12808 + }, + { + "epoch": 1.4066549527783878, + "grad_norm": 2.2702462673187256, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7207640409469604, + "num_tokens": 325406226.0, + "step": 12809 + }, + { + "epoch": 1.4067647704810016, + "grad_norm": 2.6447229385375977, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7375571727752686, + "num_tokens": 325425801.0, + "step": 12810 + }, + { + "epoch": 1.4068745881836153, + "grad_norm": 2.224915027618408, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.6946659088134766, + "num_tokens": 325453251.0, + "step": 12811 + }, + { + "epoch": 1.4069844058862289, + "grad_norm": 2.5311238765716553, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7235366702079773, + "num_tokens": 325475987.0, + "step": 12812 + }, + { + "epoch": 1.4070942235888424, + "grad_norm": 2.3514280319213867, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7191857099533081, + "num_tokens": 325499045.0, + "step": 12813 + }, + { + "epoch": 1.4072040412914562, + "grad_norm": 2.306349277496338, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.7090628147125244, + "num_tokens": 325526148.0, + "step": 12814 + }, + { + "epoch": 1.40731385899407, + "grad_norm": 2.4060800075531006, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7237105369567871, + "num_tokens": 325549871.0, + "step": 12815 + }, + { + "epoch": 1.4074236766966834, + "grad_norm": 2.163409471511841, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7150942087173462, + "num_tokens": 325577981.0, + "step": 12816 + }, + { + "epoch": 1.4075334943992972, + "grad_norm": 2.406741142272949, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7164271473884583, + "num_tokens": 325601738.0, + "step": 12817 + }, + { + "epoch": 1.4076433121019107, + "grad_norm": 2.1206676959991455, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7225900888442993, + "num_tokens": 325629376.0, + "step": 12818 + }, + { + "epoch": 1.4077531298045245, + "grad_norm": 2.0575461387634277, + "learning_rate": 1e-06, + "loss": 0.9667, + "mean_token_accuracy": 0.7034741640090942, + "num_tokens": 325660424.0, + "step": 12819 + }, + { + "epoch": 1.4078629475071383, + "grad_norm": 2.40645170211792, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7229368686676025, + "num_tokens": 325682164.0, + "step": 12820 + }, + { + "epoch": 1.4079727652097518, + "grad_norm": 2.332951307296753, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7382336854934692, + "num_tokens": 325703878.0, + "step": 12821 + }, + { + "epoch": 1.4080825829123655, + "grad_norm": 2.1836838722229004, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.731561541557312, + "num_tokens": 325728304.0, + "step": 12822 + }, + { + "epoch": 1.408192400614979, + "grad_norm": 2.2288684844970703, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.743944525718689, + "num_tokens": 325751368.0, + "step": 12823 + }, + { + "epoch": 1.4083022183175928, + "grad_norm": 2.5359139442443848, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7421143054962158, + "num_tokens": 325772832.0, + "step": 12824 + }, + { + "epoch": 1.4084120360202066, + "grad_norm": 2.781198740005493, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.732913613319397, + "num_tokens": 325790754.0, + "step": 12825 + }, + { + "epoch": 1.4085218537228201, + "grad_norm": 2.2068071365356445, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7190436124801636, + "num_tokens": 325815634.0, + "step": 12826 + }, + { + "epoch": 1.4086316714254337, + "grad_norm": 2.439694404602051, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7117878198623657, + "num_tokens": 325838284.0, + "step": 12827 + }, + { + "epoch": 1.4087414891280474, + "grad_norm": 2.139538288116455, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7170032858848572, + "num_tokens": 325866538.0, + "step": 12828 + }, + { + "epoch": 1.4088513068306612, + "grad_norm": 1.9495137929916382, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7083331942558289, + "num_tokens": 325900544.0, + "step": 12829 + }, + { + "epoch": 1.4089611245332747, + "grad_norm": 2.3817827701568604, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7384876012802124, + "num_tokens": 325923849.0, + "step": 12830 + }, + { + "epoch": 1.4090709422358885, + "grad_norm": 2.2239303588867188, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7067857980728149, + "num_tokens": 325949948.0, + "step": 12831 + }, + { + "epoch": 1.409180759938502, + "grad_norm": 2.313460350036621, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7192214131355286, + "num_tokens": 325973917.0, + "step": 12832 + }, + { + "epoch": 1.4092905776411158, + "grad_norm": 2.0458149909973145, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7127237915992737, + "num_tokens": 326003314.0, + "step": 12833 + }, + { + "epoch": 1.4094003953437295, + "grad_norm": 2.569781541824341, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7233957052230835, + "num_tokens": 326022702.0, + "step": 12834 + }, + { + "epoch": 1.409510213046343, + "grad_norm": 2.3099424839019775, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7342420816421509, + "num_tokens": 326046331.0, + "step": 12835 + }, + { + "epoch": 1.4096200307489568, + "grad_norm": 2.1594789028167725, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7295659780502319, + "num_tokens": 326072284.0, + "step": 12836 + }, + { + "epoch": 1.4097298484515703, + "grad_norm": 2.4512295722961426, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7206825613975525, + "num_tokens": 326094192.0, + "step": 12837 + }, + { + "epoch": 1.409839666154184, + "grad_norm": 2.0347232818603516, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7186814546585083, + "num_tokens": 326124194.0, + "step": 12838 + }, + { + "epoch": 1.4099494838567976, + "grad_norm": 2.197131872177124, + "learning_rate": 1e-06, + "loss": 0.8273, + "mean_token_accuracy": 0.7477889060974121, + "num_tokens": 326148970.0, + "step": 12839 + }, + { + "epoch": 1.4100593015594114, + "grad_norm": 2.0532546043395996, + "learning_rate": 1e-06, + "loss": 1.0108, + "mean_token_accuracy": 0.7018142938613892, + "num_tokens": 326179411.0, + "step": 12840 + }, + { + "epoch": 1.410169119262025, + "grad_norm": 2.275644302368164, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7295844554901123, + "num_tokens": 326203740.0, + "step": 12841 + }, + { + "epoch": 1.4102789369646387, + "grad_norm": 2.0500142574310303, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7041733860969543, + "num_tokens": 326235635.0, + "step": 12842 + }, + { + "epoch": 1.4103887546672524, + "grad_norm": 2.1260786056518555, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6968792676925659, + "num_tokens": 326262613.0, + "step": 12843 + }, + { + "epoch": 1.410498572369866, + "grad_norm": 2.235546588897705, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7049611806869507, + "num_tokens": 326289218.0, + "step": 12844 + }, + { + "epoch": 1.4106083900724797, + "grad_norm": 2.049311637878418, + "learning_rate": 1e-06, + "loss": 0.9802, + "mean_token_accuracy": 0.7023088932037354, + "num_tokens": 326318683.0, + "step": 12845 + }, + { + "epoch": 1.4107182077750933, + "grad_norm": 2.2848124504089355, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7094231843948364, + "num_tokens": 326345251.0, + "step": 12846 + }, + { + "epoch": 1.410828025477707, + "grad_norm": 2.019624948501587, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7108738422393799, + "num_tokens": 326375237.0, + "step": 12847 + }, + { + "epoch": 1.4109378431803208, + "grad_norm": 2.3053343296051025, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7033611536026001, + "num_tokens": 326400896.0, + "step": 12848 + }, + { + "epoch": 1.4110476608829343, + "grad_norm": 2.1756904125213623, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7144168615341187, + "num_tokens": 326430119.0, + "step": 12849 + }, + { + "epoch": 1.4111574785855479, + "grad_norm": 2.128859043121338, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.697331428527832, + "num_tokens": 326459603.0, + "step": 12850 + }, + { + "epoch": 1.4112672962881616, + "grad_norm": 2.3200886249542236, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7214362621307373, + "num_tokens": 326484063.0, + "step": 12851 + }, + { + "epoch": 1.4113771139907754, + "grad_norm": 2.4964640140533447, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7328106760978699, + "num_tokens": 326508352.0, + "step": 12852 + }, + { + "epoch": 1.411486931693389, + "grad_norm": 2.2086238861083984, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7134991884231567, + "num_tokens": 326536266.0, + "step": 12853 + }, + { + "epoch": 1.4115967493960027, + "grad_norm": 2.5007150173187256, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7393063306808472, + "num_tokens": 326557033.0, + "step": 12854 + }, + { + "epoch": 1.4117065670986162, + "grad_norm": 2.099635362625122, + "learning_rate": 1e-06, + "loss": 0.9066, + "mean_token_accuracy": 0.7311328649520874, + "num_tokens": 326585152.0, + "step": 12855 + }, + { + "epoch": 1.41181638480123, + "grad_norm": 1.969022274017334, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7155579328536987, + "num_tokens": 326618179.0, + "step": 12856 + }, + { + "epoch": 1.4119262025038437, + "grad_norm": 1.9044023752212524, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7313611507415771, + "num_tokens": 326651813.0, + "step": 12857 + }, + { + "epoch": 1.4120360202064572, + "grad_norm": 2.4709908962249756, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7324346899986267, + "num_tokens": 326672660.0, + "step": 12858 + }, + { + "epoch": 1.412145837909071, + "grad_norm": 2.329495668411255, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7216333150863647, + "num_tokens": 326697641.0, + "step": 12859 + }, + { + "epoch": 1.4122556556116845, + "grad_norm": 2.0746066570281982, + "learning_rate": 1e-06, + "loss": 1.0011, + "mean_token_accuracy": 0.6889669299125671, + "num_tokens": 326729552.0, + "step": 12860 + }, + { + "epoch": 1.4123654733142983, + "grad_norm": 2.192248821258545, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7163376808166504, + "num_tokens": 326756916.0, + "step": 12861 + }, + { + "epoch": 1.412475291016912, + "grad_norm": 2.2945353984832764, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7088210582733154, + "num_tokens": 326781725.0, + "step": 12862 + }, + { + "epoch": 1.4125851087195256, + "grad_norm": 2.3272483348846436, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7140491604804993, + "num_tokens": 326804932.0, + "step": 12863 + }, + { + "epoch": 1.4126949264221391, + "grad_norm": 2.1882576942443848, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7151781916618347, + "num_tokens": 326831556.0, + "step": 12864 + }, + { + "epoch": 1.4128047441247529, + "grad_norm": 2.4786734580993652, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7222322225570679, + "num_tokens": 326854187.0, + "step": 12865 + }, + { + "epoch": 1.4129145618273666, + "grad_norm": 2.223240852355957, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7234708666801453, + "num_tokens": 326879108.0, + "step": 12866 + }, + { + "epoch": 1.4130243795299802, + "grad_norm": 2.3766117095947266, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.731785774230957, + "num_tokens": 326903827.0, + "step": 12867 + }, + { + "epoch": 1.413134197232594, + "grad_norm": 2.206129312515259, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7307623028755188, + "num_tokens": 326931708.0, + "step": 12868 + }, + { + "epoch": 1.4132440149352075, + "grad_norm": 1.8236768245697021, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.716848611831665, + "num_tokens": 326970641.0, + "step": 12869 + }, + { + "epoch": 1.4133538326378212, + "grad_norm": 1.9835482835769653, + "learning_rate": 1e-06, + "loss": 1.0186, + "mean_token_accuracy": 0.6879570484161377, + "num_tokens": 327005246.0, + "step": 12870 + }, + { + "epoch": 1.413463650340435, + "grad_norm": 2.3032076358795166, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.7078229188919067, + "num_tokens": 327030595.0, + "step": 12871 + }, + { + "epoch": 1.4135734680430485, + "grad_norm": 2.563622236251831, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7153111696243286, + "num_tokens": 327050847.0, + "step": 12872 + }, + { + "epoch": 1.4136832857456623, + "grad_norm": 2.2981560230255127, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7250488996505737, + "num_tokens": 327074986.0, + "step": 12873 + }, + { + "epoch": 1.4137931034482758, + "grad_norm": 2.0940513610839844, + "learning_rate": 1e-06, + "loss": 0.8286, + "mean_token_accuracy": 0.7453442811965942, + "num_tokens": 327102480.0, + "step": 12874 + }, + { + "epoch": 1.4139029211508896, + "grad_norm": 2.379493236541748, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7263205051422119, + "num_tokens": 327124729.0, + "step": 12875 + }, + { + "epoch": 1.4140127388535033, + "grad_norm": 2.3391425609588623, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7312825918197632, + "num_tokens": 327147149.0, + "step": 12876 + }, + { + "epoch": 1.4141225565561168, + "grad_norm": 1.9747897386550903, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7291304469108582, + "num_tokens": 327176682.0, + "step": 12877 + }, + { + "epoch": 1.4142323742587304, + "grad_norm": 2.3915774822235107, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7275722622871399, + "num_tokens": 327200768.0, + "step": 12878 + }, + { + "epoch": 1.4143421919613441, + "grad_norm": 2.3026413917541504, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7113919854164124, + "num_tokens": 327226835.0, + "step": 12879 + }, + { + "epoch": 1.414452009663958, + "grad_norm": 2.172041654586792, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.7325351238250732, + "num_tokens": 327252172.0, + "step": 12880 + }, + { + "epoch": 1.4145618273665714, + "grad_norm": 1.878757357597351, + "learning_rate": 1e-06, + "loss": 0.9751, + "mean_token_accuracy": 0.7012777924537659, + "num_tokens": 327285969.0, + "step": 12881 + }, + { + "epoch": 1.4146716450691852, + "grad_norm": 2.1456615924835205, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.712901771068573, + "num_tokens": 327312711.0, + "step": 12882 + }, + { + "epoch": 1.4147814627717987, + "grad_norm": 2.4216132164001465, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7208092212677002, + "num_tokens": 327336154.0, + "step": 12883 + }, + { + "epoch": 1.4148912804744125, + "grad_norm": 1.9649419784545898, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6992758512496948, + "num_tokens": 327369451.0, + "step": 12884 + }, + { + "epoch": 1.4150010981770262, + "grad_norm": 2.375157356262207, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7352691888809204, + "num_tokens": 327393056.0, + "step": 12885 + }, + { + "epoch": 1.4151109158796398, + "grad_norm": 2.086874485015869, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7048083543777466, + "num_tokens": 327424043.0, + "step": 12886 + }, + { + "epoch": 1.4152207335822535, + "grad_norm": 2.3540353775024414, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7151522040367126, + "num_tokens": 327448659.0, + "step": 12887 + }, + { + "epoch": 1.415330551284867, + "grad_norm": 2.0927255153656006, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7151138782501221, + "num_tokens": 327476821.0, + "step": 12888 + }, + { + "epoch": 1.4154403689874808, + "grad_norm": 2.106837511062622, + "learning_rate": 1e-06, + "loss": 0.7872, + "mean_token_accuracy": 0.7572724223136902, + "num_tokens": 327504695.0, + "step": 12889 + }, + { + "epoch": 1.4155501866900946, + "grad_norm": 2.252586841583252, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7322680354118347, + "num_tokens": 327529607.0, + "step": 12890 + }, + { + "epoch": 1.4156600043927081, + "grad_norm": 2.2497708797454834, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.724534273147583, + "num_tokens": 327556105.0, + "step": 12891 + }, + { + "epoch": 1.4157698220953217, + "grad_norm": 1.883728265762329, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7262738943099976, + "num_tokens": 327590291.0, + "step": 12892 + }, + { + "epoch": 1.4158796397979354, + "grad_norm": 2.3091988563537598, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7149189114570618, + "num_tokens": 327613866.0, + "step": 12893 + }, + { + "epoch": 1.4159894575005492, + "grad_norm": 2.167769432067871, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7125247120857239, + "num_tokens": 327641623.0, + "step": 12894 + }, + { + "epoch": 1.4160992752031627, + "grad_norm": 2.4832029342651367, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7066010236740112, + "num_tokens": 327664497.0, + "step": 12895 + }, + { + "epoch": 1.4162090929057765, + "grad_norm": 2.4357244968414307, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7140042185783386, + "num_tokens": 327686022.0, + "step": 12896 + }, + { + "epoch": 1.41631891060839, + "grad_norm": 2.1000375747680664, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7045766115188599, + "num_tokens": 327715684.0, + "step": 12897 + }, + { + "epoch": 1.4164287283110037, + "grad_norm": 2.557595729827881, + "learning_rate": 1e-06, + "loss": 0.821, + "mean_token_accuracy": 0.7392501831054688, + "num_tokens": 327734846.0, + "step": 12898 + }, + { + "epoch": 1.4165385460136175, + "grad_norm": 2.0082006454467773, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.70674729347229, + "num_tokens": 327768194.0, + "step": 12899 + }, + { + "epoch": 1.416648363716231, + "grad_norm": 2.2681972980499268, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7027926445007324, + "num_tokens": 327793064.0, + "step": 12900 + }, + { + "epoch": 1.4167581814188448, + "grad_norm": 2.1455931663513184, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7203387022018433, + "num_tokens": 327819853.0, + "step": 12901 + }, + { + "epoch": 1.4168679991214583, + "grad_norm": 1.9939849376678467, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7068034410476685, + "num_tokens": 327852797.0, + "step": 12902 + }, + { + "epoch": 1.416977816824072, + "grad_norm": 2.168891191482544, + "learning_rate": 1e-06, + "loss": 0.9464, + "mean_token_accuracy": 0.7046653032302856, + "num_tokens": 327880905.0, + "step": 12903 + }, + { + "epoch": 1.4170876345266856, + "grad_norm": 2.4074771404266357, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7322942018508911, + "num_tokens": 327901901.0, + "step": 12904 + }, + { + "epoch": 1.4171974522292994, + "grad_norm": 2.1026532649993896, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7222104072570801, + "num_tokens": 327928742.0, + "step": 12905 + }, + { + "epoch": 1.417307269931913, + "grad_norm": 2.2147417068481445, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7249307632446289, + "num_tokens": 327953567.0, + "step": 12906 + }, + { + "epoch": 1.4174170876345267, + "grad_norm": 2.118975877761841, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7453116774559021, + "num_tokens": 327981764.0, + "step": 12907 + }, + { + "epoch": 1.4175269053371404, + "grad_norm": 2.3151602745056152, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7419958114624023, + "num_tokens": 328002695.0, + "step": 12908 + }, + { + "epoch": 1.417636723039754, + "grad_norm": 2.3245818614959717, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7128880023956299, + "num_tokens": 328026568.0, + "step": 12909 + }, + { + "epoch": 1.4177465407423677, + "grad_norm": 1.951700210571289, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7161073684692383, + "num_tokens": 328060755.0, + "step": 12910 + }, + { + "epoch": 1.4178563584449813, + "grad_norm": 2.1323187351226807, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7184007167816162, + "num_tokens": 328091720.0, + "step": 12911 + }, + { + "epoch": 1.417966176147595, + "grad_norm": 2.2611241340637207, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7099955677986145, + "num_tokens": 328116360.0, + "step": 12912 + }, + { + "epoch": 1.4180759938502088, + "grad_norm": 2.497849702835083, + "learning_rate": 1e-06, + "loss": 0.9746, + "mean_token_accuracy": 0.7155902981758118, + "num_tokens": 328136479.0, + "step": 12913 + }, + { + "epoch": 1.4181858115528223, + "grad_norm": 2.213531732559204, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7116421461105347, + "num_tokens": 328165895.0, + "step": 12914 + }, + { + "epoch": 1.4182956292554358, + "grad_norm": 2.132138967514038, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7091639041900635, + "num_tokens": 328193490.0, + "step": 12915 + }, + { + "epoch": 1.4184054469580496, + "grad_norm": 2.278322458267212, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7243064641952515, + "num_tokens": 328217640.0, + "step": 12916 + }, + { + "epoch": 1.4185152646606634, + "grad_norm": 2.2858939170837402, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7318457365036011, + "num_tokens": 328242212.0, + "step": 12917 + }, + { + "epoch": 1.418625082363277, + "grad_norm": 2.297908306121826, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7164720296859741, + "num_tokens": 328268444.0, + "step": 12918 + }, + { + "epoch": 1.4187349000658906, + "grad_norm": 2.1817338466644287, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7258580923080444, + "num_tokens": 328295026.0, + "step": 12919 + }, + { + "epoch": 1.4188447177685042, + "grad_norm": 2.281346559524536, + "learning_rate": 1e-06, + "loss": 0.8101, + "mean_token_accuracy": 0.7381751537322998, + "num_tokens": 328319957.0, + "step": 12920 + }, + { + "epoch": 1.418954535471118, + "grad_norm": 2.0946497917175293, + "learning_rate": 1e-06, + "loss": 0.9311, + "mean_token_accuracy": 0.7103555798530579, + "num_tokens": 328347964.0, + "step": 12921 + }, + { + "epoch": 1.4190643531737317, + "grad_norm": 2.066739082336426, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7342808246612549, + "num_tokens": 328374965.0, + "step": 12922 + }, + { + "epoch": 1.4191741708763452, + "grad_norm": 2.227356433868408, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7237488031387329, + "num_tokens": 328399468.0, + "step": 12923 + }, + { + "epoch": 1.419283988578959, + "grad_norm": 2.2955856323242188, + "learning_rate": 1e-06, + "loss": 1.014, + "mean_token_accuracy": 0.6864535212516785, + "num_tokens": 328425882.0, + "step": 12924 + }, + { + "epoch": 1.4193938062815725, + "grad_norm": 2.1116771697998047, + "learning_rate": 1e-06, + "loss": 0.7989, + "mean_token_accuracy": 0.7373948097229004, + "num_tokens": 328451970.0, + "step": 12925 + }, + { + "epoch": 1.4195036239841863, + "grad_norm": 2.3637492656707764, + "learning_rate": 1e-06, + "loss": 0.9053, + "mean_token_accuracy": 0.7219769954681396, + "num_tokens": 328474907.0, + "step": 12926 + }, + { + "epoch": 1.4196134416868, + "grad_norm": 2.0602385997772217, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.718992292881012, + "num_tokens": 328503964.0, + "step": 12927 + }, + { + "epoch": 1.4197232593894136, + "grad_norm": 2.150799036026001, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7239822149276733, + "num_tokens": 328530654.0, + "step": 12928 + }, + { + "epoch": 1.419833077092027, + "grad_norm": 2.1754093170166016, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7160744071006775, + "num_tokens": 328556530.0, + "step": 12929 + }, + { + "epoch": 1.4199428947946409, + "grad_norm": 2.0762088298797607, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7122972011566162, + "num_tokens": 328584771.0, + "step": 12930 + }, + { + "epoch": 1.4200527124972546, + "grad_norm": 2.1737425327301025, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7299250364303589, + "num_tokens": 328609686.0, + "step": 12931 + }, + { + "epoch": 1.4201625301998682, + "grad_norm": 2.5000722408294678, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7229361534118652, + "num_tokens": 328629892.0, + "step": 12932 + }, + { + "epoch": 1.420272347902482, + "grad_norm": 2.0572116374969482, + "learning_rate": 1e-06, + "loss": 1.0339, + "mean_token_accuracy": 0.6815477609634399, + "num_tokens": 328661016.0, + "step": 12933 + }, + { + "epoch": 1.4203821656050954, + "grad_norm": 2.227818012237549, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7150865793228149, + "num_tokens": 328688182.0, + "step": 12934 + }, + { + "epoch": 1.4204919833077092, + "grad_norm": 2.4709911346435547, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7192137837409973, + "num_tokens": 328709383.0, + "step": 12935 + }, + { + "epoch": 1.420601801010323, + "grad_norm": 2.425994634628296, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7337710857391357, + "num_tokens": 328731053.0, + "step": 12936 + }, + { + "epoch": 1.4207116187129365, + "grad_norm": 2.223268985748291, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7139462232589722, + "num_tokens": 328760684.0, + "step": 12937 + }, + { + "epoch": 1.4208214364155503, + "grad_norm": 2.9754321575164795, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7118266820907593, + "num_tokens": 328778279.0, + "step": 12938 + }, + { + "epoch": 1.4209312541181638, + "grad_norm": 2.5858490467071533, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7123642563819885, + "num_tokens": 328799910.0, + "step": 12939 + }, + { + "epoch": 1.4210410718207775, + "grad_norm": 2.139737367630005, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7039816975593567, + "num_tokens": 328826353.0, + "step": 12940 + }, + { + "epoch": 1.4211508895233913, + "grad_norm": 2.1577656269073486, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7018393278121948, + "num_tokens": 328855483.0, + "step": 12941 + }, + { + "epoch": 1.4212607072260048, + "grad_norm": 2.339416980743408, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7162222862243652, + "num_tokens": 328879374.0, + "step": 12942 + }, + { + "epoch": 1.4213705249286184, + "grad_norm": 2.2025773525238037, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.726582407951355, + "num_tokens": 328906294.0, + "step": 12943 + }, + { + "epoch": 1.4214803426312321, + "grad_norm": 2.123375654220581, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7402989864349365, + "num_tokens": 328933692.0, + "step": 12944 + }, + { + "epoch": 1.4215901603338459, + "grad_norm": 2.2945051193237305, + "learning_rate": 1e-06, + "loss": 0.7342, + "mean_token_accuracy": 0.7580258846282959, + "num_tokens": 328955802.0, + "step": 12945 + }, + { + "epoch": 1.4216999780364594, + "grad_norm": 2.4468791484832764, + "learning_rate": 1e-06, + "loss": 1.0051, + "mean_token_accuracy": 0.6949501037597656, + "num_tokens": 328979506.0, + "step": 12946 + }, + { + "epoch": 1.4218097957390732, + "grad_norm": 2.396595001220703, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7138345241546631, + "num_tokens": 329003582.0, + "step": 12947 + }, + { + "epoch": 1.4219196134416867, + "grad_norm": 2.4387292861938477, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7358343601226807, + "num_tokens": 329025716.0, + "step": 12948 + }, + { + "epoch": 1.4220294311443005, + "grad_norm": 2.3137481212615967, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7054319381713867, + "num_tokens": 329052918.0, + "step": 12949 + }, + { + "epoch": 1.4221392488469142, + "grad_norm": 2.471097946166992, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7031118869781494, + "num_tokens": 329075633.0, + "step": 12950 + }, + { + "epoch": 1.4222490665495278, + "grad_norm": 2.185314893722534, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7239301204681396, + "num_tokens": 329100283.0, + "step": 12951 + }, + { + "epoch": 1.4223588842521415, + "grad_norm": 2.4837141036987305, + "learning_rate": 1e-06, + "loss": 0.8676, + "mean_token_accuracy": 0.7257819175720215, + "num_tokens": 329120580.0, + "step": 12952 + }, + { + "epoch": 1.422468701954755, + "grad_norm": 2.377148389816284, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7139493823051453, + "num_tokens": 329144578.0, + "step": 12953 + }, + { + "epoch": 1.4225785196573688, + "grad_norm": 2.654392957687378, + "learning_rate": 1e-06, + "loss": 0.8057, + "mean_token_accuracy": 0.7407345771789551, + "num_tokens": 329162458.0, + "step": 12954 + }, + { + "epoch": 1.4226883373599823, + "grad_norm": 2.473278284072876, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7077871561050415, + "num_tokens": 329184932.0, + "step": 12955 + }, + { + "epoch": 1.422798155062596, + "grad_norm": 2.27447247505188, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7129960060119629, + "num_tokens": 329212821.0, + "step": 12956 + }, + { + "epoch": 1.4229079727652096, + "grad_norm": 2.3141636848449707, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7278798222541809, + "num_tokens": 329237297.0, + "step": 12957 + }, + { + "epoch": 1.4230177904678234, + "grad_norm": 2.2970871925354004, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7051980495452881, + "num_tokens": 329262227.0, + "step": 12958 + }, + { + "epoch": 1.4231276081704372, + "grad_norm": 2.421985626220703, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7211848497390747, + "num_tokens": 329284721.0, + "step": 12959 + }, + { + "epoch": 1.4232374258730507, + "grad_norm": 2.464031219482422, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7158129811286926, + "num_tokens": 329306721.0, + "step": 12960 + }, + { + "epoch": 1.4233472435756644, + "grad_norm": 2.195643663406372, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7166723012924194, + "num_tokens": 329333339.0, + "step": 12961 + }, + { + "epoch": 1.423457061278278, + "grad_norm": 2.3115696907043457, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7288142442703247, + "num_tokens": 329357902.0, + "step": 12962 + }, + { + "epoch": 1.4235668789808917, + "grad_norm": 2.3907101154327393, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7330549955368042, + "num_tokens": 329380274.0, + "step": 12963 + }, + { + "epoch": 1.4236766966835055, + "grad_norm": 2.2222037315368652, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7310289740562439, + "num_tokens": 329404874.0, + "step": 12964 + }, + { + "epoch": 1.423786514386119, + "grad_norm": 2.436392068862915, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7114817500114441, + "num_tokens": 329428877.0, + "step": 12965 + }, + { + "epoch": 1.4238963320887328, + "grad_norm": 2.483520030975342, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7430723905563354, + "num_tokens": 329450406.0, + "step": 12966 + }, + { + "epoch": 1.4240061497913463, + "grad_norm": 2.319963216781616, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7234083414077759, + "num_tokens": 329475393.0, + "step": 12967 + }, + { + "epoch": 1.42411596749396, + "grad_norm": 2.2337257862091064, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7130147218704224, + "num_tokens": 329502995.0, + "step": 12968 + }, + { + "epoch": 1.4242257851965736, + "grad_norm": 2.3009605407714844, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7112144231796265, + "num_tokens": 329527554.0, + "step": 12969 + }, + { + "epoch": 1.4243356028991874, + "grad_norm": 2.1585676670074463, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7005016207695007, + "num_tokens": 329553843.0, + "step": 12970 + }, + { + "epoch": 1.424445420601801, + "grad_norm": 2.505528211593628, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7015503644943237, + "num_tokens": 329576246.0, + "step": 12971 + }, + { + "epoch": 1.4245552383044147, + "grad_norm": 2.2969400882720947, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7188254594802856, + "num_tokens": 329599487.0, + "step": 12972 + }, + { + "epoch": 1.4246650560070284, + "grad_norm": 2.443159818649292, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7319135665893555, + "num_tokens": 329621213.0, + "step": 12973 + }, + { + "epoch": 1.424774873709642, + "grad_norm": 1.8664140701293945, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7180408239364624, + "num_tokens": 329654211.0, + "step": 12974 + }, + { + "epoch": 1.4248846914122557, + "grad_norm": 2.4661216735839844, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7263273596763611, + "num_tokens": 329678910.0, + "step": 12975 + }, + { + "epoch": 1.4249945091148692, + "grad_norm": 2.5790512561798096, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7046469449996948, + "num_tokens": 329702559.0, + "step": 12976 + }, + { + "epoch": 1.425104326817483, + "grad_norm": 2.3307077884674072, + "learning_rate": 1e-06, + "loss": 0.9702, + "mean_token_accuracy": 0.7002905011177063, + "num_tokens": 329727041.0, + "step": 12977 + }, + { + "epoch": 1.4252141445200968, + "grad_norm": 2.3519041538238525, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7108311653137207, + "num_tokens": 329751075.0, + "step": 12978 + }, + { + "epoch": 1.4253239622227103, + "grad_norm": 2.599684238433838, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7164275646209717, + "num_tokens": 329771180.0, + "step": 12979 + }, + { + "epoch": 1.4254337799253238, + "grad_norm": 2.4859540462493896, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.729146420955658, + "num_tokens": 329793055.0, + "step": 12980 + }, + { + "epoch": 1.4255435976279376, + "grad_norm": 2.4922432899475098, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7138490676879883, + "num_tokens": 329814420.0, + "step": 12981 + }, + { + "epoch": 1.4256534153305513, + "grad_norm": 2.599491596221924, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7331715822219849, + "num_tokens": 329835089.0, + "step": 12982 + }, + { + "epoch": 1.4257632330331649, + "grad_norm": 2.189901113510132, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7242011427879333, + "num_tokens": 329861719.0, + "step": 12983 + }, + { + "epoch": 1.4258730507357786, + "grad_norm": 2.205744981765747, + "learning_rate": 1e-06, + "loss": 0.8824, + "mean_token_accuracy": 0.7233155965805054, + "num_tokens": 329888722.0, + "step": 12984 + }, + { + "epoch": 1.4259828684383922, + "grad_norm": 2.22749400138855, + "learning_rate": 1e-06, + "loss": 0.9818, + "mean_token_accuracy": 0.697769284248352, + "num_tokens": 329915823.0, + "step": 12985 + }, + { + "epoch": 1.426092686141006, + "grad_norm": 2.0711236000061035, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7092875242233276, + "num_tokens": 329946841.0, + "step": 12986 + }, + { + "epoch": 1.4262025038436197, + "grad_norm": 2.29435396194458, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7370643019676208, + "num_tokens": 329971475.0, + "step": 12987 + }, + { + "epoch": 1.4263123215462332, + "grad_norm": 2.2322354316711426, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7235417366027832, + "num_tokens": 329997275.0, + "step": 12988 + }, + { + "epoch": 1.426422139248847, + "grad_norm": 2.5496127605438232, + "learning_rate": 1e-06, + "loss": 0.856, + "mean_token_accuracy": 0.7294700145721436, + "num_tokens": 330017898.0, + "step": 12989 + }, + { + "epoch": 1.4265319569514605, + "grad_norm": 2.0946946144104004, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.7017451524734497, + "num_tokens": 330047358.0, + "step": 12990 + }, + { + "epoch": 1.4266417746540743, + "grad_norm": 2.4023871421813965, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.6929776072502136, + "num_tokens": 330072439.0, + "step": 12991 + }, + { + "epoch": 1.426751592356688, + "grad_norm": 2.3689446449279785, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.730060338973999, + "num_tokens": 330096527.0, + "step": 12992 + }, + { + "epoch": 1.4268614100593016, + "grad_norm": 2.248509407043457, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7247220277786255, + "num_tokens": 330121055.0, + "step": 12993 + }, + { + "epoch": 1.426971227761915, + "grad_norm": 2.2061333656311035, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.710663914680481, + "num_tokens": 330146471.0, + "step": 12994 + }, + { + "epoch": 1.4270810454645289, + "grad_norm": 2.656703233718872, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7400984764099121, + "num_tokens": 330163604.0, + "step": 12995 + }, + { + "epoch": 1.4271908631671426, + "grad_norm": 2.431859016418457, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7251054048538208, + "num_tokens": 330186085.0, + "step": 12996 + }, + { + "epoch": 1.4273006808697561, + "grad_norm": 2.3004238605499268, + "learning_rate": 1e-06, + "loss": 1.0396, + "mean_token_accuracy": 0.6761438846588135, + "num_tokens": 330213668.0, + "step": 12997 + }, + { + "epoch": 1.42741049857237, + "grad_norm": 2.1658804416656494, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7197065949440002, + "num_tokens": 330243310.0, + "step": 12998 + }, + { + "epoch": 1.4275203162749834, + "grad_norm": 2.427183151245117, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7060939073562622, + "num_tokens": 330266787.0, + "step": 12999 + }, + { + "epoch": 1.4276301339775972, + "grad_norm": 2.2679712772369385, + "learning_rate": 1e-06, + "loss": 0.8538, + "mean_token_accuracy": 0.7297389507293701, + "num_tokens": 330290784.0, + "step": 13000 + }, + { + "epoch": 1.427739951680211, + "grad_norm": 2.0977718830108643, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7159982919692993, + "num_tokens": 330318307.0, + "step": 13001 + }, + { + "epoch": 1.4278497693828245, + "grad_norm": 2.2807486057281494, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7308189272880554, + "num_tokens": 330341457.0, + "step": 13002 + }, + { + "epoch": 1.4279595870854382, + "grad_norm": 2.2696545124053955, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7293946146965027, + "num_tokens": 330364898.0, + "step": 13003 + }, + { + "epoch": 1.4280694047880518, + "grad_norm": 2.5236191749572754, + "learning_rate": 1e-06, + "loss": 0.9777, + "mean_token_accuracy": 0.6985998153686523, + "num_tokens": 330387468.0, + "step": 13004 + }, + { + "epoch": 1.4281792224906655, + "grad_norm": 2.089092969894409, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7067092657089233, + "num_tokens": 330415976.0, + "step": 13005 + }, + { + "epoch": 1.4282890401932793, + "grad_norm": 2.3767528533935547, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.6998048424720764, + "num_tokens": 330440274.0, + "step": 13006 + }, + { + "epoch": 1.4283988578958928, + "grad_norm": 1.995283842086792, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7121055126190186, + "num_tokens": 330470577.0, + "step": 13007 + }, + { + "epoch": 1.4285086755985064, + "grad_norm": 2.387850761413574, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7216272950172424, + "num_tokens": 330492786.0, + "step": 13008 + }, + { + "epoch": 1.4286184933011201, + "grad_norm": 2.135167360305786, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7140785455703735, + "num_tokens": 330520819.0, + "step": 13009 + }, + { + "epoch": 1.4287283110037339, + "grad_norm": 2.2513179779052734, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.6999251842498779, + "num_tokens": 330546937.0, + "step": 13010 + }, + { + "epoch": 1.4288381287063474, + "grad_norm": 2.191167116165161, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7247285842895508, + "num_tokens": 330573280.0, + "step": 13011 + }, + { + "epoch": 1.4289479464089612, + "grad_norm": 2.2116761207580566, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7122492790222168, + "num_tokens": 330598739.0, + "step": 13012 + }, + { + "epoch": 1.4290577641115747, + "grad_norm": 2.137408494949341, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6968309879302979, + "num_tokens": 330628700.0, + "step": 13013 + }, + { + "epoch": 1.4291675818141885, + "grad_norm": 2.3328089714050293, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.7052825093269348, + "num_tokens": 330653884.0, + "step": 13014 + }, + { + "epoch": 1.4292773995168022, + "grad_norm": 2.3464090824127197, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7145085334777832, + "num_tokens": 330676669.0, + "step": 13015 + }, + { + "epoch": 1.4293872172194158, + "grad_norm": 2.3288168907165527, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.71776282787323, + "num_tokens": 330701976.0, + "step": 13016 + }, + { + "epoch": 1.4294970349220295, + "grad_norm": 2.1976170539855957, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7081520557403564, + "num_tokens": 330730466.0, + "step": 13017 + }, + { + "epoch": 1.429606852624643, + "grad_norm": 2.557424306869507, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7336438298225403, + "num_tokens": 330751059.0, + "step": 13018 + }, + { + "epoch": 1.4297166703272568, + "grad_norm": 2.453178644180298, + "learning_rate": 1e-06, + "loss": 0.7941, + "mean_token_accuracy": 0.7441424131393433, + "num_tokens": 330771823.0, + "step": 13019 + }, + { + "epoch": 1.4298264880298703, + "grad_norm": 2.1022729873657227, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7181843519210815, + "num_tokens": 330800814.0, + "step": 13020 + }, + { + "epoch": 1.429936305732484, + "grad_norm": 2.4799234867095947, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7164771556854248, + "num_tokens": 330823630.0, + "step": 13021 + }, + { + "epoch": 1.4300461234350976, + "grad_norm": 2.068647861480713, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7214261293411255, + "num_tokens": 330853872.0, + "step": 13022 + }, + { + "epoch": 1.4301559411377114, + "grad_norm": 2.3380422592163086, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7034755349159241, + "num_tokens": 330879783.0, + "step": 13023 + }, + { + "epoch": 1.4302657588403251, + "grad_norm": 2.471038341522217, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.7564030885696411, + "num_tokens": 330899147.0, + "step": 13024 + }, + { + "epoch": 1.4303755765429387, + "grad_norm": 2.4088234901428223, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7215961813926697, + "num_tokens": 330922640.0, + "step": 13025 + }, + { + "epoch": 1.4304853942455524, + "grad_norm": 2.186279058456421, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7099364995956421, + "num_tokens": 330950466.0, + "step": 13026 + }, + { + "epoch": 1.430595211948166, + "grad_norm": 2.269037961959839, + "learning_rate": 1e-06, + "loss": 0.8457, + "mean_token_accuracy": 0.7358647584915161, + "num_tokens": 330974387.0, + "step": 13027 + }, + { + "epoch": 1.4307050296507797, + "grad_norm": 2.165883779525757, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.726664662361145, + "num_tokens": 331001628.0, + "step": 13028 + }, + { + "epoch": 1.4308148473533935, + "grad_norm": 2.052745819091797, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6965667009353638, + "num_tokens": 331034226.0, + "step": 13029 + }, + { + "epoch": 1.430924665056007, + "grad_norm": 2.0471935272216797, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7054876089096069, + "num_tokens": 331063911.0, + "step": 13030 + }, + { + "epoch": 1.4310344827586206, + "grad_norm": 2.403930902481079, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7177157998085022, + "num_tokens": 331086417.0, + "step": 13031 + }, + { + "epoch": 1.4311443004612343, + "grad_norm": 2.1040470600128174, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7336388826370239, + "num_tokens": 331113810.0, + "step": 13032 + }, + { + "epoch": 1.431254118163848, + "grad_norm": 2.1486217975616455, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7059473991394043, + "num_tokens": 331140897.0, + "step": 13033 + }, + { + "epoch": 1.4313639358664616, + "grad_norm": 2.25289249420166, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7072135210037231, + "num_tokens": 331164761.0, + "step": 13034 + }, + { + "epoch": 1.4314737535690754, + "grad_norm": 2.5498104095458984, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7229423522949219, + "num_tokens": 331187921.0, + "step": 13035 + }, + { + "epoch": 1.431583571271689, + "grad_norm": 2.439444065093994, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7240933179855347, + "num_tokens": 331210942.0, + "step": 13036 + }, + { + "epoch": 1.4316933889743026, + "grad_norm": 2.090195417404175, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7172736525535583, + "num_tokens": 331239699.0, + "step": 13037 + }, + { + "epoch": 1.4318032066769164, + "grad_norm": 2.0962440967559814, + "learning_rate": 1e-06, + "loss": 0.8678, + "mean_token_accuracy": 0.7383368015289307, + "num_tokens": 331268045.0, + "step": 13038 + }, + { + "epoch": 1.43191302437953, + "grad_norm": 2.2268404960632324, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6986508369445801, + "num_tokens": 331294342.0, + "step": 13039 + }, + { + "epoch": 1.4320228420821437, + "grad_norm": 2.094546318054199, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7308603525161743, + "num_tokens": 331322198.0, + "step": 13040 + }, + { + "epoch": 1.4321326597847572, + "grad_norm": 2.2547786235809326, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6948853731155396, + "num_tokens": 331350868.0, + "step": 13041 + }, + { + "epoch": 1.432242477487371, + "grad_norm": 2.199033498764038, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7198443412780762, + "num_tokens": 331377883.0, + "step": 13042 + }, + { + "epoch": 1.4323522951899847, + "grad_norm": 2.116528034210205, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7369168996810913, + "num_tokens": 331403647.0, + "step": 13043 + }, + { + "epoch": 1.4324621128925983, + "grad_norm": 2.327714204788208, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7148985862731934, + "num_tokens": 331427389.0, + "step": 13044 + }, + { + "epoch": 1.4325719305952118, + "grad_norm": 2.302182197570801, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7239233255386353, + "num_tokens": 331449743.0, + "step": 13045 + }, + { + "epoch": 1.4326817482978256, + "grad_norm": 1.79323410987854, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7188535928726196, + "num_tokens": 331484772.0, + "step": 13046 + }, + { + "epoch": 1.4327915660004393, + "grad_norm": 2.1464896202087402, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7023801803588867, + "num_tokens": 331511218.0, + "step": 13047 + }, + { + "epoch": 1.4329013837030529, + "grad_norm": 2.3012213706970215, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7334970235824585, + "num_tokens": 331536136.0, + "step": 13048 + }, + { + "epoch": 1.4330112014056666, + "grad_norm": 2.593876838684082, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7160049080848694, + "num_tokens": 331557348.0, + "step": 13049 + }, + { + "epoch": 1.4331210191082802, + "grad_norm": 1.8967604637145996, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7240990996360779, + "num_tokens": 331591814.0, + "step": 13050 + }, + { + "epoch": 1.433230836810894, + "grad_norm": 2.296919107437134, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7237786054611206, + "num_tokens": 331616167.0, + "step": 13051 + }, + { + "epoch": 1.4333406545135077, + "grad_norm": 2.242704153060913, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7128733396530151, + "num_tokens": 331640888.0, + "step": 13052 + }, + { + "epoch": 1.4334504722161212, + "grad_norm": 2.3930647373199463, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7233061790466309, + "num_tokens": 331664376.0, + "step": 13053 + }, + { + "epoch": 1.433560289918735, + "grad_norm": 2.6245357990264893, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7384856343269348, + "num_tokens": 331682508.0, + "step": 13054 + }, + { + "epoch": 1.4336701076213485, + "grad_norm": 2.273308753967285, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.731345534324646, + "num_tokens": 331706612.0, + "step": 13055 + }, + { + "epoch": 1.4337799253239623, + "grad_norm": 2.3297579288482666, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7094365954399109, + "num_tokens": 331731246.0, + "step": 13056 + }, + { + "epoch": 1.433889743026576, + "grad_norm": 2.26373028755188, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.725566029548645, + "num_tokens": 331755509.0, + "step": 13057 + }, + { + "epoch": 1.4339995607291895, + "grad_norm": 2.406219959259033, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.71017986536026, + "num_tokens": 331778403.0, + "step": 13058 + }, + { + "epoch": 1.434109378431803, + "grad_norm": 2.299187183380127, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7155686020851135, + "num_tokens": 331803101.0, + "step": 13059 + }, + { + "epoch": 1.4342191961344168, + "grad_norm": 2.2679660320281982, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7083176374435425, + "num_tokens": 331831471.0, + "step": 13060 + }, + { + "epoch": 1.4343290138370306, + "grad_norm": 2.437429189682007, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7177149057388306, + "num_tokens": 331855869.0, + "step": 13061 + }, + { + "epoch": 1.4344388315396441, + "grad_norm": 2.4719960689544678, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7258037328720093, + "num_tokens": 331879993.0, + "step": 13062 + }, + { + "epoch": 1.434548649242258, + "grad_norm": 2.3720555305480957, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7024462223052979, + "num_tokens": 331905879.0, + "step": 13063 + }, + { + "epoch": 1.4346584669448714, + "grad_norm": 2.5219390392303467, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7343189120292664, + "num_tokens": 331924276.0, + "step": 13064 + }, + { + "epoch": 1.4347682846474852, + "grad_norm": 2.3207051753997803, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7168511152267456, + "num_tokens": 331948634.0, + "step": 13065 + }, + { + "epoch": 1.434878102350099, + "grad_norm": 2.300259590148926, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7222973108291626, + "num_tokens": 331973484.0, + "step": 13066 + }, + { + "epoch": 1.4349879200527125, + "grad_norm": 2.326300621032715, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7080858945846558, + "num_tokens": 331996543.0, + "step": 13067 + }, + { + "epoch": 1.4350977377553262, + "grad_norm": 2.275897264480591, + "learning_rate": 1e-06, + "loss": 0.8901, + "mean_token_accuracy": 0.7191545367240906, + "num_tokens": 332021081.0, + "step": 13068 + }, + { + "epoch": 1.4352075554579398, + "grad_norm": 2.8850460052490234, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7463867664337158, + "num_tokens": 332039416.0, + "step": 13069 + }, + { + "epoch": 1.4353173731605535, + "grad_norm": 2.047699451446533, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7315163016319275, + "num_tokens": 332068655.0, + "step": 13070 + }, + { + "epoch": 1.4354271908631673, + "grad_norm": 2.1173360347747803, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7127283811569214, + "num_tokens": 332098982.0, + "step": 13071 + }, + { + "epoch": 1.4355370085657808, + "grad_norm": 2.4337079524993896, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7203824520111084, + "num_tokens": 332122274.0, + "step": 13072 + }, + { + "epoch": 1.4356468262683943, + "grad_norm": 2.488511562347412, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7198969125747681, + "num_tokens": 332145270.0, + "step": 13073 + }, + { + "epoch": 1.435756643971008, + "grad_norm": 2.3738701343536377, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7337263822555542, + "num_tokens": 332165525.0, + "step": 13074 + }, + { + "epoch": 1.4358664616736219, + "grad_norm": 2.20259428024292, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7089014053344727, + "num_tokens": 332192362.0, + "step": 13075 + }, + { + "epoch": 1.4359762793762354, + "grad_norm": 1.9870305061340332, + "learning_rate": 1e-06, + "loss": 0.855, + "mean_token_accuracy": 0.7324857711791992, + "num_tokens": 332223923.0, + "step": 13076 + }, + { + "epoch": 1.4360860970788492, + "grad_norm": 2.0519890785217285, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6899009346961975, + "num_tokens": 332254523.0, + "step": 13077 + }, + { + "epoch": 1.4361959147814627, + "grad_norm": 2.5845165252685547, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7305009961128235, + "num_tokens": 332275653.0, + "step": 13078 + }, + { + "epoch": 1.4363057324840764, + "grad_norm": 2.1882705688476562, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7063099145889282, + "num_tokens": 332301990.0, + "step": 13079 + }, + { + "epoch": 1.4364155501866902, + "grad_norm": 2.49868106842041, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7217617630958557, + "num_tokens": 332324570.0, + "step": 13080 + }, + { + "epoch": 1.4365253678893037, + "grad_norm": 2.7682645320892334, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.731160581111908, + "num_tokens": 332343338.0, + "step": 13081 + }, + { + "epoch": 1.4366351855919175, + "grad_norm": 2.585960626602173, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7253067493438721, + "num_tokens": 332362364.0, + "step": 13082 + }, + { + "epoch": 1.436745003294531, + "grad_norm": 2.370471715927124, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7152444124221802, + "num_tokens": 332387506.0, + "step": 13083 + }, + { + "epoch": 1.4368548209971448, + "grad_norm": 2.1480600833892822, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7070181369781494, + "num_tokens": 332415279.0, + "step": 13084 + }, + { + "epoch": 1.4369646386997583, + "grad_norm": 2.1497483253479004, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7318450212478638, + "num_tokens": 332441913.0, + "step": 13085 + }, + { + "epoch": 1.437074456402372, + "grad_norm": 2.286952257156372, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7235590219497681, + "num_tokens": 332468114.0, + "step": 13086 + }, + { + "epoch": 1.4371842741049856, + "grad_norm": 2.1660547256469727, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7102020382881165, + "num_tokens": 332493363.0, + "step": 13087 + }, + { + "epoch": 1.4372940918075994, + "grad_norm": 2.569211959838867, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7047317624092102, + "num_tokens": 332513245.0, + "step": 13088 + }, + { + "epoch": 1.4374039095102131, + "grad_norm": 2.217942237854004, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7310319542884827, + "num_tokens": 332538723.0, + "step": 13089 + }, + { + "epoch": 1.4375137272128267, + "grad_norm": 2.175464630126953, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7183531522750854, + "num_tokens": 332566753.0, + "step": 13090 + }, + { + "epoch": 1.4376235449154404, + "grad_norm": 2.4117019176483154, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7387608289718628, + "num_tokens": 332589301.0, + "step": 13091 + }, + { + "epoch": 1.437733362618054, + "grad_norm": 2.432300567626953, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7149487137794495, + "num_tokens": 332611728.0, + "step": 13092 + }, + { + "epoch": 1.4378431803206677, + "grad_norm": 2.3327810764312744, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7075611352920532, + "num_tokens": 332634661.0, + "step": 13093 + }, + { + "epoch": 1.4379529980232815, + "grad_norm": 2.132688522338867, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7172636985778809, + "num_tokens": 332663153.0, + "step": 13094 + }, + { + "epoch": 1.438062815725895, + "grad_norm": 2.163182020187378, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7325263023376465, + "num_tokens": 332690294.0, + "step": 13095 + }, + { + "epoch": 1.4381726334285085, + "grad_norm": 2.320197105407715, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7227671146392822, + "num_tokens": 332713976.0, + "step": 13096 + }, + { + "epoch": 1.4382824511311223, + "grad_norm": 2.1170575618743896, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7044376134872437, + "num_tokens": 332743229.0, + "step": 13097 + }, + { + "epoch": 1.438392268833736, + "grad_norm": 2.3607916831970215, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.713797926902771, + "num_tokens": 332766842.0, + "step": 13098 + }, + { + "epoch": 1.4385020865363496, + "grad_norm": 2.764603853225708, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7270509004592896, + "num_tokens": 332786159.0, + "step": 13099 + }, + { + "epoch": 1.4386119042389633, + "grad_norm": 2.3717494010925293, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7284742593765259, + "num_tokens": 332809046.0, + "step": 13100 + }, + { + "epoch": 1.4387217219415769, + "grad_norm": 2.2443602085113525, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.721499502658844, + "num_tokens": 332833987.0, + "step": 13101 + }, + { + "epoch": 1.4388315396441906, + "grad_norm": 2.1296350955963135, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7193529605865479, + "num_tokens": 332862654.0, + "step": 13102 + }, + { + "epoch": 1.4389413573468044, + "grad_norm": 2.2908220291137695, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7140164375305176, + "num_tokens": 332886904.0, + "step": 13103 + }, + { + "epoch": 1.439051175049418, + "grad_norm": 1.9717531204223633, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7365342378616333, + "num_tokens": 332917265.0, + "step": 13104 + }, + { + "epoch": 1.4391609927520317, + "grad_norm": 1.905429482460022, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7396694421768188, + "num_tokens": 332949001.0, + "step": 13105 + }, + { + "epoch": 1.4392708104546452, + "grad_norm": 1.943644642829895, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7129133939743042, + "num_tokens": 332982091.0, + "step": 13106 + }, + { + "epoch": 1.439380628157259, + "grad_norm": 2.221280813217163, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7218601703643799, + "num_tokens": 333010628.0, + "step": 13107 + }, + { + "epoch": 1.4394904458598727, + "grad_norm": 2.0384361743927, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.7016060948371887, + "num_tokens": 333039881.0, + "step": 13108 + }, + { + "epoch": 1.4396002635624863, + "grad_norm": 2.122131109237671, + "learning_rate": 1e-06, + "loss": 1.0803, + "mean_token_accuracy": 0.6801855564117432, + "num_tokens": 333068072.0, + "step": 13109 + }, + { + "epoch": 1.4397100812650998, + "grad_norm": 2.1689646244049072, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7068057656288147, + "num_tokens": 333095929.0, + "step": 13110 + }, + { + "epoch": 1.4398198989677136, + "grad_norm": 2.160008430480957, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7370470762252808, + "num_tokens": 333120193.0, + "step": 13111 + }, + { + "epoch": 1.4399297166703273, + "grad_norm": 2.2021050453186035, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7146388292312622, + "num_tokens": 333147805.0, + "step": 13112 + }, + { + "epoch": 1.4400395343729409, + "grad_norm": 2.202949047088623, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7127038240432739, + "num_tokens": 333173071.0, + "step": 13113 + }, + { + "epoch": 1.4401493520755546, + "grad_norm": 2.3240318298339844, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7116797566413879, + "num_tokens": 333199242.0, + "step": 13114 + }, + { + "epoch": 1.4402591697781681, + "grad_norm": 2.119004726409912, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.710469126701355, + "num_tokens": 333226468.0, + "step": 13115 + }, + { + "epoch": 1.440368987480782, + "grad_norm": 2.391298532485962, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7230710983276367, + "num_tokens": 333250373.0, + "step": 13116 + }, + { + "epoch": 1.4404788051833957, + "grad_norm": 2.4045028686523438, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7375268936157227, + "num_tokens": 333272182.0, + "step": 13117 + }, + { + "epoch": 1.4405886228860092, + "grad_norm": 2.2177529335021973, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7186649441719055, + "num_tokens": 333296978.0, + "step": 13118 + }, + { + "epoch": 1.440698440588623, + "grad_norm": 2.2076191902160645, + "learning_rate": 1e-06, + "loss": 1.02, + "mean_token_accuracy": 0.697097659111023, + "num_tokens": 333325754.0, + "step": 13119 + }, + { + "epoch": 1.4408082582912365, + "grad_norm": 2.670316696166992, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7216652035713196, + "num_tokens": 333346831.0, + "step": 13120 + }, + { + "epoch": 1.4409180759938502, + "grad_norm": 2.3546955585479736, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7138947248458862, + "num_tokens": 333371195.0, + "step": 13121 + }, + { + "epoch": 1.441027893696464, + "grad_norm": 2.218768835067749, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7103391885757446, + "num_tokens": 333397238.0, + "step": 13122 + }, + { + "epoch": 1.4411377113990775, + "grad_norm": 2.5066986083984375, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.730684757232666, + "num_tokens": 333417768.0, + "step": 13123 + }, + { + "epoch": 1.441247529101691, + "grad_norm": 2.150834798812866, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7112733721733093, + "num_tokens": 333444374.0, + "step": 13124 + }, + { + "epoch": 1.4413573468043048, + "grad_norm": 2.340078592300415, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.7012701034545898, + "num_tokens": 333468761.0, + "step": 13125 + }, + { + "epoch": 1.4414671645069186, + "grad_norm": 2.0909628868103027, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7146674990653992, + "num_tokens": 333496880.0, + "step": 13126 + }, + { + "epoch": 1.4415769822095321, + "grad_norm": 2.507200002670288, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.72025066614151, + "num_tokens": 333520951.0, + "step": 13127 + }, + { + "epoch": 1.4416867999121459, + "grad_norm": 2.1779420375823975, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.706132173538208, + "num_tokens": 333546861.0, + "step": 13128 + }, + { + "epoch": 1.4417966176147594, + "grad_norm": 2.146214485168457, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.7029163241386414, + "num_tokens": 333577415.0, + "step": 13129 + }, + { + "epoch": 1.4419064353173732, + "grad_norm": 2.3589885234832764, + "learning_rate": 1e-06, + "loss": 0.8816, + "mean_token_accuracy": 0.733163595199585, + "num_tokens": 333600881.0, + "step": 13130 + }, + { + "epoch": 1.442016253019987, + "grad_norm": 2.1540236473083496, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7148769497871399, + "num_tokens": 333627918.0, + "step": 13131 + }, + { + "epoch": 1.4421260707226005, + "grad_norm": 2.267336368560791, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7229859828948975, + "num_tokens": 333650853.0, + "step": 13132 + }, + { + "epoch": 1.4422358884252142, + "grad_norm": 2.3477540016174316, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7264275550842285, + "num_tokens": 333674023.0, + "step": 13133 + }, + { + "epoch": 1.4423457061278278, + "grad_norm": 2.2160685062408447, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.707149088382721, + "num_tokens": 333701021.0, + "step": 13134 + }, + { + "epoch": 1.4424555238304415, + "grad_norm": 2.2587623596191406, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7291154861450195, + "num_tokens": 333726061.0, + "step": 13135 + }, + { + "epoch": 1.442565341533055, + "grad_norm": 2.1099565029144287, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7208791971206665, + "num_tokens": 333754006.0, + "step": 13136 + }, + { + "epoch": 1.4426751592356688, + "grad_norm": 2.2383475303649902, + "learning_rate": 1e-06, + "loss": 0.9411, + "mean_token_accuracy": 0.7014576196670532, + "num_tokens": 333779306.0, + "step": 13137 + }, + { + "epoch": 1.4427849769382823, + "grad_norm": 2.442553997039795, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7256972789764404, + "num_tokens": 333801381.0, + "step": 13138 + }, + { + "epoch": 1.442894794640896, + "grad_norm": 2.1677072048187256, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7348428964614868, + "num_tokens": 333827820.0, + "step": 13139 + }, + { + "epoch": 1.4430046123435099, + "grad_norm": 2.423279285430908, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7203776836395264, + "num_tokens": 333851264.0, + "step": 13140 + }, + { + "epoch": 1.4431144300461234, + "grad_norm": 2.6540913581848145, + "learning_rate": 1e-06, + "loss": 0.8446, + "mean_token_accuracy": 0.7328522205352783, + "num_tokens": 333870296.0, + "step": 13141 + }, + { + "epoch": 1.4432242477487371, + "grad_norm": 1.9935506582260132, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.6992337703704834, + "num_tokens": 333904125.0, + "step": 13142 + }, + { + "epoch": 1.4433340654513507, + "grad_norm": 2.505788564682007, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7239776253700256, + "num_tokens": 333925982.0, + "step": 13143 + }, + { + "epoch": 1.4434438831539644, + "grad_norm": 2.382019519805908, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.7203716039657593, + "num_tokens": 333950921.0, + "step": 13144 + }, + { + "epoch": 1.4435537008565782, + "grad_norm": 1.9624733924865723, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7064802050590515, + "num_tokens": 333987412.0, + "step": 13145 + }, + { + "epoch": 1.4436635185591917, + "grad_norm": 2.3240911960601807, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7084782719612122, + "num_tokens": 334012978.0, + "step": 13146 + }, + { + "epoch": 1.4437733362618055, + "grad_norm": 2.638535261154175, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.7495803833007812, + "num_tokens": 334030570.0, + "step": 13147 + }, + { + "epoch": 1.443883153964419, + "grad_norm": 2.0710837841033936, + "learning_rate": 1e-06, + "loss": 1.0007, + "mean_token_accuracy": 0.6917949318885803, + "num_tokens": 334058603.0, + "step": 13148 + }, + { + "epoch": 1.4439929716670328, + "grad_norm": 2.093837022781372, + "learning_rate": 1e-06, + "loss": 0.8581, + "mean_token_accuracy": 0.7334824800491333, + "num_tokens": 334086911.0, + "step": 13149 + }, + { + "epoch": 1.4441027893696463, + "grad_norm": 2.3731706142425537, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7058156728744507, + "num_tokens": 334110455.0, + "step": 13150 + }, + { + "epoch": 1.44421260707226, + "grad_norm": 2.087130308151245, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.726630687713623, + "num_tokens": 334138178.0, + "step": 13151 + }, + { + "epoch": 1.4443224247748736, + "grad_norm": 2.086232900619507, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.721899688243866, + "num_tokens": 334167032.0, + "step": 13152 + }, + { + "epoch": 1.4444322424774874, + "grad_norm": 2.2807939052581787, + "learning_rate": 1e-06, + "loss": 0.8563, + "mean_token_accuracy": 0.7296252250671387, + "num_tokens": 334191778.0, + "step": 13153 + }, + { + "epoch": 1.4445420601801011, + "grad_norm": 2.765078067779541, + "learning_rate": 1e-06, + "loss": 0.767, + "mean_token_accuracy": 0.7461156249046326, + "num_tokens": 334209911.0, + "step": 13154 + }, + { + "epoch": 1.4446518778827147, + "grad_norm": 2.5351169109344482, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7225813269615173, + "num_tokens": 334232557.0, + "step": 13155 + }, + { + "epoch": 1.4447616955853284, + "grad_norm": 2.048694372177124, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6927130222320557, + "num_tokens": 334263260.0, + "step": 13156 + }, + { + "epoch": 1.444871513287942, + "grad_norm": 2.4134109020233154, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7355483770370483, + "num_tokens": 334285654.0, + "step": 13157 + }, + { + "epoch": 1.4449813309905557, + "grad_norm": 2.1423468589782715, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7324152588844299, + "num_tokens": 334314178.0, + "step": 13158 + }, + { + "epoch": 1.4450911486931695, + "grad_norm": 2.5498363971710205, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7387291789054871, + "num_tokens": 334336285.0, + "step": 13159 + }, + { + "epoch": 1.445200966395783, + "grad_norm": 2.0936005115509033, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7323406934738159, + "num_tokens": 334362302.0, + "step": 13160 + }, + { + "epoch": 1.4453107840983965, + "grad_norm": 2.2895379066467285, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.7330565452575684, + "num_tokens": 334385449.0, + "step": 13161 + }, + { + "epoch": 1.4454206018010103, + "grad_norm": 2.4731101989746094, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7316575050354004, + "num_tokens": 334408384.0, + "step": 13162 + }, + { + "epoch": 1.445530419503624, + "grad_norm": 2.239696979522705, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7255356311798096, + "num_tokens": 334435861.0, + "step": 13163 + }, + { + "epoch": 1.4456402372062376, + "grad_norm": 2.0950663089752197, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.7269669771194458, + "num_tokens": 334465940.0, + "step": 13164 + }, + { + "epoch": 1.4457500549088513, + "grad_norm": 2.3226685523986816, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7333714365959167, + "num_tokens": 334490317.0, + "step": 13165 + }, + { + "epoch": 1.4458598726114649, + "grad_norm": 2.6407406330108643, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7427799701690674, + "num_tokens": 334509440.0, + "step": 13166 + }, + { + "epoch": 1.4459696903140786, + "grad_norm": 2.3352065086364746, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.726313591003418, + "num_tokens": 334532928.0, + "step": 13167 + }, + { + "epoch": 1.4460795080166924, + "grad_norm": 2.280364513397217, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7115286588668823, + "num_tokens": 334558559.0, + "step": 13168 + }, + { + "epoch": 1.446189325719306, + "grad_norm": 2.174065351486206, + "learning_rate": 1e-06, + "loss": 0.8779, + "mean_token_accuracy": 0.7258560657501221, + "num_tokens": 334583752.0, + "step": 13169 + }, + { + "epoch": 1.4462991434219197, + "grad_norm": 2.268136501312256, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.7023163437843323, + "num_tokens": 334611751.0, + "step": 13170 + }, + { + "epoch": 1.4464089611245332, + "grad_norm": 2.512422800064087, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7084982991218567, + "num_tokens": 334633158.0, + "step": 13171 + }, + { + "epoch": 1.446518778827147, + "grad_norm": 2.383873462677002, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7274309396743774, + "num_tokens": 334656474.0, + "step": 13172 + }, + { + "epoch": 1.4466285965297607, + "grad_norm": 2.174593210220337, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7334635257720947, + "num_tokens": 334682587.0, + "step": 13173 + }, + { + "epoch": 1.4467384142323743, + "grad_norm": 2.635576009750366, + "learning_rate": 1e-06, + "loss": 0.8019, + "mean_token_accuracy": 0.7434493899345398, + "num_tokens": 334700620.0, + "step": 13174 + }, + { + "epoch": 1.4468482319349878, + "grad_norm": 2.2034454345703125, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.7113845348358154, + "num_tokens": 334730661.0, + "step": 13175 + }, + { + "epoch": 1.4469580496376016, + "grad_norm": 2.151308536529541, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7339366674423218, + "num_tokens": 334760471.0, + "step": 13176 + }, + { + "epoch": 1.4470678673402153, + "grad_norm": 2.364544153213501, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7134115099906921, + "num_tokens": 334783007.0, + "step": 13177 + }, + { + "epoch": 1.4471776850428288, + "grad_norm": 2.232260227203369, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7272618412971497, + "num_tokens": 334808162.0, + "step": 13178 + }, + { + "epoch": 1.4472875027454426, + "grad_norm": 2.64444637298584, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.726449728012085, + "num_tokens": 334828689.0, + "step": 13179 + }, + { + "epoch": 1.4473973204480561, + "grad_norm": 2.1657774448394775, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7141625881195068, + "num_tokens": 334857170.0, + "step": 13180 + }, + { + "epoch": 1.44750713815067, + "grad_norm": 2.7620081901550293, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7454844117164612, + "num_tokens": 334875744.0, + "step": 13181 + }, + { + "epoch": 1.4476169558532836, + "grad_norm": 2.3434195518493652, + "learning_rate": 1e-06, + "loss": 0.819, + "mean_token_accuracy": 0.7385575771331787, + "num_tokens": 334897664.0, + "step": 13182 + }, + { + "epoch": 1.4477267735558972, + "grad_norm": 2.254432439804077, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7129102945327759, + "num_tokens": 334923286.0, + "step": 13183 + }, + { + "epoch": 1.447836591258511, + "grad_norm": 2.166545867919922, + "learning_rate": 1e-06, + "loss": 0.7793, + "mean_token_accuracy": 0.7475835084915161, + "num_tokens": 334948763.0, + "step": 13184 + }, + { + "epoch": 1.4479464089611245, + "grad_norm": 2.0443601608276367, + "learning_rate": 1e-06, + "loss": 0.827, + "mean_token_accuracy": 0.7414445877075195, + "num_tokens": 334976201.0, + "step": 13185 + }, + { + "epoch": 1.4480562266637382, + "grad_norm": 2.2816436290740967, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7238169312477112, + "num_tokens": 335002395.0, + "step": 13186 + }, + { + "epoch": 1.448166044366352, + "grad_norm": 2.3037285804748535, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.731853187084198, + "num_tokens": 335026205.0, + "step": 13187 + }, + { + "epoch": 1.4482758620689655, + "grad_norm": 2.2492988109588623, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7125478982925415, + "num_tokens": 335053292.0, + "step": 13188 + }, + { + "epoch": 1.448385679771579, + "grad_norm": 2.016256809234619, + "learning_rate": 1e-06, + "loss": 0.9516, + "mean_token_accuracy": 0.7099500894546509, + "num_tokens": 335081806.0, + "step": 13189 + }, + { + "epoch": 1.4484954974741928, + "grad_norm": 2.162283182144165, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7247642278671265, + "num_tokens": 335108895.0, + "step": 13190 + }, + { + "epoch": 1.4486053151768066, + "grad_norm": 2.152660369873047, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7034337520599365, + "num_tokens": 335135111.0, + "step": 13191 + }, + { + "epoch": 1.44871513287942, + "grad_norm": 2.2384724617004395, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7292615175247192, + "num_tokens": 335159923.0, + "step": 13192 + }, + { + "epoch": 1.4488249505820339, + "grad_norm": 2.2565362453460693, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7277616262435913, + "num_tokens": 335184975.0, + "step": 13193 + }, + { + "epoch": 1.4489347682846474, + "grad_norm": 2.157000780105591, + "learning_rate": 1e-06, + "loss": 1.0085, + "mean_token_accuracy": 0.6894016265869141, + "num_tokens": 335213968.0, + "step": 13194 + }, + { + "epoch": 1.4490445859872612, + "grad_norm": 2.0302939414978027, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.7045232653617859, + "num_tokens": 335245149.0, + "step": 13195 + }, + { + "epoch": 1.449154403689875, + "grad_norm": 2.4009366035461426, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7222939133644104, + "num_tokens": 335268068.0, + "step": 13196 + }, + { + "epoch": 1.4492642213924884, + "grad_norm": 2.143508195877075, + "learning_rate": 1e-06, + "loss": 0.9561, + "mean_token_accuracy": 0.7092016935348511, + "num_tokens": 335295384.0, + "step": 13197 + }, + { + "epoch": 1.4493740390951022, + "grad_norm": 2.1687333583831787, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7422460317611694, + "num_tokens": 335323153.0, + "step": 13198 + }, + { + "epoch": 1.4494838567977157, + "grad_norm": 2.1976277828216553, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7135578393936157, + "num_tokens": 335349291.0, + "step": 13199 + }, + { + "epoch": 1.4495936745003295, + "grad_norm": 1.9666568040847778, + "learning_rate": 1e-06, + "loss": 0.7856, + "mean_token_accuracy": 0.7599840760231018, + "num_tokens": 335377229.0, + "step": 13200 + }, + { + "epoch": 1.449703492202943, + "grad_norm": 2.368623733520508, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7290477156639099, + "num_tokens": 335400702.0, + "step": 13201 + }, + { + "epoch": 1.4498133099055568, + "grad_norm": 2.26190447807312, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.6981754302978516, + "num_tokens": 335426473.0, + "step": 13202 + }, + { + "epoch": 1.4499231276081703, + "grad_norm": 2.12873911857605, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7022930383682251, + "num_tokens": 335455272.0, + "step": 13203 + }, + { + "epoch": 1.450032945310784, + "grad_norm": 2.090697765350342, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7158037424087524, + "num_tokens": 335484425.0, + "step": 13204 + }, + { + "epoch": 1.4501427630133978, + "grad_norm": 2.3319942951202393, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7035343647003174, + "num_tokens": 335508737.0, + "step": 13205 + }, + { + "epoch": 1.4502525807160114, + "grad_norm": 2.5513157844543457, + "learning_rate": 1e-06, + "loss": 0.7956, + "mean_token_accuracy": 0.7507783770561218, + "num_tokens": 335533077.0, + "step": 13206 + }, + { + "epoch": 1.4503623984186251, + "grad_norm": 2.2230074405670166, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.693806529045105, + "num_tokens": 335559393.0, + "step": 13207 + }, + { + "epoch": 1.4504722161212387, + "grad_norm": 2.0372307300567627, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7306145429611206, + "num_tokens": 335588830.0, + "step": 13208 + }, + { + "epoch": 1.4505820338238524, + "grad_norm": 2.0194952487945557, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7304943799972534, + "num_tokens": 335617815.0, + "step": 13209 + }, + { + "epoch": 1.4506918515264662, + "grad_norm": 2.3473565578460693, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.729614794254303, + "num_tokens": 335640584.0, + "step": 13210 + }, + { + "epoch": 1.4508016692290797, + "grad_norm": 2.2980005741119385, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7141989469528198, + "num_tokens": 335665310.0, + "step": 13211 + }, + { + "epoch": 1.4509114869316933, + "grad_norm": 2.274576425552368, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7096299529075623, + "num_tokens": 335692629.0, + "step": 13212 + }, + { + "epoch": 1.451021304634307, + "grad_norm": 2.3293845653533936, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7155897617340088, + "num_tokens": 335717285.0, + "step": 13213 + }, + { + "epoch": 1.4511311223369208, + "grad_norm": 2.182373285293579, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7228028774261475, + "num_tokens": 335744644.0, + "step": 13214 + }, + { + "epoch": 1.4512409400395343, + "grad_norm": 2.0944089889526367, + "learning_rate": 1e-06, + "loss": 1.0568, + "mean_token_accuracy": 0.6784110069274902, + "num_tokens": 335774563.0, + "step": 13215 + }, + { + "epoch": 1.451350757742148, + "grad_norm": 2.4973461627960205, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7222923040390015, + "num_tokens": 335795988.0, + "step": 13216 + }, + { + "epoch": 1.4514605754447616, + "grad_norm": 2.0582079887390137, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7204703092575073, + "num_tokens": 335824808.0, + "step": 13217 + }, + { + "epoch": 1.4515703931473753, + "grad_norm": 2.3522346019744873, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7228431701660156, + "num_tokens": 335848278.0, + "step": 13218 + }, + { + "epoch": 1.451680210849989, + "grad_norm": 2.337364435195923, + "learning_rate": 1e-06, + "loss": 0.7454, + "mean_token_accuracy": 0.7579529285430908, + "num_tokens": 335870950.0, + "step": 13219 + }, + { + "epoch": 1.4517900285526026, + "grad_norm": 2.6083366870880127, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7210326194763184, + "num_tokens": 335891640.0, + "step": 13220 + }, + { + "epoch": 1.4518998462552164, + "grad_norm": 2.172581195831299, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7168334126472473, + "num_tokens": 335918508.0, + "step": 13221 + }, + { + "epoch": 1.45200966395783, + "grad_norm": 2.458073854446411, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6898669004440308, + "num_tokens": 335942388.0, + "step": 13222 + }, + { + "epoch": 1.4521194816604437, + "grad_norm": 2.2911105155944824, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7126859426498413, + "num_tokens": 335966524.0, + "step": 13223 + }, + { + "epoch": 1.4522292993630574, + "grad_norm": 2.1036906242370605, + "learning_rate": 1e-06, + "loss": 1.0722, + "mean_token_accuracy": 0.6785098910331726, + "num_tokens": 335995859.0, + "step": 13224 + }, + { + "epoch": 1.452339117065671, + "grad_norm": 2.147672176361084, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.6972846984863281, + "num_tokens": 336024647.0, + "step": 13225 + }, + { + "epoch": 1.4524489347682845, + "grad_norm": 2.5826478004455566, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.726593017578125, + "num_tokens": 336046292.0, + "step": 13226 + }, + { + "epoch": 1.4525587524708983, + "grad_norm": 2.357264280319214, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7283982038497925, + "num_tokens": 336067354.0, + "step": 13227 + }, + { + "epoch": 1.452668570173512, + "grad_norm": 2.0231664180755615, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7246778607368469, + "num_tokens": 336097148.0, + "step": 13228 + }, + { + "epoch": 1.4527783878761256, + "grad_norm": 2.3481709957122803, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.719826877117157, + "num_tokens": 336124000.0, + "step": 13229 + }, + { + "epoch": 1.4528882055787393, + "grad_norm": 2.363278388977051, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.716805636882782, + "num_tokens": 336149088.0, + "step": 13230 + }, + { + "epoch": 1.4529980232813529, + "grad_norm": 2.4317378997802734, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7131551504135132, + "num_tokens": 336171625.0, + "step": 13231 + }, + { + "epoch": 1.4531078409839666, + "grad_norm": 2.158097982406616, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.702180027961731, + "num_tokens": 336198323.0, + "step": 13232 + }, + { + "epoch": 1.4532176586865804, + "grad_norm": 2.3013651371002197, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7346408367156982, + "num_tokens": 336220916.0, + "step": 13233 + }, + { + "epoch": 1.453327476389194, + "grad_norm": 1.89368736743927, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7270424365997314, + "num_tokens": 336252558.0, + "step": 13234 + }, + { + "epoch": 1.4534372940918077, + "grad_norm": 2.3343071937561035, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7115397453308105, + "num_tokens": 336278146.0, + "step": 13235 + }, + { + "epoch": 1.4535471117944212, + "grad_norm": 2.0188488960266113, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.731467604637146, + "num_tokens": 336306846.0, + "step": 13236 + }, + { + "epoch": 1.453656929497035, + "grad_norm": 2.0602774620056152, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7027311325073242, + "num_tokens": 336338219.0, + "step": 13237 + }, + { + "epoch": 1.4537667471996487, + "grad_norm": 1.9196794033050537, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7295581102371216, + "num_tokens": 336371656.0, + "step": 13238 + }, + { + "epoch": 1.4538765649022622, + "grad_norm": 2.115910530090332, + "learning_rate": 1e-06, + "loss": 1.0203, + "mean_token_accuracy": 0.6907272338867188, + "num_tokens": 336401291.0, + "step": 13239 + }, + { + "epoch": 1.4539863826048758, + "grad_norm": 2.170536756515503, + "learning_rate": 1e-06, + "loss": 0.8463, + "mean_token_accuracy": 0.7302299737930298, + "num_tokens": 336428496.0, + "step": 13240 + }, + { + "epoch": 1.4540962003074895, + "grad_norm": 2.213994264602661, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7237319350242615, + "num_tokens": 336455108.0, + "step": 13241 + }, + { + "epoch": 1.4542060180101033, + "grad_norm": 2.314239263534546, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.6978117227554321, + "num_tokens": 336480402.0, + "step": 13242 + }, + { + "epoch": 1.4543158357127168, + "grad_norm": 2.336376428604126, + "learning_rate": 1e-06, + "loss": 0.8063, + "mean_token_accuracy": 0.7403408885002136, + "num_tokens": 336504579.0, + "step": 13243 + }, + { + "epoch": 1.4544256534153306, + "grad_norm": 2.221428394317627, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7133955955505371, + "num_tokens": 336530821.0, + "step": 13244 + }, + { + "epoch": 1.4545354711179441, + "grad_norm": 2.419886827468872, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7344380021095276, + "num_tokens": 336554816.0, + "step": 13245 + }, + { + "epoch": 1.4546452888205579, + "grad_norm": 2.0845792293548584, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7182638645172119, + "num_tokens": 336583957.0, + "step": 13246 + }, + { + "epoch": 1.4547551065231716, + "grad_norm": 2.2937474250793457, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7307829856872559, + "num_tokens": 336608715.0, + "step": 13247 + }, + { + "epoch": 1.4548649242257852, + "grad_norm": 2.3178369998931885, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7101230621337891, + "num_tokens": 336632988.0, + "step": 13248 + }, + { + "epoch": 1.454974741928399, + "grad_norm": 2.0341410636901855, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7109548449516296, + "num_tokens": 336664345.0, + "step": 13249 + }, + { + "epoch": 1.4550845596310125, + "grad_norm": 2.304046154022217, + "learning_rate": 1e-06, + "loss": 0.8967, + "mean_token_accuracy": 0.7155098915100098, + "num_tokens": 336689407.0, + "step": 13250 + }, + { + "epoch": 1.4551943773336262, + "grad_norm": 2.0558245182037354, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7226225137710571, + "num_tokens": 336718513.0, + "step": 13251 + }, + { + "epoch": 1.45530419503624, + "grad_norm": 2.1930224895477295, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7049200534820557, + "num_tokens": 336745114.0, + "step": 13252 + }, + { + "epoch": 1.4554140127388535, + "grad_norm": 2.2775373458862305, + "learning_rate": 1e-06, + "loss": 0.773, + "mean_token_accuracy": 0.750367283821106, + "num_tokens": 336767429.0, + "step": 13253 + }, + { + "epoch": 1.455523830441467, + "grad_norm": 2.012932538986206, + "learning_rate": 1e-06, + "loss": 1.0202, + "mean_token_accuracy": 0.6949540376663208, + "num_tokens": 336798853.0, + "step": 13254 + }, + { + "epoch": 1.4556336481440808, + "grad_norm": 2.0435478687286377, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7035268545150757, + "num_tokens": 336829634.0, + "step": 13255 + }, + { + "epoch": 1.4557434658466946, + "grad_norm": 2.31961989402771, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7240492105484009, + "num_tokens": 336852298.0, + "step": 13256 + }, + { + "epoch": 1.455853283549308, + "grad_norm": 2.099264621734619, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.7393910884857178, + "num_tokens": 336878535.0, + "step": 13257 + }, + { + "epoch": 1.4559631012519219, + "grad_norm": 2.23077392578125, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7102898359298706, + "num_tokens": 336904587.0, + "step": 13258 + }, + { + "epoch": 1.4560729189545354, + "grad_norm": 2.3600471019744873, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7334591150283813, + "num_tokens": 336928079.0, + "step": 13259 + }, + { + "epoch": 1.4561827366571491, + "grad_norm": 2.425286054611206, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7508772611618042, + "num_tokens": 336950304.0, + "step": 13260 + }, + { + "epoch": 1.456292554359763, + "grad_norm": 2.42600154876709, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7238091826438904, + "num_tokens": 336974841.0, + "step": 13261 + }, + { + "epoch": 1.4564023720623764, + "grad_norm": 2.2685327529907227, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7214981913566589, + "num_tokens": 337001529.0, + "step": 13262 + }, + { + "epoch": 1.4565121897649902, + "grad_norm": 2.630835771560669, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7284196615219116, + "num_tokens": 337022072.0, + "step": 13263 + }, + { + "epoch": 1.4566220074676037, + "grad_norm": 2.2790350914001465, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.736478328704834, + "num_tokens": 337046172.0, + "step": 13264 + }, + { + "epoch": 1.4567318251702175, + "grad_norm": 2.279672384262085, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.6955305933952332, + "num_tokens": 337071502.0, + "step": 13265 + }, + { + "epoch": 1.456841642872831, + "grad_norm": 2.432757616043091, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7154571413993835, + "num_tokens": 337094208.0, + "step": 13266 + }, + { + "epoch": 1.4569514605754448, + "grad_norm": 2.2215073108673096, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7350523471832275, + "num_tokens": 337118497.0, + "step": 13267 + }, + { + "epoch": 1.4570612782780583, + "grad_norm": 1.8623138666152954, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.693597137928009, + "num_tokens": 337156945.0, + "step": 13268 + }, + { + "epoch": 1.457171095980672, + "grad_norm": 2.2435812950134277, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7260972857475281, + "num_tokens": 337182096.0, + "step": 13269 + }, + { + "epoch": 1.4572809136832858, + "grad_norm": 2.1257617473602295, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7185596227645874, + "num_tokens": 337210280.0, + "step": 13270 + }, + { + "epoch": 1.4573907313858994, + "grad_norm": 2.2488584518432617, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7218774557113647, + "num_tokens": 337234793.0, + "step": 13271 + }, + { + "epoch": 1.4575005490885131, + "grad_norm": 2.118940830230713, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7138229608535767, + "num_tokens": 337262236.0, + "step": 13272 + }, + { + "epoch": 1.4576103667911267, + "grad_norm": 2.3281707763671875, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.70563805103302, + "num_tokens": 337288954.0, + "step": 13273 + }, + { + "epoch": 1.4577201844937404, + "grad_norm": 2.313676595687866, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7068409323692322, + "num_tokens": 337314978.0, + "step": 13274 + }, + { + "epoch": 1.4578300021963542, + "grad_norm": 2.124727964401245, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7282331585884094, + "num_tokens": 337344236.0, + "step": 13275 + }, + { + "epoch": 1.4579398198989677, + "grad_norm": 2.1107611656188965, + "learning_rate": 1e-06, + "loss": 0.9026, + "mean_token_accuracy": 0.7253860235214233, + "num_tokens": 337370725.0, + "step": 13276 + }, + { + "epoch": 1.4580496376015812, + "grad_norm": 1.9624967575073242, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7151697874069214, + "num_tokens": 337403800.0, + "step": 13277 + }, + { + "epoch": 1.458159455304195, + "grad_norm": 2.0704121589660645, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.7412885427474976, + "num_tokens": 337432139.0, + "step": 13278 + }, + { + "epoch": 1.4582692730068088, + "grad_norm": 2.2653250694274902, + "learning_rate": 1e-06, + "loss": 0.9853, + "mean_token_accuracy": 0.6937882304191589, + "num_tokens": 337459408.0, + "step": 13279 + }, + { + "epoch": 1.4583790907094223, + "grad_norm": 2.471972703933716, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7221953868865967, + "num_tokens": 337481510.0, + "step": 13280 + }, + { + "epoch": 1.458488908412036, + "grad_norm": 2.1188745498657227, + "learning_rate": 1e-06, + "loss": 0.9662, + "mean_token_accuracy": 0.7052721977233887, + "num_tokens": 337510585.0, + "step": 13281 + }, + { + "epoch": 1.4585987261146496, + "grad_norm": 2.363974094390869, + "learning_rate": 1e-06, + "loss": 0.9844, + "mean_token_accuracy": 0.7049703598022461, + "num_tokens": 337534666.0, + "step": 13282 + }, + { + "epoch": 1.4587085438172633, + "grad_norm": 2.3355681896209717, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7167925834655762, + "num_tokens": 337559189.0, + "step": 13283 + }, + { + "epoch": 1.458818361519877, + "grad_norm": 2.3849167823791504, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7281430959701538, + "num_tokens": 337583165.0, + "step": 13284 + }, + { + "epoch": 1.4589281792224906, + "grad_norm": 2.520113945007324, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.728062629699707, + "num_tokens": 337603568.0, + "step": 13285 + }, + { + "epoch": 1.4590379969251044, + "grad_norm": 2.537750244140625, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7211053371429443, + "num_tokens": 337625100.0, + "step": 13286 + }, + { + "epoch": 1.459147814627718, + "grad_norm": 2.378365993499756, + "learning_rate": 1e-06, + "loss": 0.9598, + "mean_token_accuracy": 0.7013057470321655, + "num_tokens": 337650222.0, + "step": 13287 + }, + { + "epoch": 1.4592576323303317, + "grad_norm": 2.2185425758361816, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7202161550521851, + "num_tokens": 337675296.0, + "step": 13288 + }, + { + "epoch": 1.4593674500329454, + "grad_norm": 2.213189125061035, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7223423719406128, + "num_tokens": 337702978.0, + "step": 13289 + }, + { + "epoch": 1.459477267735559, + "grad_norm": 2.3033251762390137, + "learning_rate": 1e-06, + "loss": 1.007, + "mean_token_accuracy": 0.6987158060073853, + "num_tokens": 337729165.0, + "step": 13290 + }, + { + "epoch": 1.4595870854381725, + "grad_norm": 2.008924961090088, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6968269348144531, + "num_tokens": 337759013.0, + "step": 13291 + }, + { + "epoch": 1.4596969031407863, + "grad_norm": 2.1696646213531494, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7307093143463135, + "num_tokens": 337785476.0, + "step": 13292 + }, + { + "epoch": 1.4598067208434, + "grad_norm": 2.364793062210083, + "learning_rate": 1e-06, + "loss": 0.9291, + "mean_token_accuracy": 0.7179794311523438, + "num_tokens": 337807705.0, + "step": 13293 + }, + { + "epoch": 1.4599165385460136, + "grad_norm": 2.0144028663635254, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7367405891418457, + "num_tokens": 337835123.0, + "step": 13294 + }, + { + "epoch": 1.4600263562486273, + "grad_norm": 2.4912214279174805, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.751313328742981, + "num_tokens": 337853897.0, + "step": 13295 + }, + { + "epoch": 1.4601361739512408, + "grad_norm": 2.04475998878479, + "learning_rate": 1e-06, + "loss": 1.0212, + "mean_token_accuracy": 0.6857904195785522, + "num_tokens": 337886230.0, + "step": 13296 + }, + { + "epoch": 1.4602459916538546, + "grad_norm": 2.412846326828003, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7094054222106934, + "num_tokens": 337913714.0, + "step": 13297 + }, + { + "epoch": 1.4603558093564684, + "grad_norm": 2.18892240524292, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7302961349487305, + "num_tokens": 337943133.0, + "step": 13298 + }, + { + "epoch": 1.460465627059082, + "grad_norm": 1.865149974822998, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7297103404998779, + "num_tokens": 337979346.0, + "step": 13299 + }, + { + "epoch": 1.4605754447616957, + "grad_norm": 2.4665379524230957, + "learning_rate": 1e-06, + "loss": 0.8493, + "mean_token_accuracy": 0.7311871647834778, + "num_tokens": 338000672.0, + "step": 13300 + }, + { + "epoch": 1.4606852624643092, + "grad_norm": 1.9907722473144531, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7174873352050781, + "num_tokens": 338029690.0, + "step": 13301 + }, + { + "epoch": 1.460795080166923, + "grad_norm": 2.4104886054992676, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7224447727203369, + "num_tokens": 338054052.0, + "step": 13302 + }, + { + "epoch": 1.4609048978695367, + "grad_norm": 2.1600120067596436, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7255353331565857, + "num_tokens": 338081754.0, + "step": 13303 + }, + { + "epoch": 1.4610147155721502, + "grad_norm": 2.4969379901885986, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7254250645637512, + "num_tokens": 338102524.0, + "step": 13304 + }, + { + "epoch": 1.4611245332747638, + "grad_norm": 2.3194713592529297, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7172824740409851, + "num_tokens": 338127488.0, + "step": 13305 + }, + { + "epoch": 1.4612343509773775, + "grad_norm": 2.239612579345703, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.6926949620246887, + "num_tokens": 338157511.0, + "step": 13306 + }, + { + "epoch": 1.4613441686799913, + "grad_norm": 2.161633253097534, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7210413217544556, + "num_tokens": 338184537.0, + "step": 13307 + }, + { + "epoch": 1.4614539863826048, + "grad_norm": 2.0796115398406982, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7396796941757202, + "num_tokens": 338212179.0, + "step": 13308 + }, + { + "epoch": 1.4615638040852186, + "grad_norm": 2.5571742057800293, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7464313507080078, + "num_tokens": 338231214.0, + "step": 13309 + }, + { + "epoch": 1.4616736217878321, + "grad_norm": 2.17466139793396, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7136210203170776, + "num_tokens": 338255985.0, + "step": 13310 + }, + { + "epoch": 1.4617834394904459, + "grad_norm": 2.0672855377197266, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7018275260925293, + "num_tokens": 338286358.0, + "step": 13311 + }, + { + "epoch": 1.4618932571930596, + "grad_norm": 2.172422170639038, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7160555720329285, + "num_tokens": 338314182.0, + "step": 13312 + }, + { + "epoch": 1.4620030748956732, + "grad_norm": 2.238114833831787, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7173476219177246, + "num_tokens": 338340676.0, + "step": 13313 + }, + { + "epoch": 1.462112892598287, + "grad_norm": 2.387511968612671, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6931630373001099, + "num_tokens": 338364376.0, + "step": 13314 + }, + { + "epoch": 1.4622227103009005, + "grad_norm": 2.2681734561920166, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7162883281707764, + "num_tokens": 338389990.0, + "step": 13315 + }, + { + "epoch": 1.4623325280035142, + "grad_norm": 2.362915515899658, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7177109718322754, + "num_tokens": 338413612.0, + "step": 13316 + }, + { + "epoch": 1.462442345706128, + "grad_norm": 2.38999080657959, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7201479077339172, + "num_tokens": 338437892.0, + "step": 13317 + }, + { + "epoch": 1.4625521634087415, + "grad_norm": 2.3665897846221924, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.711609423160553, + "num_tokens": 338462321.0, + "step": 13318 + }, + { + "epoch": 1.462661981111355, + "grad_norm": 2.4115679264068604, + "learning_rate": 1e-06, + "loss": 0.9333, + "mean_token_accuracy": 0.7151381969451904, + "num_tokens": 338487416.0, + "step": 13319 + }, + { + "epoch": 1.4627717988139688, + "grad_norm": 2.273916244506836, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7230005264282227, + "num_tokens": 338515260.0, + "step": 13320 + }, + { + "epoch": 1.4628816165165826, + "grad_norm": 2.3921942710876465, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7291258573532104, + "num_tokens": 338537612.0, + "step": 13321 + }, + { + "epoch": 1.462991434219196, + "grad_norm": 2.0535390377044678, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7107968330383301, + "num_tokens": 338570366.0, + "step": 13322 + }, + { + "epoch": 1.4631012519218098, + "grad_norm": 2.2852792739868164, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.717623233795166, + "num_tokens": 338594987.0, + "step": 13323 + }, + { + "epoch": 1.4632110696244234, + "grad_norm": 2.457146167755127, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7449610829353333, + "num_tokens": 338618143.0, + "step": 13324 + }, + { + "epoch": 1.4633208873270371, + "grad_norm": 2.4225378036499023, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7286173105239868, + "num_tokens": 338641631.0, + "step": 13325 + }, + { + "epoch": 1.463430705029651, + "grad_norm": 2.4801769256591797, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7321081161499023, + "num_tokens": 338662708.0, + "step": 13326 + }, + { + "epoch": 1.4635405227322644, + "grad_norm": 2.194370985031128, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.714404821395874, + "num_tokens": 338690064.0, + "step": 13327 + }, + { + "epoch": 1.4636503404348782, + "grad_norm": 2.1459920406341553, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7092396020889282, + "num_tokens": 338718992.0, + "step": 13328 + }, + { + "epoch": 1.4637601581374917, + "grad_norm": 2.2640035152435303, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7139843106269836, + "num_tokens": 338743575.0, + "step": 13329 + }, + { + "epoch": 1.4638699758401055, + "grad_norm": 2.3771238327026367, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7382775545120239, + "num_tokens": 338765647.0, + "step": 13330 + }, + { + "epoch": 1.463979793542719, + "grad_norm": 2.3341994285583496, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7162512540817261, + "num_tokens": 338792588.0, + "step": 13331 + }, + { + "epoch": 1.4640896112453328, + "grad_norm": 2.2420506477355957, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7122335433959961, + "num_tokens": 338819348.0, + "step": 13332 + }, + { + "epoch": 1.4641994289479463, + "grad_norm": 2.541492223739624, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7195771932601929, + "num_tokens": 338840647.0, + "step": 13333 + }, + { + "epoch": 1.46430924665056, + "grad_norm": 2.3712894916534424, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7316958904266357, + "num_tokens": 338864038.0, + "step": 13334 + }, + { + "epoch": 1.4644190643531738, + "grad_norm": 2.330364942550659, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7114783525466919, + "num_tokens": 338889148.0, + "step": 13335 + }, + { + "epoch": 1.4645288820557874, + "grad_norm": 2.5117199420928955, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7368383407592773, + "num_tokens": 338910241.0, + "step": 13336 + }, + { + "epoch": 1.464638699758401, + "grad_norm": 2.206070899963379, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7272599339485168, + "num_tokens": 338935893.0, + "step": 13337 + }, + { + "epoch": 1.4647485174610146, + "grad_norm": 2.3109896183013916, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7469353079795837, + "num_tokens": 338959364.0, + "step": 13338 + }, + { + "epoch": 1.4648583351636284, + "grad_norm": 2.1010732650756836, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7177444696426392, + "num_tokens": 338989240.0, + "step": 13339 + }, + { + "epoch": 1.4649681528662422, + "grad_norm": 2.582502841949463, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7146506309509277, + "num_tokens": 339010054.0, + "step": 13340 + }, + { + "epoch": 1.4650779705688557, + "grad_norm": 2.5991547107696533, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7214937806129456, + "num_tokens": 339030802.0, + "step": 13341 + }, + { + "epoch": 1.4651877882714692, + "grad_norm": 2.5287938117980957, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7024243474006653, + "num_tokens": 339053278.0, + "step": 13342 + }, + { + "epoch": 1.465297605974083, + "grad_norm": 2.230700731277466, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7118172645568848, + "num_tokens": 339079950.0, + "step": 13343 + }, + { + "epoch": 1.4654074236766967, + "grad_norm": 2.24342679977417, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7203218936920166, + "num_tokens": 339107947.0, + "step": 13344 + }, + { + "epoch": 1.4655172413793103, + "grad_norm": 2.0484681129455566, + "learning_rate": 1e-06, + "loss": 0.9968, + "mean_token_accuracy": 0.6946390271186829, + "num_tokens": 339138743.0, + "step": 13345 + }, + { + "epoch": 1.465627059081924, + "grad_norm": 2.1332900524139404, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7226086854934692, + "num_tokens": 339166647.0, + "step": 13346 + }, + { + "epoch": 1.4657368767845376, + "grad_norm": 2.4960873126983643, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.7450437545776367, + "num_tokens": 339186276.0, + "step": 13347 + }, + { + "epoch": 1.4658466944871513, + "grad_norm": 2.099522352218628, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7153360843658447, + "num_tokens": 339214866.0, + "step": 13348 + }, + { + "epoch": 1.465956512189765, + "grad_norm": 2.174769163131714, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7214569449424744, + "num_tokens": 339241392.0, + "step": 13349 + }, + { + "epoch": 1.4660663298923786, + "grad_norm": 2.188535213470459, + "learning_rate": 1e-06, + "loss": 0.7823, + "mean_token_accuracy": 0.7430210113525391, + "num_tokens": 339266657.0, + "step": 13350 + }, + { + "epoch": 1.4661761475949924, + "grad_norm": 2.053079843521118, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.7117725610733032, + "num_tokens": 339297211.0, + "step": 13351 + }, + { + "epoch": 1.466285965297606, + "grad_norm": 2.2999374866485596, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7141131162643433, + "num_tokens": 339322593.0, + "step": 13352 + }, + { + "epoch": 1.4663957830002197, + "grad_norm": 2.4829297065734863, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7260745763778687, + "num_tokens": 339343722.0, + "step": 13353 + }, + { + "epoch": 1.4665056007028334, + "grad_norm": 2.20951771736145, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7176103591918945, + "num_tokens": 339371613.0, + "step": 13354 + }, + { + "epoch": 1.466615418405447, + "grad_norm": 2.1770589351654053, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7336097955703735, + "num_tokens": 339400376.0, + "step": 13355 + }, + { + "epoch": 1.4667252361080605, + "grad_norm": 2.290573835372925, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7212470173835754, + "num_tokens": 339426852.0, + "step": 13356 + }, + { + "epoch": 1.4668350538106742, + "grad_norm": 2.1921424865722656, + "learning_rate": 1e-06, + "loss": 0.9813, + "mean_token_accuracy": 0.6963989734649658, + "num_tokens": 339457226.0, + "step": 13357 + }, + { + "epoch": 1.466944871513288, + "grad_norm": 2.2365221977233887, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7118179798126221, + "num_tokens": 339483796.0, + "step": 13358 + }, + { + "epoch": 1.4670546892159015, + "grad_norm": 2.458311080932617, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7402815818786621, + "num_tokens": 339504949.0, + "step": 13359 + }, + { + "epoch": 1.4671645069185153, + "grad_norm": 2.352428674697876, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7073659300804138, + "num_tokens": 339529216.0, + "step": 13360 + }, + { + "epoch": 1.4672743246211288, + "grad_norm": 2.537731885910034, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7034275531768799, + "num_tokens": 339549785.0, + "step": 13361 + }, + { + "epoch": 1.4673841423237426, + "grad_norm": 2.374525547027588, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7387283444404602, + "num_tokens": 339571212.0, + "step": 13362 + }, + { + "epoch": 1.4674939600263563, + "grad_norm": 2.0256128311157227, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.694282054901123, + "num_tokens": 339599585.0, + "step": 13363 + }, + { + "epoch": 1.4676037777289699, + "grad_norm": 2.245084762573242, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.72733074426651, + "num_tokens": 339624869.0, + "step": 13364 + }, + { + "epoch": 1.4677135954315836, + "grad_norm": 1.927734375, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7285842299461365, + "num_tokens": 339655842.0, + "step": 13365 + }, + { + "epoch": 1.4678234131341972, + "grad_norm": 2.169858455657959, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7005144357681274, + "num_tokens": 339683442.0, + "step": 13366 + }, + { + "epoch": 1.467933230836811, + "grad_norm": 2.3816540241241455, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7309690713882446, + "num_tokens": 339706696.0, + "step": 13367 + }, + { + "epoch": 1.4680430485394247, + "grad_norm": 2.1889522075653076, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7100930213928223, + "num_tokens": 339732595.0, + "step": 13368 + }, + { + "epoch": 1.4681528662420382, + "grad_norm": 2.5684080123901367, + "learning_rate": 1e-06, + "loss": 0.9775, + "mean_token_accuracy": 0.7039944529533386, + "num_tokens": 339753555.0, + "step": 13369 + }, + { + "epoch": 1.4682626839446518, + "grad_norm": 2.228301525115967, + "learning_rate": 1e-06, + "loss": 0.8802, + "mean_token_accuracy": 0.7223457098007202, + "num_tokens": 339779948.0, + "step": 13370 + }, + { + "epoch": 1.4683725016472655, + "grad_norm": 2.27968430519104, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.7323828935623169, + "num_tokens": 339805359.0, + "step": 13371 + }, + { + "epoch": 1.4684823193498793, + "grad_norm": 2.381749391555786, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7247636318206787, + "num_tokens": 339827418.0, + "step": 13372 + }, + { + "epoch": 1.4685921370524928, + "grad_norm": 2.1139883995056152, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7053098082542419, + "num_tokens": 339856245.0, + "step": 13373 + }, + { + "epoch": 1.4687019547551066, + "grad_norm": 2.4813339710235596, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7135816812515259, + "num_tokens": 339877455.0, + "step": 13374 + }, + { + "epoch": 1.46881177245772, + "grad_norm": 2.2907886505126953, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7148957252502441, + "num_tokens": 339902874.0, + "step": 13375 + }, + { + "epoch": 1.4689215901603339, + "grad_norm": 2.3322107791900635, + "learning_rate": 1e-06, + "loss": 0.9738, + "mean_token_accuracy": 0.7060613632202148, + "num_tokens": 339929271.0, + "step": 13376 + }, + { + "epoch": 1.4690314078629476, + "grad_norm": 2.044529914855957, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7155268788337708, + "num_tokens": 339960138.0, + "step": 13377 + }, + { + "epoch": 1.4691412255655611, + "grad_norm": 2.3394253253936768, + "learning_rate": 1e-06, + "loss": 0.7756, + "mean_token_accuracy": 0.7489423155784607, + "num_tokens": 339981764.0, + "step": 13378 + }, + { + "epoch": 1.469251043268175, + "grad_norm": 2.1894447803497314, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7182590365409851, + "num_tokens": 340007467.0, + "step": 13379 + }, + { + "epoch": 1.4693608609707884, + "grad_norm": 2.2661020755767822, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.7372217178344727, + "num_tokens": 340033279.0, + "step": 13380 + }, + { + "epoch": 1.4694706786734022, + "grad_norm": 2.2828433513641357, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7179825305938721, + "num_tokens": 340057835.0, + "step": 13381 + }, + { + "epoch": 1.4695804963760157, + "grad_norm": 2.3480398654937744, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7282307147979736, + "num_tokens": 340081866.0, + "step": 13382 + }, + { + "epoch": 1.4696903140786295, + "grad_norm": 2.0558056831359863, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7172398567199707, + "num_tokens": 340113022.0, + "step": 13383 + }, + { + "epoch": 1.469800131781243, + "grad_norm": 1.9387235641479492, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7233281135559082, + "num_tokens": 340145252.0, + "step": 13384 + }, + { + "epoch": 1.4699099494838568, + "grad_norm": 2.1045451164245605, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7171063423156738, + "num_tokens": 340171621.0, + "step": 13385 + }, + { + "epoch": 1.4700197671864705, + "grad_norm": 2.0331013202667236, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.6975691318511963, + "num_tokens": 340203367.0, + "step": 13386 + }, + { + "epoch": 1.470129584889084, + "grad_norm": 2.2767093181610107, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.717081606388092, + "num_tokens": 340229790.0, + "step": 13387 + }, + { + "epoch": 1.4702394025916978, + "grad_norm": 2.2096309661865234, + "learning_rate": 1e-06, + "loss": 0.8552, + "mean_token_accuracy": 0.7324657440185547, + "num_tokens": 340254157.0, + "step": 13388 + }, + { + "epoch": 1.4703492202943114, + "grad_norm": 2.0393643379211426, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7056142687797546, + "num_tokens": 340287218.0, + "step": 13389 + }, + { + "epoch": 1.4704590379969251, + "grad_norm": 2.3850085735321045, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7251062393188477, + "num_tokens": 340312007.0, + "step": 13390 + }, + { + "epoch": 1.4705688556995389, + "grad_norm": 2.4231374263763428, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7112035155296326, + "num_tokens": 340337303.0, + "step": 13391 + }, + { + "epoch": 1.4706786734021524, + "grad_norm": 2.1186647415161133, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7200354933738708, + "num_tokens": 340365751.0, + "step": 13392 + }, + { + "epoch": 1.4707884911047662, + "grad_norm": 2.322205066680908, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7130019068717957, + "num_tokens": 340389454.0, + "step": 13393 + }, + { + "epoch": 1.4708983088073797, + "grad_norm": 2.0377848148345947, + "learning_rate": 1e-06, + "loss": 1.008, + "mean_token_accuracy": 0.695030927658081, + "num_tokens": 340420491.0, + "step": 13394 + }, + { + "epoch": 1.4710081265099935, + "grad_norm": 2.7382216453552246, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7365220785140991, + "num_tokens": 340439159.0, + "step": 13395 + }, + { + "epoch": 1.471117944212607, + "grad_norm": 2.154590606689453, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7139067053794861, + "num_tokens": 340467585.0, + "step": 13396 + }, + { + "epoch": 1.4712277619152208, + "grad_norm": 2.556180000305176, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7195268869400024, + "num_tokens": 340488544.0, + "step": 13397 + }, + { + "epoch": 1.4713375796178343, + "grad_norm": 2.1622226238250732, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7162214517593384, + "num_tokens": 340515051.0, + "step": 13398 + }, + { + "epoch": 1.471447397320448, + "grad_norm": 2.6472554206848145, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.7221208810806274, + "num_tokens": 340533227.0, + "step": 13399 + }, + { + "epoch": 1.4715572150230618, + "grad_norm": 1.988856554031372, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7135785222053528, + "num_tokens": 340564729.0, + "step": 13400 + }, + { + "epoch": 1.4716670327256753, + "grad_norm": 2.5487139225006104, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7354381084442139, + "num_tokens": 340584852.0, + "step": 13401 + }, + { + "epoch": 1.471776850428289, + "grad_norm": 2.1687824726104736, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7241696119308472, + "num_tokens": 340611656.0, + "step": 13402 + }, + { + "epoch": 1.4718866681309026, + "grad_norm": 2.44856858253479, + "learning_rate": 1e-06, + "loss": 0.7453, + "mean_token_accuracy": 0.7675607204437256, + "num_tokens": 340632809.0, + "step": 13403 + }, + { + "epoch": 1.4719964858335164, + "grad_norm": 2.4265682697296143, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7076356410980225, + "num_tokens": 340657614.0, + "step": 13404 + }, + { + "epoch": 1.4721063035361301, + "grad_norm": 2.3179681301116943, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7121721506118774, + "num_tokens": 340682295.0, + "step": 13405 + }, + { + "epoch": 1.4722161212387437, + "grad_norm": 2.6137278079986572, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7148537039756775, + "num_tokens": 340703085.0, + "step": 13406 + }, + { + "epoch": 1.4723259389413572, + "grad_norm": 2.38826322555542, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7262066602706909, + "num_tokens": 340726833.0, + "step": 13407 + }, + { + "epoch": 1.472435756643971, + "grad_norm": 2.204346179962158, + "learning_rate": 1e-06, + "loss": 0.9714, + "mean_token_accuracy": 0.6970899701118469, + "num_tokens": 340757653.0, + "step": 13408 + }, + { + "epoch": 1.4725455743465847, + "grad_norm": 2.0709526538848877, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7283835411071777, + "num_tokens": 340784936.0, + "step": 13409 + }, + { + "epoch": 1.4726553920491983, + "grad_norm": 2.022365093231201, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7142313718795776, + "num_tokens": 340815586.0, + "step": 13410 + }, + { + "epoch": 1.472765209751812, + "grad_norm": 2.3312275409698486, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7198997735977173, + "num_tokens": 340838538.0, + "step": 13411 + }, + { + "epoch": 1.4728750274544256, + "grad_norm": 1.8409569263458252, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7129770517349243, + "num_tokens": 340871504.0, + "step": 13412 + }, + { + "epoch": 1.4729848451570393, + "grad_norm": 2.108306884765625, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.7280178666114807, + "num_tokens": 340899447.0, + "step": 13413 + }, + { + "epoch": 1.473094662859653, + "grad_norm": 2.042212724685669, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7184613943099976, + "num_tokens": 340927265.0, + "step": 13414 + }, + { + "epoch": 1.4732044805622666, + "grad_norm": 2.072002649307251, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7123630046844482, + "num_tokens": 340957799.0, + "step": 13415 + }, + { + "epoch": 1.4733142982648804, + "grad_norm": 2.2515432834625244, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.745099663734436, + "num_tokens": 340982826.0, + "step": 13416 + }, + { + "epoch": 1.473424115967494, + "grad_norm": 2.348496198654175, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7411003708839417, + "num_tokens": 341005601.0, + "step": 13417 + }, + { + "epoch": 1.4735339336701077, + "grad_norm": 2.1048107147216797, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7179750800132751, + "num_tokens": 341032486.0, + "step": 13418 + }, + { + "epoch": 1.4736437513727214, + "grad_norm": 1.9735580682754517, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7131178379058838, + "num_tokens": 341064078.0, + "step": 13419 + }, + { + "epoch": 1.473753569075335, + "grad_norm": 2.1133522987365723, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7160249948501587, + "num_tokens": 341090396.0, + "step": 13420 + }, + { + "epoch": 1.4738633867779485, + "grad_norm": 2.3637471199035645, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7171786427497864, + "num_tokens": 341114264.0, + "step": 13421 + }, + { + "epoch": 1.4739732044805622, + "grad_norm": 2.0776684284210205, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7064579725265503, + "num_tokens": 341144421.0, + "step": 13422 + }, + { + "epoch": 1.474083022183176, + "grad_norm": 2.7952840328216553, + "learning_rate": 1e-06, + "loss": 0.8258, + "mean_token_accuracy": 0.736814022064209, + "num_tokens": 341160046.0, + "step": 13423 + }, + { + "epoch": 1.4741928398857895, + "grad_norm": 2.5558226108551025, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7200908660888672, + "num_tokens": 341181873.0, + "step": 13424 + }, + { + "epoch": 1.4743026575884033, + "grad_norm": 1.9904190301895142, + "learning_rate": 1e-06, + "loss": 1.0113, + "mean_token_accuracy": 0.6825448870658875, + "num_tokens": 341214766.0, + "step": 13425 + }, + { + "epoch": 1.4744124752910168, + "grad_norm": 2.2731261253356934, + "learning_rate": 1e-06, + "loss": 0.9663, + "mean_token_accuracy": 0.7027249932289124, + "num_tokens": 341241250.0, + "step": 13426 + }, + { + "epoch": 1.4745222929936306, + "grad_norm": 2.4826529026031494, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7134031057357788, + "num_tokens": 341263757.0, + "step": 13427 + }, + { + "epoch": 1.4746321106962443, + "grad_norm": 2.1239068508148193, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.716541051864624, + "num_tokens": 341292241.0, + "step": 13428 + }, + { + "epoch": 1.4747419283988579, + "grad_norm": 2.072871208190918, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7226418852806091, + "num_tokens": 341320461.0, + "step": 13429 + }, + { + "epoch": 1.4748517461014716, + "grad_norm": 2.278954029083252, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.705105721950531, + "num_tokens": 341347050.0, + "step": 13430 + }, + { + "epoch": 1.4749615638040852, + "grad_norm": 2.2778515815734863, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7221947908401489, + "num_tokens": 341371742.0, + "step": 13431 + }, + { + "epoch": 1.475071381506699, + "grad_norm": 2.5397138595581055, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7197636365890503, + "num_tokens": 341391903.0, + "step": 13432 + }, + { + "epoch": 1.4751811992093127, + "grad_norm": 2.279210090637207, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7120782136917114, + "num_tokens": 341416665.0, + "step": 13433 + }, + { + "epoch": 1.4752910169119262, + "grad_norm": 2.0019893646240234, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.7023434638977051, + "num_tokens": 341447161.0, + "step": 13434 + }, + { + "epoch": 1.4754008346145397, + "grad_norm": 2.4893274307250977, + "learning_rate": 1e-06, + "loss": 0.804, + "mean_token_accuracy": 0.7439851760864258, + "num_tokens": 341467662.0, + "step": 13435 + }, + { + "epoch": 1.4755106523171535, + "grad_norm": 2.4529242515563965, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7343639731407166, + "num_tokens": 341489016.0, + "step": 13436 + }, + { + "epoch": 1.4756204700197673, + "grad_norm": 2.2088119983673096, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.716711163520813, + "num_tokens": 341515453.0, + "step": 13437 + }, + { + "epoch": 1.4757302877223808, + "grad_norm": 2.4028520584106445, + "learning_rate": 1e-06, + "loss": 0.7937, + "mean_token_accuracy": 0.744507908821106, + "num_tokens": 341536817.0, + "step": 13438 + }, + { + "epoch": 1.4758401054249946, + "grad_norm": 2.094419240951538, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7031615972518921, + "num_tokens": 341565281.0, + "step": 13439 + }, + { + "epoch": 1.475949923127608, + "grad_norm": 2.0344204902648926, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7239282727241516, + "num_tokens": 341596783.0, + "step": 13440 + }, + { + "epoch": 1.4760597408302218, + "grad_norm": 2.5245254039764404, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7201946973800659, + "num_tokens": 341618111.0, + "step": 13441 + }, + { + "epoch": 1.4761695585328356, + "grad_norm": 2.1060547828674316, + "learning_rate": 1e-06, + "loss": 1.0171, + "mean_token_accuracy": 0.6902040839195251, + "num_tokens": 341648836.0, + "step": 13442 + }, + { + "epoch": 1.4762793762354491, + "grad_norm": 2.4961209297180176, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7242550849914551, + "num_tokens": 341671006.0, + "step": 13443 + }, + { + "epoch": 1.476389193938063, + "grad_norm": 2.060297966003418, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7305247783660889, + "num_tokens": 341700241.0, + "step": 13444 + }, + { + "epoch": 1.4764990116406764, + "grad_norm": 2.4552500247955322, + "learning_rate": 1e-06, + "loss": 0.9523, + "mean_token_accuracy": 0.7076328992843628, + "num_tokens": 341724494.0, + "step": 13445 + }, + { + "epoch": 1.4766088293432902, + "grad_norm": 2.312962293624878, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7185624837875366, + "num_tokens": 341751039.0, + "step": 13446 + }, + { + "epoch": 1.4767186470459037, + "grad_norm": 2.5277554988861084, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7313293814659119, + "num_tokens": 341771151.0, + "step": 13447 + }, + { + "epoch": 1.4768284647485175, + "grad_norm": 2.258317232131958, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.720758318901062, + "num_tokens": 341796863.0, + "step": 13448 + }, + { + "epoch": 1.476938282451131, + "grad_norm": 2.2137486934661865, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7277059555053711, + "num_tokens": 341823483.0, + "step": 13449 + }, + { + "epoch": 1.4770481001537448, + "grad_norm": 2.791154623031616, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.7419012784957886, + "num_tokens": 341840645.0, + "step": 13450 + }, + { + "epoch": 1.4771579178563585, + "grad_norm": 2.4311676025390625, + "learning_rate": 1e-06, + "loss": 0.9581, + "mean_token_accuracy": 0.7173751592636108, + "num_tokens": 341864709.0, + "step": 13451 + }, + { + "epoch": 1.477267735558972, + "grad_norm": 2.4565417766571045, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7195643782615662, + "num_tokens": 341886277.0, + "step": 13452 + }, + { + "epoch": 1.4773775532615858, + "grad_norm": 2.5179898738861084, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7294566035270691, + "num_tokens": 341906891.0, + "step": 13453 + }, + { + "epoch": 1.4774873709641994, + "grad_norm": 2.086296796798706, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.7026751041412354, + "num_tokens": 341935457.0, + "step": 13454 + }, + { + "epoch": 1.477597188666813, + "grad_norm": 2.639235496520996, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.7252115607261658, + "num_tokens": 341954752.0, + "step": 13455 + }, + { + "epoch": 1.4777070063694269, + "grad_norm": 2.6391618251800537, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7063926458358765, + "num_tokens": 341974964.0, + "step": 13456 + }, + { + "epoch": 1.4778168240720404, + "grad_norm": 1.9593886137008667, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7109909653663635, + "num_tokens": 342005607.0, + "step": 13457 + }, + { + "epoch": 1.477926641774654, + "grad_norm": 2.1715986728668213, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7295076847076416, + "num_tokens": 342031735.0, + "step": 13458 + }, + { + "epoch": 1.4780364594772677, + "grad_norm": 2.253894329071045, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.721428632736206, + "num_tokens": 342055621.0, + "step": 13459 + }, + { + "epoch": 1.4781462771798815, + "grad_norm": 1.8242015838623047, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.705254077911377, + "num_tokens": 342091730.0, + "step": 13460 + }, + { + "epoch": 1.478256094882495, + "grad_norm": 2.166159152984619, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7010527849197388, + "num_tokens": 342118496.0, + "step": 13461 + }, + { + "epoch": 1.4783659125851087, + "grad_norm": 2.232905864715576, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7037989497184753, + "num_tokens": 342144853.0, + "step": 13462 + }, + { + "epoch": 1.4784757302877223, + "grad_norm": 2.0778486728668213, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7176930904388428, + "num_tokens": 342173023.0, + "step": 13463 + }, + { + "epoch": 1.478585547990336, + "grad_norm": 2.2778472900390625, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7084506154060364, + "num_tokens": 342198767.0, + "step": 13464 + }, + { + "epoch": 1.4786953656929498, + "grad_norm": 2.4618215560913086, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7331559658050537, + "num_tokens": 342221055.0, + "step": 13465 + }, + { + "epoch": 1.4788051833955633, + "grad_norm": 2.1515755653381348, + "learning_rate": 1e-06, + "loss": 0.9989, + "mean_token_accuracy": 0.6936843395233154, + "num_tokens": 342249072.0, + "step": 13466 + }, + { + "epoch": 1.478915001098177, + "grad_norm": 2.1544041633605957, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7386641502380371, + "num_tokens": 342275464.0, + "step": 13467 + }, + { + "epoch": 1.4790248188007906, + "grad_norm": 2.2802090644836426, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7215256094932556, + "num_tokens": 342300453.0, + "step": 13468 + }, + { + "epoch": 1.4791346365034044, + "grad_norm": 2.204206943511963, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.7059964537620544, + "num_tokens": 342327728.0, + "step": 13469 + }, + { + "epoch": 1.4792444542060181, + "grad_norm": 2.0799734592437744, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7069348692893982, + "num_tokens": 342359952.0, + "step": 13470 + }, + { + "epoch": 1.4793542719086317, + "grad_norm": 2.100010871887207, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7293247580528259, + "num_tokens": 342386750.0, + "step": 13471 + }, + { + "epoch": 1.4794640896112452, + "grad_norm": 2.436617851257324, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7048648595809937, + "num_tokens": 342412777.0, + "step": 13472 + }, + { + "epoch": 1.479573907313859, + "grad_norm": 2.4344987869262695, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.736847996711731, + "num_tokens": 342436180.0, + "step": 13473 + }, + { + "epoch": 1.4796837250164727, + "grad_norm": 2.422481060028076, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7412111759185791, + "num_tokens": 342456817.0, + "step": 13474 + }, + { + "epoch": 1.4797935427190863, + "grad_norm": 2.070880174636841, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.72234046459198, + "num_tokens": 342488726.0, + "step": 13475 + }, + { + "epoch": 1.4799033604217, + "grad_norm": 2.1718642711639404, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7321977615356445, + "num_tokens": 342513897.0, + "step": 13476 + }, + { + "epoch": 1.4800131781243135, + "grad_norm": 2.782223701477051, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7046195268630981, + "num_tokens": 342533391.0, + "step": 13477 + }, + { + "epoch": 1.4801229958269273, + "grad_norm": 2.2272891998291016, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.720454216003418, + "num_tokens": 342562241.0, + "step": 13478 + }, + { + "epoch": 1.480232813529541, + "grad_norm": 2.147202730178833, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7238174676895142, + "num_tokens": 342587581.0, + "step": 13479 + }, + { + "epoch": 1.4803426312321546, + "grad_norm": 2.5845367908477783, + "learning_rate": 1e-06, + "loss": 0.8371, + "mean_token_accuracy": 0.7366623878479004, + "num_tokens": 342607165.0, + "step": 13480 + }, + { + "epoch": 1.4804524489347684, + "grad_norm": 2.5864176750183105, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.730103611946106, + "num_tokens": 342627813.0, + "step": 13481 + }, + { + "epoch": 1.4805622666373819, + "grad_norm": 2.138585329055786, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6921202540397644, + "num_tokens": 342657883.0, + "step": 13482 + }, + { + "epoch": 1.4806720843399956, + "grad_norm": 2.5050060749053955, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7214646339416504, + "num_tokens": 342678892.0, + "step": 13483 + }, + { + "epoch": 1.4807819020426094, + "grad_norm": 2.1337952613830566, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7221410274505615, + "num_tokens": 342708743.0, + "step": 13484 + }, + { + "epoch": 1.480891719745223, + "grad_norm": 2.3205673694610596, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7391578555107117, + "num_tokens": 342733069.0, + "step": 13485 + }, + { + "epoch": 1.4810015374478365, + "grad_norm": 2.395998239517212, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7306249737739563, + "num_tokens": 342755322.0, + "step": 13486 + }, + { + "epoch": 1.4811113551504502, + "grad_norm": 2.2392654418945312, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7102571725845337, + "num_tokens": 342781645.0, + "step": 13487 + }, + { + "epoch": 1.481221172853064, + "grad_norm": 2.3706021308898926, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7295361757278442, + "num_tokens": 342804383.0, + "step": 13488 + }, + { + "epoch": 1.4813309905556775, + "grad_norm": 2.216395139694214, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7230387926101685, + "num_tokens": 342827843.0, + "step": 13489 + }, + { + "epoch": 1.4814408082582913, + "grad_norm": 2.2004082202911377, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7187685966491699, + "num_tokens": 342851592.0, + "step": 13490 + }, + { + "epoch": 1.4815506259609048, + "grad_norm": 2.4106929302215576, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7268333435058594, + "num_tokens": 342874837.0, + "step": 13491 + }, + { + "epoch": 1.4816604436635186, + "grad_norm": 2.351865291595459, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7230437994003296, + "num_tokens": 342898363.0, + "step": 13492 + }, + { + "epoch": 1.4817702613661323, + "grad_norm": 2.177371025085449, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7140058279037476, + "num_tokens": 342925272.0, + "step": 13493 + }, + { + "epoch": 1.4818800790687459, + "grad_norm": 2.7679760456085205, + "learning_rate": 1e-06, + "loss": 0.8654, + "mean_token_accuracy": 0.73293536901474, + "num_tokens": 342943908.0, + "step": 13494 + }, + { + "epoch": 1.4819898967713596, + "grad_norm": 2.4003727436065674, + "learning_rate": 1e-06, + "loss": 0.7387, + "mean_token_accuracy": 0.7577813267707825, + "num_tokens": 342963315.0, + "step": 13495 + }, + { + "epoch": 1.4820997144739732, + "grad_norm": 2.1440513134002686, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7171355485916138, + "num_tokens": 342990266.0, + "step": 13496 + }, + { + "epoch": 1.482209532176587, + "grad_norm": 2.1712968349456787, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7279994487762451, + "num_tokens": 343017186.0, + "step": 13497 + }, + { + "epoch": 1.4823193498792007, + "grad_norm": 2.3531532287597656, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7399758100509644, + "num_tokens": 343039999.0, + "step": 13498 + }, + { + "epoch": 1.4824291675818142, + "grad_norm": 2.279827356338501, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7173923254013062, + "num_tokens": 343066279.0, + "step": 13499 + }, + { + "epoch": 1.4825389852844277, + "grad_norm": 2.1735870838165283, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7083946466445923, + "num_tokens": 343093913.0, + "step": 13500 + }, + { + "epoch": 1.4826488029870415, + "grad_norm": 2.3241217136383057, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7201759219169617, + "num_tokens": 343117884.0, + "step": 13501 + }, + { + "epoch": 1.4827586206896552, + "grad_norm": 2.172333240509033, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7236390113830566, + "num_tokens": 343145609.0, + "step": 13502 + }, + { + "epoch": 1.4828684383922688, + "grad_norm": 2.3852086067199707, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7231241464614868, + "num_tokens": 343169330.0, + "step": 13503 + }, + { + "epoch": 1.4829782560948825, + "grad_norm": 2.136561155319214, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7087061405181885, + "num_tokens": 343198477.0, + "step": 13504 + }, + { + "epoch": 1.483088073797496, + "grad_norm": 2.0178210735321045, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7181051969528198, + "num_tokens": 343229279.0, + "step": 13505 + }, + { + "epoch": 1.4831978915001098, + "grad_norm": 2.078202962875366, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7088755369186401, + "num_tokens": 343258422.0, + "step": 13506 + }, + { + "epoch": 1.4833077092027236, + "grad_norm": 2.082697629928589, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.6958187222480774, + "num_tokens": 343289995.0, + "step": 13507 + }, + { + "epoch": 1.4834175269053371, + "grad_norm": 2.289395332336426, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7103526592254639, + "num_tokens": 343313807.0, + "step": 13508 + }, + { + "epoch": 1.4835273446079509, + "grad_norm": 2.066793918609619, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7114503383636475, + "num_tokens": 343344322.0, + "step": 13509 + }, + { + "epoch": 1.4836371623105644, + "grad_norm": 2.224278211593628, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7347549200057983, + "num_tokens": 343370471.0, + "step": 13510 + }, + { + "epoch": 1.4837469800131782, + "grad_norm": 1.9937231540679932, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7192195653915405, + "num_tokens": 343400510.0, + "step": 13511 + }, + { + "epoch": 1.4838567977157917, + "grad_norm": 2.8659682273864746, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7318639755249023, + "num_tokens": 343419035.0, + "step": 13512 + }, + { + "epoch": 1.4839666154184055, + "grad_norm": 2.4355127811431885, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7339133620262146, + "num_tokens": 343441671.0, + "step": 13513 + }, + { + "epoch": 1.484076433121019, + "grad_norm": 2.2182037830352783, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7324166297912598, + "num_tokens": 343468085.0, + "step": 13514 + }, + { + "epoch": 1.4841862508236328, + "grad_norm": 2.2397804260253906, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7185401320457458, + "num_tokens": 343493534.0, + "step": 13515 + }, + { + "epoch": 1.4842960685262465, + "grad_norm": 2.501230478286743, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7167264223098755, + "num_tokens": 343516836.0, + "step": 13516 + }, + { + "epoch": 1.48440588622886, + "grad_norm": 2.5265002250671387, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7383396625518799, + "num_tokens": 343538189.0, + "step": 13517 + }, + { + "epoch": 1.4845157039314738, + "grad_norm": 2.201880931854248, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7156063318252563, + "num_tokens": 343563540.0, + "step": 13518 + }, + { + "epoch": 1.4846255216340873, + "grad_norm": 2.280977487564087, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.723675549030304, + "num_tokens": 343587768.0, + "step": 13519 + }, + { + "epoch": 1.484735339336701, + "grad_norm": 2.3937535285949707, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7356048822402954, + "num_tokens": 343610515.0, + "step": 13520 + }, + { + "epoch": 1.4848451570393149, + "grad_norm": 2.4392073154449463, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7299662828445435, + "num_tokens": 343632378.0, + "step": 13521 + }, + { + "epoch": 1.4849549747419284, + "grad_norm": 2.1054766178131104, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7211414575576782, + "num_tokens": 343660230.0, + "step": 13522 + }, + { + "epoch": 1.485064792444542, + "grad_norm": 2.2238214015960693, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7137463688850403, + "num_tokens": 343688141.0, + "step": 13523 + }, + { + "epoch": 1.4851746101471557, + "grad_norm": 2.0921685695648193, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7142391204833984, + "num_tokens": 343714822.0, + "step": 13524 + }, + { + "epoch": 1.4852844278497694, + "grad_norm": 2.2997703552246094, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7021239399909973, + "num_tokens": 343742489.0, + "step": 13525 + }, + { + "epoch": 1.485394245552383, + "grad_norm": 2.3925046920776367, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7101415395736694, + "num_tokens": 343765937.0, + "step": 13526 + }, + { + "epoch": 1.4855040632549967, + "grad_norm": 2.3389554023742676, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7274764776229858, + "num_tokens": 343788828.0, + "step": 13527 + }, + { + "epoch": 1.4856138809576103, + "grad_norm": 2.2895305156707764, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.704530656337738, + "num_tokens": 343814766.0, + "step": 13528 + }, + { + "epoch": 1.485723698660224, + "grad_norm": 2.3429579734802246, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7063345313072205, + "num_tokens": 343840018.0, + "step": 13529 + }, + { + "epoch": 1.4858335163628378, + "grad_norm": 2.283643960952759, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7069039344787598, + "num_tokens": 343865047.0, + "step": 13530 + }, + { + "epoch": 1.4859433340654513, + "grad_norm": 2.425096273422241, + "learning_rate": 1e-06, + "loss": 0.912, + "mean_token_accuracy": 0.7175268530845642, + "num_tokens": 343888218.0, + "step": 13531 + }, + { + "epoch": 1.486053151768065, + "grad_norm": 2.457811117172241, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7210652828216553, + "num_tokens": 343909015.0, + "step": 13532 + }, + { + "epoch": 1.4861629694706786, + "grad_norm": 2.4228484630584717, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7103841304779053, + "num_tokens": 343931448.0, + "step": 13533 + }, + { + "epoch": 1.4862727871732924, + "grad_norm": 2.1537554264068604, + "learning_rate": 1e-06, + "loss": 0.9044, + "mean_token_accuracy": 0.7188625335693359, + "num_tokens": 343961159.0, + "step": 13534 + }, + { + "epoch": 1.4863826048759061, + "grad_norm": 2.3535640239715576, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7377855181694031, + "num_tokens": 343985638.0, + "step": 13535 + }, + { + "epoch": 1.4864924225785197, + "grad_norm": 2.2463724613189697, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7038238644599915, + "num_tokens": 344014401.0, + "step": 13536 + }, + { + "epoch": 1.4866022402811332, + "grad_norm": 2.2134292125701904, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7339571714401245, + "num_tokens": 344037223.0, + "step": 13537 + }, + { + "epoch": 1.486712057983747, + "grad_norm": 2.247314691543579, + "learning_rate": 1e-06, + "loss": 0.7802, + "mean_token_accuracy": 0.7600813508033752, + "num_tokens": 344062678.0, + "step": 13538 + }, + { + "epoch": 1.4868218756863607, + "grad_norm": 2.1078693866729736, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7228220701217651, + "num_tokens": 344091970.0, + "step": 13539 + }, + { + "epoch": 1.4869316933889742, + "grad_norm": 2.060560703277588, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.704124391078949, + "num_tokens": 344123395.0, + "step": 13540 + }, + { + "epoch": 1.487041511091588, + "grad_norm": 2.1059730052948, + "learning_rate": 1e-06, + "loss": 0.986, + "mean_token_accuracy": 0.6945439577102661, + "num_tokens": 344152741.0, + "step": 13541 + }, + { + "epoch": 1.4871513287942015, + "grad_norm": 2.25093936920166, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7155050039291382, + "num_tokens": 344176737.0, + "step": 13542 + }, + { + "epoch": 1.4872611464968153, + "grad_norm": 1.9719359874725342, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7156482934951782, + "num_tokens": 344209483.0, + "step": 13543 + }, + { + "epoch": 1.487370964199429, + "grad_norm": 2.505704879760742, + "learning_rate": 1e-06, + "loss": 0.7981, + "mean_token_accuracy": 0.7522832155227661, + "num_tokens": 344229965.0, + "step": 13544 + }, + { + "epoch": 1.4874807819020426, + "grad_norm": 2.350907325744629, + "learning_rate": 1e-06, + "loss": 0.7565, + "mean_token_accuracy": 0.7514744400978088, + "num_tokens": 344252599.0, + "step": 13545 + }, + { + "epoch": 1.4875905996046563, + "grad_norm": 2.0606086254119873, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7098312377929688, + "num_tokens": 344281277.0, + "step": 13546 + }, + { + "epoch": 1.4877004173072699, + "grad_norm": 2.221133232116699, + "learning_rate": 1e-06, + "loss": 0.944, + "mean_token_accuracy": 0.7090599536895752, + "num_tokens": 344309006.0, + "step": 13547 + }, + { + "epoch": 1.4878102350098836, + "grad_norm": 2.4526679515838623, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.719740629196167, + "num_tokens": 344330480.0, + "step": 13548 + }, + { + "epoch": 1.4879200527124974, + "grad_norm": 2.314992666244507, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7096893787384033, + "num_tokens": 344355891.0, + "step": 13549 + }, + { + "epoch": 1.488029870415111, + "grad_norm": 2.45426082611084, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7225208282470703, + "num_tokens": 344377641.0, + "step": 13550 + }, + { + "epoch": 1.4881396881177245, + "grad_norm": 2.1868343353271484, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7184177041053772, + "num_tokens": 344403572.0, + "step": 13551 + }, + { + "epoch": 1.4882495058203382, + "grad_norm": 2.23360538482666, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7170343399047852, + "num_tokens": 344427654.0, + "step": 13552 + }, + { + "epoch": 1.488359323522952, + "grad_norm": 2.245502471923828, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7334980368614197, + "num_tokens": 344452481.0, + "step": 13553 + }, + { + "epoch": 1.4884691412255655, + "grad_norm": 2.463348627090454, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7253310084342957, + "num_tokens": 344474078.0, + "step": 13554 + }, + { + "epoch": 1.4885789589281793, + "grad_norm": 2.388700008392334, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7216380834579468, + "num_tokens": 344499276.0, + "step": 13555 + }, + { + "epoch": 1.4886887766307928, + "grad_norm": 2.3444695472717285, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7139514684677124, + "num_tokens": 344523562.0, + "step": 13556 + }, + { + "epoch": 1.4887985943334066, + "grad_norm": 2.0211706161499023, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7078657150268555, + "num_tokens": 344555175.0, + "step": 13557 + }, + { + "epoch": 1.4889084120360203, + "grad_norm": 2.375821828842163, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7139072418212891, + "num_tokens": 344579115.0, + "step": 13558 + }, + { + "epoch": 1.4890182297386338, + "grad_norm": 2.364990234375, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7173788547515869, + "num_tokens": 344602202.0, + "step": 13559 + }, + { + "epoch": 1.4891280474412476, + "grad_norm": 2.3355495929718018, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7180081605911255, + "num_tokens": 344625578.0, + "step": 13560 + }, + { + "epoch": 1.4892378651438611, + "grad_norm": 2.2662339210510254, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6901199221611023, + "num_tokens": 344652519.0, + "step": 13561 + }, + { + "epoch": 1.489347682846475, + "grad_norm": 2.189552068710327, + "learning_rate": 1e-06, + "loss": 0.954, + "mean_token_accuracy": 0.7081592082977295, + "num_tokens": 344679180.0, + "step": 13562 + }, + { + "epoch": 1.4894575005490884, + "grad_norm": 2.4609732627868652, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.7055842876434326, + "num_tokens": 344703405.0, + "step": 13563 + }, + { + "epoch": 1.4895673182517022, + "grad_norm": 2.522233247756958, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7205788493156433, + "num_tokens": 344726491.0, + "step": 13564 + }, + { + "epoch": 1.4896771359543157, + "grad_norm": 2.13775897026062, + "learning_rate": 1e-06, + "loss": 0.9615, + "mean_token_accuracy": 0.6973607540130615, + "num_tokens": 344755885.0, + "step": 13565 + }, + { + "epoch": 1.4897869536569295, + "grad_norm": 2.5269253253936768, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7257491946220398, + "num_tokens": 344777566.0, + "step": 13566 + }, + { + "epoch": 1.4898967713595432, + "grad_norm": 2.288841962814331, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7181776762008667, + "num_tokens": 344801891.0, + "step": 13567 + }, + { + "epoch": 1.4900065890621568, + "grad_norm": 2.017005443572998, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7304766178131104, + "num_tokens": 344833305.0, + "step": 13568 + }, + { + "epoch": 1.4901164067647705, + "grad_norm": 2.0579962730407715, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7204852104187012, + "num_tokens": 344863631.0, + "step": 13569 + }, + { + "epoch": 1.490226224467384, + "grad_norm": 2.285670042037964, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.721432626247406, + "num_tokens": 344889419.0, + "step": 13570 + }, + { + "epoch": 1.4903360421699978, + "grad_norm": 2.224564790725708, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7157389521598816, + "num_tokens": 344915987.0, + "step": 13571 + }, + { + "epoch": 1.4904458598726116, + "grad_norm": 2.651844024658203, + "learning_rate": 1e-06, + "loss": 0.7544, + "mean_token_accuracy": 0.7545421123504639, + "num_tokens": 344934879.0, + "step": 13572 + }, + { + "epoch": 1.4905556775752251, + "grad_norm": 1.9367867708206177, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7221871614456177, + "num_tokens": 344966191.0, + "step": 13573 + }, + { + "epoch": 1.4906654952778389, + "grad_norm": 2.329036235809326, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7127498388290405, + "num_tokens": 344991264.0, + "step": 13574 + }, + { + "epoch": 1.4907753129804524, + "grad_norm": 2.6511738300323486, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7085031270980835, + "num_tokens": 345010549.0, + "step": 13575 + }, + { + "epoch": 1.4908851306830662, + "grad_norm": 2.2226991653442383, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7134541273117065, + "num_tokens": 345035249.0, + "step": 13576 + }, + { + "epoch": 1.4909949483856797, + "grad_norm": 2.1596035957336426, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.7059929370880127, + "num_tokens": 345063365.0, + "step": 13577 + }, + { + "epoch": 1.4911047660882935, + "grad_norm": 2.0452609062194824, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7089398503303528, + "num_tokens": 345095870.0, + "step": 13578 + }, + { + "epoch": 1.491214583790907, + "grad_norm": 2.065944194793701, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7174389958381653, + "num_tokens": 345125319.0, + "step": 13579 + }, + { + "epoch": 1.4913244014935207, + "grad_norm": 2.347407102584839, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.6987102031707764, + "num_tokens": 345148856.0, + "step": 13580 + }, + { + "epoch": 1.4914342191961345, + "grad_norm": 2.612274408340454, + "learning_rate": 1e-06, + "loss": 0.7736, + "mean_token_accuracy": 0.7496176958084106, + "num_tokens": 345167817.0, + "step": 13581 + }, + { + "epoch": 1.491544036898748, + "grad_norm": 2.47800612449646, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7347419261932373, + "num_tokens": 345189479.0, + "step": 13582 + }, + { + "epoch": 1.4916538546013618, + "grad_norm": 2.3061091899871826, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7198966145515442, + "num_tokens": 345213834.0, + "step": 13583 + }, + { + "epoch": 1.4917636723039753, + "grad_norm": 2.440305471420288, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7360074520111084, + "num_tokens": 345236218.0, + "step": 13584 + }, + { + "epoch": 1.491873490006589, + "grad_norm": 2.254408597946167, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6941275596618652, + "num_tokens": 345262636.0, + "step": 13585 + }, + { + "epoch": 1.4919833077092028, + "grad_norm": 2.1727383136749268, + "learning_rate": 1e-06, + "loss": 0.8257, + "mean_token_accuracy": 0.7405268549919128, + "num_tokens": 345289016.0, + "step": 13586 + }, + { + "epoch": 1.4920931254118164, + "grad_norm": 2.447404623031616, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7216871380805969, + "num_tokens": 345312421.0, + "step": 13587 + }, + { + "epoch": 1.49220294311443, + "grad_norm": 2.367326021194458, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7208583354949951, + "num_tokens": 345336501.0, + "step": 13588 + }, + { + "epoch": 1.4923127608170437, + "grad_norm": 2.223435401916504, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7093002200126648, + "num_tokens": 345363288.0, + "step": 13589 + }, + { + "epoch": 1.4924225785196574, + "grad_norm": 2.2734086513519287, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.727108359336853, + "num_tokens": 345388667.0, + "step": 13590 + }, + { + "epoch": 1.492532396222271, + "grad_norm": 2.047584056854248, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7242599725723267, + "num_tokens": 345417304.0, + "step": 13591 + }, + { + "epoch": 1.4926422139248847, + "grad_norm": 2.384803295135498, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7420700192451477, + "num_tokens": 345439848.0, + "step": 13592 + }, + { + "epoch": 1.4927520316274983, + "grad_norm": 2.5249271392822266, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7307132482528687, + "num_tokens": 345460507.0, + "step": 13593 + }, + { + "epoch": 1.492861849330112, + "grad_norm": 2.2182750701904297, + "learning_rate": 1e-06, + "loss": 0.7749, + "mean_token_accuracy": 0.7527529001235962, + "num_tokens": 345486579.0, + "step": 13594 + }, + { + "epoch": 1.4929716670327258, + "grad_norm": 2.3445699214935303, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.7256803512573242, + "num_tokens": 345512014.0, + "step": 13595 + }, + { + "epoch": 1.4930814847353393, + "grad_norm": 2.234962224960327, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7152233123779297, + "num_tokens": 345537232.0, + "step": 13596 + }, + { + "epoch": 1.493191302437953, + "grad_norm": 2.31821346282959, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7391296029090881, + "num_tokens": 345560872.0, + "step": 13597 + }, + { + "epoch": 1.4933011201405666, + "grad_norm": 2.3391709327697754, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7344363331794739, + "num_tokens": 345584100.0, + "step": 13598 + }, + { + "epoch": 1.4934109378431804, + "grad_norm": 2.20436429977417, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7047057747840881, + "num_tokens": 345612785.0, + "step": 13599 + }, + { + "epoch": 1.493520755545794, + "grad_norm": 2.1502811908721924, + "learning_rate": 1e-06, + "loss": 1.0142, + "mean_token_accuracy": 0.6933727860450745, + "num_tokens": 345641108.0, + "step": 13600 + }, + { + "epoch": 1.4936305732484076, + "grad_norm": 2.6280629634857178, + "learning_rate": 1e-06, + "loss": 0.875, + "mean_token_accuracy": 0.7261354327201843, + "num_tokens": 345661117.0, + "step": 13601 + }, + { + "epoch": 1.4937403909510212, + "grad_norm": 2.0993893146514893, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.736523449420929, + "num_tokens": 345688104.0, + "step": 13602 + }, + { + "epoch": 1.493850208653635, + "grad_norm": 2.0660240650177, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7105128169059753, + "num_tokens": 345718861.0, + "step": 13603 + }, + { + "epoch": 1.4939600263562487, + "grad_norm": 2.4810426235198975, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7281795144081116, + "num_tokens": 345738654.0, + "step": 13604 + }, + { + "epoch": 1.4940698440588622, + "grad_norm": 2.1142797470092773, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7194763422012329, + "num_tokens": 345766567.0, + "step": 13605 + }, + { + "epoch": 1.494179661761476, + "grad_norm": 2.224907159805298, + "learning_rate": 1e-06, + "loss": 0.7914, + "mean_token_accuracy": 0.7557360529899597, + "num_tokens": 345790919.0, + "step": 13606 + }, + { + "epoch": 1.4942894794640895, + "grad_norm": 2.254068374633789, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7317357659339905, + "num_tokens": 345815193.0, + "step": 13607 + }, + { + "epoch": 1.4943992971667033, + "grad_norm": 2.2760021686553955, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.716166615486145, + "num_tokens": 345842149.0, + "step": 13608 + }, + { + "epoch": 1.494509114869317, + "grad_norm": 2.2736337184906006, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.725471019744873, + "num_tokens": 345869622.0, + "step": 13609 + }, + { + "epoch": 1.4946189325719306, + "grad_norm": 2.308483362197876, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.7090628147125244, + "num_tokens": 345894181.0, + "step": 13610 + }, + { + "epoch": 1.4947287502745443, + "grad_norm": 2.5090138912200928, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7121967077255249, + "num_tokens": 345914677.0, + "step": 13611 + }, + { + "epoch": 1.4948385679771579, + "grad_norm": 2.074256658554077, + "learning_rate": 1e-06, + "loss": 1.0216, + "mean_token_accuracy": 0.6926815509796143, + "num_tokens": 345944203.0, + "step": 13612 + }, + { + "epoch": 1.4949483856797716, + "grad_norm": 2.118736982345581, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.701869010925293, + "num_tokens": 345973880.0, + "step": 13613 + }, + { + "epoch": 1.4950582033823854, + "grad_norm": 2.2568867206573486, + "learning_rate": 1e-06, + "loss": 0.9227, + "mean_token_accuracy": 0.7186571955680847, + "num_tokens": 345998770.0, + "step": 13614 + }, + { + "epoch": 1.495168021084999, + "grad_norm": 2.196380615234375, + "learning_rate": 1e-06, + "loss": 1.0107, + "mean_token_accuracy": 0.697939395904541, + "num_tokens": 346026889.0, + "step": 13615 + }, + { + "epoch": 1.4952778387876124, + "grad_norm": 2.461125135421753, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.726110577583313, + "num_tokens": 346049207.0, + "step": 13616 + }, + { + "epoch": 1.4953876564902262, + "grad_norm": 2.235987901687622, + "learning_rate": 1e-06, + "loss": 0.9855, + "mean_token_accuracy": 0.6950520277023315, + "num_tokens": 346076287.0, + "step": 13617 + }, + { + "epoch": 1.49549747419284, + "grad_norm": 2.2375729084014893, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7323254346847534, + "num_tokens": 346102334.0, + "step": 13618 + }, + { + "epoch": 1.4956072918954535, + "grad_norm": 2.0866236686706543, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7037146091461182, + "num_tokens": 346130875.0, + "step": 13619 + }, + { + "epoch": 1.4957171095980673, + "grad_norm": 2.3294479846954346, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.6985040903091431, + "num_tokens": 346157087.0, + "step": 13620 + }, + { + "epoch": 1.4958269273006808, + "grad_norm": 2.3362014293670654, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7195928692817688, + "num_tokens": 346182589.0, + "step": 13621 + }, + { + "epoch": 1.4959367450032945, + "grad_norm": 2.324244499206543, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7340461015701294, + "num_tokens": 346206908.0, + "step": 13622 + }, + { + "epoch": 1.4960465627059083, + "grad_norm": 2.177330493927002, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7257300615310669, + "num_tokens": 346234681.0, + "step": 13623 + }, + { + "epoch": 1.4961563804085218, + "grad_norm": 2.3546202182769775, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7424542307853699, + "num_tokens": 346258348.0, + "step": 13624 + }, + { + "epoch": 1.4962661981111356, + "grad_norm": 2.03183650970459, + "learning_rate": 1e-06, + "loss": 1.0255, + "mean_token_accuracy": 0.6856909990310669, + "num_tokens": 346290851.0, + "step": 13625 + }, + { + "epoch": 1.4963760158137491, + "grad_norm": 2.060124158859253, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7139201760292053, + "num_tokens": 346320485.0, + "step": 13626 + }, + { + "epoch": 1.4964858335163629, + "grad_norm": 2.105156660079956, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7114877700805664, + "num_tokens": 346351144.0, + "step": 13627 + }, + { + "epoch": 1.4965956512189764, + "grad_norm": 2.4807980060577393, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7259487509727478, + "num_tokens": 346373224.0, + "step": 13628 + }, + { + "epoch": 1.4967054689215902, + "grad_norm": 2.5900697708129883, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7200583219528198, + "num_tokens": 346394488.0, + "step": 13629 + }, + { + "epoch": 1.4968152866242037, + "grad_norm": 2.82136869430542, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7266101837158203, + "num_tokens": 346412140.0, + "step": 13630 + }, + { + "epoch": 1.4969251043268175, + "grad_norm": 2.0841658115386963, + "learning_rate": 1e-06, + "loss": 1.0023, + "mean_token_accuracy": 0.6835678815841675, + "num_tokens": 346441050.0, + "step": 13631 + }, + { + "epoch": 1.4970349220294312, + "grad_norm": 2.3702685832977295, + "learning_rate": 1e-06, + "loss": 0.7562, + "mean_token_accuracy": 0.7548625469207764, + "num_tokens": 346462435.0, + "step": 13632 + }, + { + "epoch": 1.4971447397320448, + "grad_norm": 2.5753085613250732, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7197551727294922, + "num_tokens": 346484332.0, + "step": 13633 + }, + { + "epoch": 1.4972545574346585, + "grad_norm": 2.6516764163970947, + "learning_rate": 1e-06, + "loss": 0.7939, + "mean_token_accuracy": 0.7418263554573059, + "num_tokens": 346503522.0, + "step": 13634 + }, + { + "epoch": 1.497364375137272, + "grad_norm": 2.043417453765869, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7201014161109924, + "num_tokens": 346534435.0, + "step": 13635 + }, + { + "epoch": 1.4974741928398858, + "grad_norm": 2.368584156036377, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7347126007080078, + "num_tokens": 346557246.0, + "step": 13636 + }, + { + "epoch": 1.4975840105424996, + "grad_norm": 2.4509499073028564, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7264785766601562, + "num_tokens": 346579911.0, + "step": 13637 + }, + { + "epoch": 1.497693828245113, + "grad_norm": 2.115959644317627, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7192169427871704, + "num_tokens": 346609828.0, + "step": 13638 + }, + { + "epoch": 1.4978036459477266, + "grad_norm": 2.1085808277130127, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.6976732611656189, + "num_tokens": 346638882.0, + "step": 13639 + }, + { + "epoch": 1.4979134636503404, + "grad_norm": 2.199147939682007, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7361716032028198, + "num_tokens": 346666576.0, + "step": 13640 + }, + { + "epoch": 1.4980232813529542, + "grad_norm": 1.975532054901123, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.711471438407898, + "num_tokens": 346700374.0, + "step": 13641 + }, + { + "epoch": 1.4981330990555677, + "grad_norm": 2.7344977855682373, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7188395261764526, + "num_tokens": 346719626.0, + "step": 13642 + }, + { + "epoch": 1.4982429167581814, + "grad_norm": 2.709036350250244, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.736028790473938, + "num_tokens": 346738491.0, + "step": 13643 + }, + { + "epoch": 1.498352734460795, + "grad_norm": 2.3048627376556396, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7189996838569641, + "num_tokens": 346760991.0, + "step": 13644 + }, + { + "epoch": 1.4984625521634087, + "grad_norm": 2.123922348022461, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7297415733337402, + "num_tokens": 346787977.0, + "step": 13645 + }, + { + "epoch": 1.4985723698660225, + "grad_norm": 2.4858510494232178, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.7378770709037781, + "num_tokens": 346807318.0, + "step": 13646 + }, + { + "epoch": 1.498682187568636, + "grad_norm": 2.199659585952759, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7336738109588623, + "num_tokens": 346832007.0, + "step": 13647 + }, + { + "epoch": 1.4987920052712498, + "grad_norm": 2.254065752029419, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7149146199226379, + "num_tokens": 346859179.0, + "step": 13648 + }, + { + "epoch": 1.4989018229738633, + "grad_norm": 2.7517521381378174, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7189328074455261, + "num_tokens": 346877002.0, + "step": 13649 + }, + { + "epoch": 1.499011640676477, + "grad_norm": 2.5028278827667236, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7248055934906006, + "num_tokens": 346897422.0, + "step": 13650 + }, + { + "epoch": 1.4991214583790908, + "grad_norm": 2.145240306854248, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.724444568157196, + "num_tokens": 346924441.0, + "step": 13651 + }, + { + "epoch": 1.4992312760817044, + "grad_norm": 2.2885618209838867, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7070504426956177, + "num_tokens": 346952152.0, + "step": 13652 + }, + { + "epoch": 1.499341093784318, + "grad_norm": 2.555307626724243, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7197743058204651, + "num_tokens": 346975563.0, + "step": 13653 + }, + { + "epoch": 1.4994509114869317, + "grad_norm": 2.032515048980713, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7122063636779785, + "num_tokens": 347003361.0, + "step": 13654 + }, + { + "epoch": 1.4995607291895454, + "grad_norm": 2.538226366043091, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7293874025344849, + "num_tokens": 347025083.0, + "step": 13655 + }, + { + "epoch": 1.499670546892159, + "grad_norm": 2.3548457622528076, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7277508974075317, + "num_tokens": 347048612.0, + "step": 13656 + }, + { + "epoch": 1.4997803645947727, + "grad_norm": 2.209120750427246, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7182223200798035, + "num_tokens": 347073978.0, + "step": 13657 + }, + { + "epoch": 1.4998901822973862, + "grad_norm": 2.298196792602539, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7187192440032959, + "num_tokens": 347098141.0, + "step": 13658 + }, + { + "epoch": 1.5, + "grad_norm": 2.3188021183013916, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7166779041290283, + "num_tokens": 347123204.0, + "step": 13659 + }, + { + "epoch": 1.5001098177026138, + "grad_norm": 2.402740001678467, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7080893516540527, + "num_tokens": 347146779.0, + "step": 13660 + }, + { + "epoch": 1.5002196354052273, + "grad_norm": 2.521789073944092, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7280921936035156, + "num_tokens": 347168141.0, + "step": 13661 + }, + { + "epoch": 1.5003294531078408, + "grad_norm": 1.781054139137268, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7261117696762085, + "num_tokens": 347204393.0, + "step": 13662 + }, + { + "epoch": 1.5004392708104546, + "grad_norm": 2.256472587585449, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.726161539554596, + "num_tokens": 347228892.0, + "step": 13663 + }, + { + "epoch": 1.5005490885130683, + "grad_norm": 2.095043420791626, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7066773772239685, + "num_tokens": 347256257.0, + "step": 13664 + }, + { + "epoch": 1.500658906215682, + "grad_norm": 2.068676233291626, + "learning_rate": 1e-06, + "loss": 0.9314, + "mean_token_accuracy": 0.7155250906944275, + "num_tokens": 347285497.0, + "step": 13665 + }, + { + "epoch": 1.5007687239182956, + "grad_norm": 2.0123705863952637, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7121542692184448, + "num_tokens": 347317205.0, + "step": 13666 + }, + { + "epoch": 1.5008785416209092, + "grad_norm": 2.258693218231201, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6965348720550537, + "num_tokens": 347344590.0, + "step": 13667 + }, + { + "epoch": 1.500988359323523, + "grad_norm": 1.9757275581359863, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7103672623634338, + "num_tokens": 347377344.0, + "step": 13668 + }, + { + "epoch": 1.5010981770261367, + "grad_norm": 2.50335955619812, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7274460792541504, + "num_tokens": 347397676.0, + "step": 13669 + }, + { + "epoch": 1.5012079947287504, + "grad_norm": 2.2904202938079834, + "learning_rate": 1e-06, + "loss": 0.7894, + "mean_token_accuracy": 0.748738169670105, + "num_tokens": 347422200.0, + "step": 13670 + }, + { + "epoch": 1.501317812431364, + "grad_norm": 2.4059605598449707, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7269350290298462, + "num_tokens": 347444142.0, + "step": 13671 + }, + { + "epoch": 1.5014276301339775, + "grad_norm": 2.2831156253814697, + "learning_rate": 1e-06, + "loss": 0.8743, + "mean_token_accuracy": 0.7280775308609009, + "num_tokens": 347470457.0, + "step": 13672 + }, + { + "epoch": 1.5015374478365913, + "grad_norm": 2.4566686153411865, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7077987194061279, + "num_tokens": 347494147.0, + "step": 13673 + }, + { + "epoch": 1.501647265539205, + "grad_norm": 2.154121160507202, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7207980155944824, + "num_tokens": 347522040.0, + "step": 13674 + }, + { + "epoch": 1.5017570832418186, + "grad_norm": 2.852792978286743, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7351330518722534, + "num_tokens": 347538878.0, + "step": 13675 + }, + { + "epoch": 1.501866900944432, + "grad_norm": 2.202366590499878, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7224404811859131, + "num_tokens": 347564338.0, + "step": 13676 + }, + { + "epoch": 1.5019767186470458, + "grad_norm": 2.448981523513794, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7187291383743286, + "num_tokens": 347585947.0, + "step": 13677 + }, + { + "epoch": 1.5020865363496596, + "grad_norm": 2.668522834777832, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7358928918838501, + "num_tokens": 347604649.0, + "step": 13678 + }, + { + "epoch": 1.5021963540522734, + "grad_norm": 2.4173402786254883, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7194977402687073, + "num_tokens": 347629165.0, + "step": 13679 + }, + { + "epoch": 1.502306171754887, + "grad_norm": 2.1723005771636963, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6929024457931519, + "num_tokens": 347658396.0, + "step": 13680 + }, + { + "epoch": 1.5024159894575004, + "grad_norm": 2.7388007640838623, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7240469455718994, + "num_tokens": 347676297.0, + "step": 13681 + }, + { + "epoch": 1.5025258071601142, + "grad_norm": 2.005070447921753, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7355101108551025, + "num_tokens": 347706736.0, + "step": 13682 + }, + { + "epoch": 1.502635624862728, + "grad_norm": 2.8256025314331055, + "learning_rate": 1e-06, + "loss": 0.7751, + "mean_token_accuracy": 0.7535993456840515, + "num_tokens": 347723326.0, + "step": 13683 + }, + { + "epoch": 1.5027454425653415, + "grad_norm": 2.268198013305664, + "learning_rate": 1e-06, + "loss": 0.881, + "mean_token_accuracy": 0.720775306224823, + "num_tokens": 347747757.0, + "step": 13684 + }, + { + "epoch": 1.5028552602679552, + "grad_norm": 2.6199071407318115, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7131431698799133, + "num_tokens": 347768364.0, + "step": 13685 + }, + { + "epoch": 1.5029650779705688, + "grad_norm": 2.1261658668518066, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7435764074325562, + "num_tokens": 347795732.0, + "step": 13686 + }, + { + "epoch": 1.5030748956731825, + "grad_norm": 2.37577223777771, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7196215391159058, + "num_tokens": 347819509.0, + "step": 13687 + }, + { + "epoch": 1.5031847133757963, + "grad_norm": 2.699099540710449, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.715157687664032, + "num_tokens": 347839400.0, + "step": 13688 + }, + { + "epoch": 1.5032945310784098, + "grad_norm": 2.5319278240203857, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7333739995956421, + "num_tokens": 347861431.0, + "step": 13689 + }, + { + "epoch": 1.5034043487810234, + "grad_norm": 2.4612252712249756, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7312926054000854, + "num_tokens": 347882512.0, + "step": 13690 + }, + { + "epoch": 1.5035141664836371, + "grad_norm": 2.2803914546966553, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.728398859500885, + "num_tokens": 347906198.0, + "step": 13691 + }, + { + "epoch": 1.5036239841862509, + "grad_norm": 2.003135919570923, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7206923961639404, + "num_tokens": 347937085.0, + "step": 13692 + }, + { + "epoch": 1.5037338018888646, + "grad_norm": 2.050110340118408, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7091429233551025, + "num_tokens": 347968078.0, + "step": 13693 + }, + { + "epoch": 1.5038436195914782, + "grad_norm": 2.305220603942871, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7170277237892151, + "num_tokens": 347992230.0, + "step": 13694 + }, + { + "epoch": 1.5039534372940917, + "grad_norm": 2.311505079269409, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.7306643724441528, + "num_tokens": 348016648.0, + "step": 13695 + }, + { + "epoch": 1.5040632549967055, + "grad_norm": 2.0905702114105225, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7101029753684998, + "num_tokens": 348045453.0, + "step": 13696 + }, + { + "epoch": 1.5041730726993192, + "grad_norm": 2.26985764503479, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7176904678344727, + "num_tokens": 348072277.0, + "step": 13697 + }, + { + "epoch": 1.5042828904019327, + "grad_norm": 2.0832066535949707, + "learning_rate": 1e-06, + "loss": 0.9336, + "mean_token_accuracy": 0.7084023952484131, + "num_tokens": 348102364.0, + "step": 13698 + }, + { + "epoch": 1.5043927081045465, + "grad_norm": 2.18078875541687, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7297263145446777, + "num_tokens": 348129145.0, + "step": 13699 + }, + { + "epoch": 1.50450252580716, + "grad_norm": 2.1882550716400146, + "learning_rate": 1e-06, + "loss": 0.9638, + "mean_token_accuracy": 0.7118340730667114, + "num_tokens": 348155758.0, + "step": 13700 + }, + { + "epoch": 1.5046123435097738, + "grad_norm": 1.9644867181777954, + "learning_rate": 1e-06, + "loss": 0.9716, + "mean_token_accuracy": 0.7038326263427734, + "num_tokens": 348187726.0, + "step": 13701 + }, + { + "epoch": 1.5047221612123876, + "grad_norm": 2.372544527053833, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7208329439163208, + "num_tokens": 348212166.0, + "step": 13702 + }, + { + "epoch": 1.504831978915001, + "grad_norm": 2.094308853149414, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7287659049034119, + "num_tokens": 348239799.0, + "step": 13703 + }, + { + "epoch": 1.5049417966176146, + "grad_norm": 2.464963674545288, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7456585168838501, + "num_tokens": 348260405.0, + "step": 13704 + }, + { + "epoch": 1.5050516143202284, + "grad_norm": 2.583895206451416, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7293268442153931, + "num_tokens": 348279921.0, + "step": 13705 + }, + { + "epoch": 1.5051614320228421, + "grad_norm": 2.363238573074341, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7393258213996887, + "num_tokens": 348301906.0, + "step": 13706 + }, + { + "epoch": 1.505271249725456, + "grad_norm": 2.477538824081421, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7225795984268188, + "num_tokens": 348324269.0, + "step": 13707 + }, + { + "epoch": 1.5053810674280694, + "grad_norm": 2.420381546020508, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.7458916902542114, + "num_tokens": 348345866.0, + "step": 13708 + }, + { + "epoch": 1.505490885130683, + "grad_norm": 2.442894458770752, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7205402851104736, + "num_tokens": 348368732.0, + "step": 13709 + }, + { + "epoch": 1.5056007028332967, + "grad_norm": 2.6162514686584473, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7215847969055176, + "num_tokens": 348388876.0, + "step": 13710 + }, + { + "epoch": 1.5057105205359105, + "grad_norm": 2.44439959526062, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7337668538093567, + "num_tokens": 348409506.0, + "step": 13711 + }, + { + "epoch": 1.505820338238524, + "grad_norm": 2.094261646270752, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7334868907928467, + "num_tokens": 348436297.0, + "step": 13712 + }, + { + "epoch": 1.5059301559411375, + "grad_norm": 2.3651278018951416, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.731033980846405, + "num_tokens": 348460259.0, + "step": 13713 + }, + { + "epoch": 1.5060399736437513, + "grad_norm": 2.2258946895599365, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7074840068817139, + "num_tokens": 348486656.0, + "step": 13714 + }, + { + "epoch": 1.506149791346365, + "grad_norm": 2.1369354724884033, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7283012866973877, + "num_tokens": 348512584.0, + "step": 13715 + }, + { + "epoch": 1.5062596090489788, + "grad_norm": 2.436739206314087, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7358353734016418, + "num_tokens": 348535229.0, + "step": 13716 + }, + { + "epoch": 1.5063694267515924, + "grad_norm": 2.567336082458496, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7186802625656128, + "num_tokens": 348556651.0, + "step": 13717 + }, + { + "epoch": 1.506479244454206, + "grad_norm": 1.9906399250030518, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.737359881401062, + "num_tokens": 348585564.0, + "step": 13718 + }, + { + "epoch": 1.5065890621568196, + "grad_norm": 2.1334598064422607, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7137727737426758, + "num_tokens": 348611766.0, + "step": 13719 + }, + { + "epoch": 1.5066988798594334, + "grad_norm": 2.243943691253662, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7149733304977417, + "num_tokens": 348636911.0, + "step": 13720 + }, + { + "epoch": 1.5068086975620472, + "grad_norm": 2.23750638961792, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.719131350517273, + "num_tokens": 348662413.0, + "step": 13721 + }, + { + "epoch": 1.5069185152646607, + "grad_norm": 1.9509615898132324, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7079219222068787, + "num_tokens": 348694381.0, + "step": 13722 + }, + { + "epoch": 1.5070283329672742, + "grad_norm": 2.8137078285217285, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7226089835166931, + "num_tokens": 348712920.0, + "step": 13723 + }, + { + "epoch": 1.507138150669888, + "grad_norm": 2.2658095359802246, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7142705917358398, + "num_tokens": 348739282.0, + "step": 13724 + }, + { + "epoch": 1.5072479683725017, + "grad_norm": 2.100200653076172, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7143905162811279, + "num_tokens": 348766638.0, + "step": 13725 + }, + { + "epoch": 1.5073577860751153, + "grad_norm": 2.2104690074920654, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7252532243728638, + "num_tokens": 348794123.0, + "step": 13726 + }, + { + "epoch": 1.5074676037777288, + "grad_norm": 2.2874860763549805, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7130473852157593, + "num_tokens": 348820832.0, + "step": 13727 + }, + { + "epoch": 1.5075774214803426, + "grad_norm": 2.08677339553833, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7150406837463379, + "num_tokens": 348849883.0, + "step": 13728 + }, + { + "epoch": 1.5076872391829563, + "grad_norm": 2.1016037464141846, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7116009593009949, + "num_tokens": 348877236.0, + "step": 13729 + }, + { + "epoch": 1.50779705688557, + "grad_norm": 2.1899008750915527, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.718692421913147, + "num_tokens": 348904858.0, + "step": 13730 + }, + { + "epoch": 1.5079068745881836, + "grad_norm": 2.036839485168457, + "learning_rate": 1e-06, + "loss": 0.8104, + "mean_token_accuracy": 0.7449077367782593, + "num_tokens": 348935697.0, + "step": 13731 + }, + { + "epoch": 1.5080166922907972, + "grad_norm": 2.0545945167541504, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7133249640464783, + "num_tokens": 348965448.0, + "step": 13732 + }, + { + "epoch": 1.508126509993411, + "grad_norm": 2.576799154281616, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7295122146606445, + "num_tokens": 348986140.0, + "step": 13733 + }, + { + "epoch": 1.5082363276960247, + "grad_norm": 2.11173152923584, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7094185948371887, + "num_tokens": 349013792.0, + "step": 13734 + }, + { + "epoch": 1.5083461453986382, + "grad_norm": 2.255448818206787, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7160894274711609, + "num_tokens": 349037999.0, + "step": 13735 + }, + { + "epoch": 1.508455963101252, + "grad_norm": 2.265191078186035, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7108403444290161, + "num_tokens": 349065613.0, + "step": 13736 + }, + { + "epoch": 1.5085657808038655, + "grad_norm": 2.297189712524414, + "learning_rate": 1e-06, + "loss": 1.0048, + "mean_token_accuracy": 0.6969115138053894, + "num_tokens": 349092833.0, + "step": 13737 + }, + { + "epoch": 1.5086755985064793, + "grad_norm": 2.225858449935913, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7265251874923706, + "num_tokens": 349117491.0, + "step": 13738 + }, + { + "epoch": 1.508785416209093, + "grad_norm": 2.132476568222046, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7173249125480652, + "num_tokens": 349147411.0, + "step": 13739 + }, + { + "epoch": 1.5088952339117065, + "grad_norm": 1.874207854270935, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.6964077949523926, + "num_tokens": 349185393.0, + "step": 13740 + }, + { + "epoch": 1.50900505161432, + "grad_norm": 2.48399019241333, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7282947301864624, + "num_tokens": 349207626.0, + "step": 13741 + }, + { + "epoch": 1.5091148693169338, + "grad_norm": 2.3465464115142822, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7093816995620728, + "num_tokens": 349232242.0, + "step": 13742 + }, + { + "epoch": 1.5092246870195476, + "grad_norm": 2.1188392639160156, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7017804384231567, + "num_tokens": 349261764.0, + "step": 13743 + }, + { + "epoch": 1.5093345047221614, + "grad_norm": 2.3316123485565186, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.6968648433685303, + "num_tokens": 349286760.0, + "step": 13744 + }, + { + "epoch": 1.5094443224247749, + "grad_norm": 2.263028144836426, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7166613340377808, + "num_tokens": 349312566.0, + "step": 13745 + }, + { + "epoch": 1.5095541401273884, + "grad_norm": 2.239335536956787, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7224951982498169, + "num_tokens": 349337758.0, + "step": 13746 + }, + { + "epoch": 1.5096639578300022, + "grad_norm": 2.104830265045166, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7294588685035706, + "num_tokens": 349364305.0, + "step": 13747 + }, + { + "epoch": 1.509773775532616, + "grad_norm": 2.111816644668579, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7204003930091858, + "num_tokens": 349394211.0, + "step": 13748 + }, + { + "epoch": 1.5098835932352295, + "grad_norm": 2.209554672241211, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7197002172470093, + "num_tokens": 349420227.0, + "step": 13749 + }, + { + "epoch": 1.5099934109378432, + "grad_norm": 2.2884573936462402, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7361196279525757, + "num_tokens": 349444328.0, + "step": 13750 + }, + { + "epoch": 1.5101032286404568, + "grad_norm": 2.2087037563323975, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7178956270217896, + "num_tokens": 349472127.0, + "step": 13751 + }, + { + "epoch": 1.5102130463430705, + "grad_norm": 2.0360453128814697, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7233047485351562, + "num_tokens": 349503457.0, + "step": 13752 + }, + { + "epoch": 1.5103228640456843, + "grad_norm": 2.304091691970825, + "learning_rate": 1e-06, + "loss": 0.8071, + "mean_token_accuracy": 0.7450186014175415, + "num_tokens": 349526786.0, + "step": 13753 + }, + { + "epoch": 1.5104326817482978, + "grad_norm": 2.0600602626800537, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7041101455688477, + "num_tokens": 349557458.0, + "step": 13754 + }, + { + "epoch": 1.5105424994509113, + "grad_norm": 2.129149913787842, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7213389873504639, + "num_tokens": 349584873.0, + "step": 13755 + }, + { + "epoch": 1.510652317153525, + "grad_norm": 2.5192923545837402, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7367408275604248, + "num_tokens": 349606269.0, + "step": 13756 + }, + { + "epoch": 1.5107621348561389, + "grad_norm": 2.253741979598999, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7179422378540039, + "num_tokens": 349629678.0, + "step": 13757 + }, + { + "epoch": 1.5108719525587526, + "grad_norm": 2.0689191818237305, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7017841935157776, + "num_tokens": 349657286.0, + "step": 13758 + }, + { + "epoch": 1.5109817702613662, + "grad_norm": 2.578314781188965, + "learning_rate": 1e-06, + "loss": 0.7403, + "mean_token_accuracy": 0.7599296569824219, + "num_tokens": 349677391.0, + "step": 13759 + }, + { + "epoch": 1.5110915879639797, + "grad_norm": 2.225538492202759, + "learning_rate": 1e-06, + "loss": 1.0667, + "mean_token_accuracy": 0.6784566044807434, + "num_tokens": 349704986.0, + "step": 13760 + }, + { + "epoch": 1.5112014056665934, + "grad_norm": 2.3598947525024414, + "learning_rate": 1e-06, + "loss": 0.7905, + "mean_token_accuracy": 0.7408562302589417, + "num_tokens": 349727469.0, + "step": 13761 + }, + { + "epoch": 1.5113112233692072, + "grad_norm": 2.434356212615967, + "learning_rate": 1e-06, + "loss": 0.8373, + "mean_token_accuracy": 0.7345898151397705, + "num_tokens": 349750951.0, + "step": 13762 + }, + { + "epoch": 1.5114210410718207, + "grad_norm": 2.1435706615448, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6906329393386841, + "num_tokens": 349780005.0, + "step": 13763 + }, + { + "epoch": 1.5115308587744345, + "grad_norm": 2.4406938552856445, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7210391759872437, + "num_tokens": 349805897.0, + "step": 13764 + }, + { + "epoch": 1.511640676477048, + "grad_norm": 2.173335075378418, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.722099781036377, + "num_tokens": 349833133.0, + "step": 13765 + }, + { + "epoch": 1.5117504941796618, + "grad_norm": 2.1758196353912354, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7379917502403259, + "num_tokens": 349858052.0, + "step": 13766 + }, + { + "epoch": 1.5118603118822755, + "grad_norm": 2.1561317443847656, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.71805340051651, + "num_tokens": 349885196.0, + "step": 13767 + }, + { + "epoch": 1.511970129584889, + "grad_norm": 1.9877245426177979, + "learning_rate": 1e-06, + "loss": 0.8664, + "mean_token_accuracy": 0.7253193259239197, + "num_tokens": 349915433.0, + "step": 13768 + }, + { + "epoch": 1.5120799472875026, + "grad_norm": 2.6316468715667725, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7108683586120605, + "num_tokens": 349935926.0, + "step": 13769 + }, + { + "epoch": 1.5121897649901164, + "grad_norm": 2.3968677520751953, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7166876196861267, + "num_tokens": 349959040.0, + "step": 13770 + }, + { + "epoch": 1.5122995826927301, + "grad_norm": 1.962367296218872, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7163130044937134, + "num_tokens": 349992241.0, + "step": 13771 + }, + { + "epoch": 1.5124094003953439, + "grad_norm": 2.3947577476501465, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7201941013336182, + "num_tokens": 350014515.0, + "step": 13772 + }, + { + "epoch": 1.5125192180979574, + "grad_norm": 2.102790117263794, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7038478255271912, + "num_tokens": 350042266.0, + "step": 13773 + }, + { + "epoch": 1.512629035800571, + "grad_norm": 2.1726584434509277, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.712699294090271, + "num_tokens": 350070835.0, + "step": 13774 + }, + { + "epoch": 1.5127388535031847, + "grad_norm": 2.573544979095459, + "learning_rate": 1e-06, + "loss": 0.8424, + "mean_token_accuracy": 0.7304229736328125, + "num_tokens": 350090752.0, + "step": 13775 + }, + { + "epoch": 1.5128486712057985, + "grad_norm": 1.9132031202316284, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7196296453475952, + "num_tokens": 350122114.0, + "step": 13776 + }, + { + "epoch": 1.512958488908412, + "grad_norm": 2.121920347213745, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7229773998260498, + "num_tokens": 350149191.0, + "step": 13777 + }, + { + "epoch": 1.5130683066110255, + "grad_norm": 2.4912219047546387, + "learning_rate": 1e-06, + "loss": 0.9769, + "mean_token_accuracy": 0.7134518623352051, + "num_tokens": 350171817.0, + "step": 13778 + }, + { + "epoch": 1.5131781243136393, + "grad_norm": 2.495962619781494, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7107638120651245, + "num_tokens": 350194462.0, + "step": 13779 + }, + { + "epoch": 1.513287942016253, + "grad_norm": 2.44775128364563, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7334752082824707, + "num_tokens": 350217003.0, + "step": 13780 + }, + { + "epoch": 1.5133977597188668, + "grad_norm": 2.3703534603118896, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7418004274368286, + "num_tokens": 350240669.0, + "step": 13781 + }, + { + "epoch": 1.5135075774214803, + "grad_norm": 2.344945192337036, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7237305641174316, + "num_tokens": 350264489.0, + "step": 13782 + }, + { + "epoch": 1.5136173951240939, + "grad_norm": 2.0786209106445312, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7312084436416626, + "num_tokens": 350293746.0, + "step": 13783 + }, + { + "epoch": 1.5137272128267076, + "grad_norm": 2.5536296367645264, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7293439507484436, + "num_tokens": 350314351.0, + "step": 13784 + }, + { + "epoch": 1.5138370305293214, + "grad_norm": 1.9860658645629883, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.6942564249038696, + "num_tokens": 350345244.0, + "step": 13785 + }, + { + "epoch": 1.5139468482319351, + "grad_norm": 2.318516969680786, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7404270768165588, + "num_tokens": 350370381.0, + "step": 13786 + }, + { + "epoch": 1.5140566659345487, + "grad_norm": 2.4818570613861084, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7302014231681824, + "num_tokens": 350391255.0, + "step": 13787 + }, + { + "epoch": 1.5141664836371622, + "grad_norm": 2.3829705715179443, + "learning_rate": 1e-06, + "loss": 0.9672, + "mean_token_accuracy": 0.7058302164077759, + "num_tokens": 350415673.0, + "step": 13788 + }, + { + "epoch": 1.514276301339776, + "grad_norm": 2.2288296222686768, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7235172986984253, + "num_tokens": 350442261.0, + "step": 13789 + }, + { + "epoch": 1.5143861190423897, + "grad_norm": 1.9785940647125244, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7284888029098511, + "num_tokens": 350472071.0, + "step": 13790 + }, + { + "epoch": 1.5144959367450033, + "grad_norm": 2.296855926513672, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7026112079620361, + "num_tokens": 350496839.0, + "step": 13791 + }, + { + "epoch": 1.5146057544476168, + "grad_norm": 2.1022822856903076, + "learning_rate": 1e-06, + "loss": 1.0017, + "mean_token_accuracy": 0.7026129961013794, + "num_tokens": 350524687.0, + "step": 13792 + }, + { + "epoch": 1.5147155721502306, + "grad_norm": 2.2952358722686768, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7114582657814026, + "num_tokens": 350550230.0, + "step": 13793 + }, + { + "epoch": 1.5148253898528443, + "grad_norm": 1.9873850345611572, + "learning_rate": 1e-06, + "loss": 0.9896, + "mean_token_accuracy": 0.7059375047683716, + "num_tokens": 350584351.0, + "step": 13794 + }, + { + "epoch": 1.514935207555458, + "grad_norm": 2.3148717880249023, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7282370328903198, + "num_tokens": 350607723.0, + "step": 13795 + }, + { + "epoch": 1.5150450252580716, + "grad_norm": 2.361147165298462, + "learning_rate": 1e-06, + "loss": 0.8009, + "mean_token_accuracy": 0.74242103099823, + "num_tokens": 350631512.0, + "step": 13796 + }, + { + "epoch": 1.5151548429606851, + "grad_norm": 2.2336673736572266, + "learning_rate": 1e-06, + "loss": 0.9805, + "mean_token_accuracy": 0.7031494379043579, + "num_tokens": 350661094.0, + "step": 13797 + }, + { + "epoch": 1.515264660663299, + "grad_norm": 2.195587158203125, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7199246883392334, + "num_tokens": 350685549.0, + "step": 13798 + }, + { + "epoch": 1.5153744783659127, + "grad_norm": 2.575282573699951, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7384419441223145, + "num_tokens": 350704789.0, + "step": 13799 + }, + { + "epoch": 1.5154842960685262, + "grad_norm": 2.2086410522460938, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7237343788146973, + "num_tokens": 350729851.0, + "step": 13800 + }, + { + "epoch": 1.51559411377114, + "grad_norm": 2.248128890991211, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7182064056396484, + "num_tokens": 350754575.0, + "step": 13801 + }, + { + "epoch": 1.5157039314737535, + "grad_norm": 2.249660015106201, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7132366895675659, + "num_tokens": 350781710.0, + "step": 13802 + }, + { + "epoch": 1.5158137491763672, + "grad_norm": 2.133484125137329, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7245326638221741, + "num_tokens": 350809046.0, + "step": 13803 + }, + { + "epoch": 1.515923566878981, + "grad_norm": 2.0928847789764404, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7092456817626953, + "num_tokens": 350837531.0, + "step": 13804 + }, + { + "epoch": 1.5160333845815945, + "grad_norm": 2.0704197883605957, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.71098792552948, + "num_tokens": 350866886.0, + "step": 13805 + }, + { + "epoch": 1.516143202284208, + "grad_norm": 2.568444013595581, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7311052680015564, + "num_tokens": 350887468.0, + "step": 13806 + }, + { + "epoch": 1.5162530199868218, + "grad_norm": 2.250732660293579, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.7166801691055298, + "num_tokens": 350913730.0, + "step": 13807 + }, + { + "epoch": 1.5163628376894356, + "grad_norm": 2.3301026821136475, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7235677242279053, + "num_tokens": 350936675.0, + "step": 13808 + }, + { + "epoch": 1.5164726553920493, + "grad_norm": 2.3742387294769287, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7192943096160889, + "num_tokens": 350960668.0, + "step": 13809 + }, + { + "epoch": 1.5165824730946629, + "grad_norm": 2.9772069454193115, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7419085502624512, + "num_tokens": 350977910.0, + "step": 13810 + }, + { + "epoch": 1.5166922907972764, + "grad_norm": 2.344567060470581, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7209365963935852, + "num_tokens": 351001282.0, + "step": 13811 + }, + { + "epoch": 1.5168021084998902, + "grad_norm": 2.260211706161499, + "learning_rate": 1e-06, + "loss": 0.8241, + "mean_token_accuracy": 0.7345975041389465, + "num_tokens": 351025203.0, + "step": 13812 + }, + { + "epoch": 1.516911926202504, + "grad_norm": 2.132270574569702, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.7090087532997131, + "num_tokens": 351053106.0, + "step": 13813 + }, + { + "epoch": 1.5170217439051175, + "grad_norm": 2.1607155799865723, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7323665618896484, + "num_tokens": 351080426.0, + "step": 13814 + }, + { + "epoch": 1.5171315616077312, + "grad_norm": 2.2490344047546387, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7080155611038208, + "num_tokens": 351105605.0, + "step": 13815 + }, + { + "epoch": 1.5172413793103448, + "grad_norm": 2.2490224838256836, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7266649603843689, + "num_tokens": 351130593.0, + "step": 13816 + }, + { + "epoch": 1.5173511970129585, + "grad_norm": 2.113635301589966, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7147111296653748, + "num_tokens": 351162157.0, + "step": 13817 + }, + { + "epoch": 1.5174610147155723, + "grad_norm": 2.244135856628418, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7130041122436523, + "num_tokens": 351189146.0, + "step": 13818 + }, + { + "epoch": 1.5175708324181858, + "grad_norm": 2.2942330837249756, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7211217880249023, + "num_tokens": 351213351.0, + "step": 13819 + }, + { + "epoch": 1.5176806501207993, + "grad_norm": 2.2844345569610596, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7238823771476746, + "num_tokens": 351236642.0, + "step": 13820 + }, + { + "epoch": 1.517790467823413, + "grad_norm": 2.1934995651245117, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7358294725418091, + "num_tokens": 351262706.0, + "step": 13821 + }, + { + "epoch": 1.5179002855260268, + "grad_norm": 2.162240743637085, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.720577597618103, + "num_tokens": 351289605.0, + "step": 13822 + }, + { + "epoch": 1.5180101032286406, + "grad_norm": 2.1265790462493896, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.722590446472168, + "num_tokens": 351318608.0, + "step": 13823 + }, + { + "epoch": 1.5181199209312541, + "grad_norm": 2.7657885551452637, + "learning_rate": 1e-06, + "loss": 0.795, + "mean_token_accuracy": 0.7471334934234619, + "num_tokens": 351336913.0, + "step": 13824 + }, + { + "epoch": 1.5182297386338677, + "grad_norm": 2.70582914352417, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7165681719779968, + "num_tokens": 351356376.0, + "step": 13825 + }, + { + "epoch": 1.5183395563364814, + "grad_norm": 2.2472007274627686, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7219053506851196, + "num_tokens": 351382934.0, + "step": 13826 + }, + { + "epoch": 1.5184493740390952, + "grad_norm": 2.3925607204437256, + "learning_rate": 1e-06, + "loss": 0.8007, + "mean_token_accuracy": 0.7480854392051697, + "num_tokens": 351405249.0, + "step": 13827 + }, + { + "epoch": 1.5185591917417087, + "grad_norm": 2.149362802505493, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7226181030273438, + "num_tokens": 351431857.0, + "step": 13828 + }, + { + "epoch": 1.5186690094443223, + "grad_norm": 2.4805784225463867, + "learning_rate": 1e-06, + "loss": 0.9734, + "mean_token_accuracy": 0.7020484209060669, + "num_tokens": 351454004.0, + "step": 13829 + }, + { + "epoch": 1.518778827146936, + "grad_norm": 2.524407386779785, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.7476229667663574, + "num_tokens": 351474149.0, + "step": 13830 + }, + { + "epoch": 1.5188886448495498, + "grad_norm": 2.3386778831481934, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7397130727767944, + "num_tokens": 351496347.0, + "step": 13831 + }, + { + "epoch": 1.5189984625521635, + "grad_norm": 2.172079563140869, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7409125566482544, + "num_tokens": 351522203.0, + "step": 13832 + }, + { + "epoch": 1.519108280254777, + "grad_norm": 2.6472110748291016, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7216687202453613, + "num_tokens": 351543544.0, + "step": 13833 + }, + { + "epoch": 1.5192180979573906, + "grad_norm": 2.3759701251983643, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7120765447616577, + "num_tokens": 351568776.0, + "step": 13834 + }, + { + "epoch": 1.5193279156600044, + "grad_norm": 2.3361856937408447, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.714680552482605, + "num_tokens": 351592400.0, + "step": 13835 + }, + { + "epoch": 1.5194377333626181, + "grad_norm": 2.2458133697509766, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7327243089675903, + "num_tokens": 351615363.0, + "step": 13836 + }, + { + "epoch": 1.5195475510652319, + "grad_norm": 2.3557474613189697, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7210321426391602, + "num_tokens": 351638520.0, + "step": 13837 + }, + { + "epoch": 1.5196573687678454, + "grad_norm": 2.3869848251342773, + "learning_rate": 1e-06, + "loss": 0.8036, + "mean_token_accuracy": 0.7406039237976074, + "num_tokens": 351659521.0, + "step": 13838 + }, + { + "epoch": 1.519767186470459, + "grad_norm": 2.2936618328094482, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7155848145484924, + "num_tokens": 351685865.0, + "step": 13839 + }, + { + "epoch": 1.5198770041730727, + "grad_norm": 2.31892466545105, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7172104716300964, + "num_tokens": 351708522.0, + "step": 13840 + }, + { + "epoch": 1.5199868218756865, + "grad_norm": 2.0273475646972656, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7224951982498169, + "num_tokens": 351738283.0, + "step": 13841 + }, + { + "epoch": 1.5200966395783, + "grad_norm": 2.1579322814941406, + "learning_rate": 1e-06, + "loss": 1.0053, + "mean_token_accuracy": 0.6916286945343018, + "num_tokens": 351765837.0, + "step": 13842 + }, + { + "epoch": 1.5202064572809135, + "grad_norm": 2.2305033206939697, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7001246213912964, + "num_tokens": 351791984.0, + "step": 13843 + }, + { + "epoch": 1.5203162749835273, + "grad_norm": 2.1556482315063477, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7111831903457642, + "num_tokens": 351820030.0, + "step": 13844 + }, + { + "epoch": 1.520426092686141, + "grad_norm": 2.351616859436035, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7441572546958923, + "num_tokens": 351842194.0, + "step": 13845 + }, + { + "epoch": 1.5205359103887548, + "grad_norm": 2.351649045944214, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.70795738697052, + "num_tokens": 351868442.0, + "step": 13846 + }, + { + "epoch": 1.5206457280913683, + "grad_norm": 2.2042505741119385, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7158013582229614, + "num_tokens": 351894324.0, + "step": 13847 + }, + { + "epoch": 1.5207555457939819, + "grad_norm": 2.0561363697052, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7154548168182373, + "num_tokens": 351923943.0, + "step": 13848 + }, + { + "epoch": 1.5208653634965956, + "grad_norm": 2.315462827682495, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7005429267883301, + "num_tokens": 351950249.0, + "step": 13849 + }, + { + "epoch": 1.5209751811992094, + "grad_norm": 2.3754117488861084, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.726294755935669, + "num_tokens": 351972804.0, + "step": 13850 + }, + { + "epoch": 1.5210849989018231, + "grad_norm": 2.0071845054626465, + "learning_rate": 1e-06, + "loss": 1.0086, + "mean_token_accuracy": 0.6949266791343689, + "num_tokens": 352005858.0, + "step": 13851 + }, + { + "epoch": 1.5211948166044367, + "grad_norm": 2.4158477783203125, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7188035249710083, + "num_tokens": 352029991.0, + "step": 13852 + }, + { + "epoch": 1.5213046343070502, + "grad_norm": 2.4127931594848633, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.7284390926361084, + "num_tokens": 352053230.0, + "step": 13853 + }, + { + "epoch": 1.521414452009664, + "grad_norm": 2.47942852973938, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.693149209022522, + "num_tokens": 352076059.0, + "step": 13854 + }, + { + "epoch": 1.5215242697122777, + "grad_norm": 2.0609800815582275, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7238534688949585, + "num_tokens": 352106622.0, + "step": 13855 + }, + { + "epoch": 1.5216340874148913, + "grad_norm": 2.7486226558685303, + "learning_rate": 1e-06, + "loss": 0.8658, + "mean_token_accuracy": 0.7238918542861938, + "num_tokens": 352124801.0, + "step": 13856 + }, + { + "epoch": 1.5217439051175048, + "grad_norm": 2.6208176612854004, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7325961589813232, + "num_tokens": 352144641.0, + "step": 13857 + }, + { + "epoch": 1.5218537228201185, + "grad_norm": 2.398346424102783, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.721735954284668, + "num_tokens": 352167640.0, + "step": 13858 + }, + { + "epoch": 1.5219635405227323, + "grad_norm": 2.245194911956787, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7154631018638611, + "num_tokens": 352193152.0, + "step": 13859 + }, + { + "epoch": 1.522073358225346, + "grad_norm": 2.1894619464874268, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7061225175857544, + "num_tokens": 352219598.0, + "step": 13860 + }, + { + "epoch": 1.5221831759279596, + "grad_norm": 2.319835901260376, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7052322626113892, + "num_tokens": 352244567.0, + "step": 13861 + }, + { + "epoch": 1.5222929936305731, + "grad_norm": 2.3157670497894287, + "learning_rate": 1e-06, + "loss": 0.8049, + "mean_token_accuracy": 0.748133659362793, + "num_tokens": 352265869.0, + "step": 13862 + }, + { + "epoch": 1.522402811333187, + "grad_norm": 2.455130100250244, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.711840033531189, + "num_tokens": 352287474.0, + "step": 13863 + }, + { + "epoch": 1.5225126290358006, + "grad_norm": 1.9784289598464966, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7123465538024902, + "num_tokens": 352318320.0, + "step": 13864 + }, + { + "epoch": 1.5226224467384142, + "grad_norm": 1.989259958267212, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7012173533439636, + "num_tokens": 352348820.0, + "step": 13865 + }, + { + "epoch": 1.522732264441028, + "grad_norm": 2.2801549434661865, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7102575302124023, + "num_tokens": 352373148.0, + "step": 13866 + }, + { + "epoch": 1.5228420821436415, + "grad_norm": 2.1404383182525635, + "learning_rate": 1e-06, + "loss": 1.0109, + "mean_token_accuracy": 0.6984252333641052, + "num_tokens": 352400819.0, + "step": 13867 + }, + { + "epoch": 1.5229518998462552, + "grad_norm": 2.146252155303955, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7113056182861328, + "num_tokens": 352427036.0, + "step": 13868 + }, + { + "epoch": 1.523061717548869, + "grad_norm": 2.319823980331421, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7370383739471436, + "num_tokens": 352449242.0, + "step": 13869 + }, + { + "epoch": 1.5231715352514825, + "grad_norm": 2.5725913047790527, + "learning_rate": 1e-06, + "loss": 0.8354, + "mean_token_accuracy": 0.7354718446731567, + "num_tokens": 352470348.0, + "step": 13870 + }, + { + "epoch": 1.523281352954096, + "grad_norm": 2.5852057933807373, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7300541996955872, + "num_tokens": 352491519.0, + "step": 13871 + }, + { + "epoch": 1.5233911706567098, + "grad_norm": 2.3949859142303467, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7303097248077393, + "num_tokens": 352512801.0, + "step": 13872 + }, + { + "epoch": 1.5235009883593236, + "grad_norm": 2.2752578258514404, + "learning_rate": 1e-06, + "loss": 0.9684, + "mean_token_accuracy": 0.7046327590942383, + "num_tokens": 352539703.0, + "step": 13873 + }, + { + "epoch": 1.5236108060619373, + "grad_norm": 2.35878849029541, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7355959415435791, + "num_tokens": 352561892.0, + "step": 13874 + }, + { + "epoch": 1.5237206237645509, + "grad_norm": 2.2999446392059326, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7105515003204346, + "num_tokens": 352587612.0, + "step": 13875 + }, + { + "epoch": 1.5238304414671644, + "grad_norm": 2.301492691040039, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7020893096923828, + "num_tokens": 352617467.0, + "step": 13876 + }, + { + "epoch": 1.5239402591697782, + "grad_norm": 2.614975690841675, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.724850058555603, + "num_tokens": 352637138.0, + "step": 13877 + }, + { + "epoch": 1.524050076872392, + "grad_norm": 2.273186445236206, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7102817296981812, + "num_tokens": 352661930.0, + "step": 13878 + }, + { + "epoch": 1.5241598945750054, + "grad_norm": 2.080758810043335, + "learning_rate": 1e-06, + "loss": 0.9637, + "mean_token_accuracy": 0.6975984573364258, + "num_tokens": 352692781.0, + "step": 13879 + }, + { + "epoch": 1.5242697122776192, + "grad_norm": 2.2983787059783936, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7022237181663513, + "num_tokens": 352715691.0, + "step": 13880 + }, + { + "epoch": 1.5243795299802327, + "grad_norm": 2.3031561374664307, + "learning_rate": 1e-06, + "loss": 0.7333, + "mean_token_accuracy": 0.758393406867981, + "num_tokens": 352737077.0, + "step": 13881 + }, + { + "epoch": 1.5244893476828465, + "grad_norm": 2.3363423347473145, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.715813398361206, + "num_tokens": 352760791.0, + "step": 13882 + }, + { + "epoch": 1.5245991653854603, + "grad_norm": 2.421271562576294, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.7041729688644409, + "num_tokens": 352782881.0, + "step": 13883 + }, + { + "epoch": 1.5247089830880738, + "grad_norm": 2.4663286209106445, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.7244244813919067, + "num_tokens": 352804280.0, + "step": 13884 + }, + { + "epoch": 1.5248188007906873, + "grad_norm": 2.297654628753662, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7277535796165466, + "num_tokens": 352828933.0, + "step": 13885 + }, + { + "epoch": 1.524928618493301, + "grad_norm": 2.426774740219116, + "learning_rate": 1e-06, + "loss": 0.7808, + "mean_token_accuracy": 0.7490942478179932, + "num_tokens": 352849877.0, + "step": 13886 + }, + { + "epoch": 1.5250384361959148, + "grad_norm": 2.422220230102539, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7224311828613281, + "num_tokens": 352870910.0, + "step": 13887 + }, + { + "epoch": 1.5251482538985286, + "grad_norm": 2.300276517868042, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7130646109580994, + "num_tokens": 352894855.0, + "step": 13888 + }, + { + "epoch": 1.5252580716011421, + "grad_norm": 2.2554843425750732, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7227861285209656, + "num_tokens": 352919767.0, + "step": 13889 + }, + { + "epoch": 1.5253678893037557, + "grad_norm": 2.2337167263031006, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7151666879653931, + "num_tokens": 352946129.0, + "step": 13890 + }, + { + "epoch": 1.5254777070063694, + "grad_norm": 2.2813658714294434, + "learning_rate": 1e-06, + "loss": 0.9797, + "mean_token_accuracy": 0.7005521059036255, + "num_tokens": 352970798.0, + "step": 13891 + }, + { + "epoch": 1.5255875247089832, + "grad_norm": 2.4356563091278076, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7135671377182007, + "num_tokens": 352993618.0, + "step": 13892 + }, + { + "epoch": 1.5256973424115967, + "grad_norm": 2.398926019668579, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.715085506439209, + "num_tokens": 353017361.0, + "step": 13893 + }, + { + "epoch": 1.5258071601142102, + "grad_norm": 2.246375560760498, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.6996834874153137, + "num_tokens": 353042990.0, + "step": 13894 + }, + { + "epoch": 1.525916977816824, + "grad_norm": 2.141552686691284, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7115453481674194, + "num_tokens": 353070273.0, + "step": 13895 + }, + { + "epoch": 1.5260267955194378, + "grad_norm": 2.0777618885040283, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.707009494304657, + "num_tokens": 353098657.0, + "step": 13896 + }, + { + "epoch": 1.5261366132220515, + "grad_norm": 2.4477145671844482, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7309699058532715, + "num_tokens": 353120004.0, + "step": 13897 + }, + { + "epoch": 1.526246430924665, + "grad_norm": 2.5654475688934326, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7226985692977905, + "num_tokens": 353140269.0, + "step": 13898 + }, + { + "epoch": 1.5263562486272786, + "grad_norm": 2.2902698516845703, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7150071859359741, + "num_tokens": 353163999.0, + "step": 13899 + }, + { + "epoch": 1.5264660663298923, + "grad_norm": 2.4885716438293457, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.7319884300231934, + "num_tokens": 353184601.0, + "step": 13900 + }, + { + "epoch": 1.526575884032506, + "grad_norm": 2.3306243419647217, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7113913297653198, + "num_tokens": 353208619.0, + "step": 13901 + }, + { + "epoch": 1.5266857017351199, + "grad_norm": 2.370434284210205, + "learning_rate": 1e-06, + "loss": 0.8167, + "mean_token_accuracy": 0.7454757690429688, + "num_tokens": 353231117.0, + "step": 13902 + }, + { + "epoch": 1.5267955194377334, + "grad_norm": 2.3221049308776855, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7140575647354126, + "num_tokens": 353255790.0, + "step": 13903 + }, + { + "epoch": 1.526905337140347, + "grad_norm": 2.2254638671875, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7121057510375977, + "num_tokens": 353280853.0, + "step": 13904 + }, + { + "epoch": 1.5270151548429607, + "grad_norm": 2.5609352588653564, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7158288955688477, + "num_tokens": 353301696.0, + "step": 13905 + }, + { + "epoch": 1.5271249725455744, + "grad_norm": 2.3960442543029785, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7168669700622559, + "num_tokens": 353324303.0, + "step": 13906 + }, + { + "epoch": 1.527234790248188, + "grad_norm": 2.037194013595581, + "learning_rate": 1e-06, + "loss": 0.8608, + "mean_token_accuracy": 0.7310001254081726, + "num_tokens": 353354316.0, + "step": 13907 + }, + { + "epoch": 1.5273446079508015, + "grad_norm": 2.483278512954712, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7283673286437988, + "num_tokens": 353376623.0, + "step": 13908 + }, + { + "epoch": 1.5274544256534153, + "grad_norm": 2.341320514678955, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7441115379333496, + "num_tokens": 353400281.0, + "step": 13909 + }, + { + "epoch": 1.527564243356029, + "grad_norm": 2.7296860218048096, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7369222640991211, + "num_tokens": 353419175.0, + "step": 13910 + }, + { + "epoch": 1.5276740610586428, + "grad_norm": 2.420318603515625, + "learning_rate": 1e-06, + "loss": 0.8299, + "mean_token_accuracy": 0.7366039156913757, + "num_tokens": 353441353.0, + "step": 13911 + }, + { + "epoch": 1.5277838787612563, + "grad_norm": 2.121161937713623, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.724037766456604, + "num_tokens": 353468318.0, + "step": 13912 + }, + { + "epoch": 1.5278936964638699, + "grad_norm": 2.2256991863250732, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7139778733253479, + "num_tokens": 353494872.0, + "step": 13913 + }, + { + "epoch": 1.5280035141664836, + "grad_norm": 2.604320526123047, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7275522947311401, + "num_tokens": 353515222.0, + "step": 13914 + }, + { + "epoch": 1.5281133318690974, + "grad_norm": 2.213792324066162, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7253987789154053, + "num_tokens": 353541462.0, + "step": 13915 + }, + { + "epoch": 1.5282231495717111, + "grad_norm": 2.3414692878723145, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7089099287986755, + "num_tokens": 353567336.0, + "step": 13916 + }, + { + "epoch": 1.5283329672743247, + "grad_norm": 2.1056270599365234, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7105942368507385, + "num_tokens": 353596374.0, + "step": 13917 + }, + { + "epoch": 1.5284427849769382, + "grad_norm": 2.4057157039642334, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7266556024551392, + "num_tokens": 353619620.0, + "step": 13918 + }, + { + "epoch": 1.528552602679552, + "grad_norm": 2.63213849067688, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7434544563293457, + "num_tokens": 353640775.0, + "step": 13919 + }, + { + "epoch": 1.5286624203821657, + "grad_norm": 2.40234375, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7263867259025574, + "num_tokens": 353663481.0, + "step": 13920 + }, + { + "epoch": 1.5287722380847792, + "grad_norm": 1.821236252784729, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7047855854034424, + "num_tokens": 353701279.0, + "step": 13921 + }, + { + "epoch": 1.5288820557873928, + "grad_norm": 2.4601104259490967, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7298907041549683, + "num_tokens": 353722634.0, + "step": 13922 + }, + { + "epoch": 1.5289918734900065, + "grad_norm": 2.4074254035949707, + "learning_rate": 1e-06, + "loss": 0.7657, + "mean_token_accuracy": 0.7541816234588623, + "num_tokens": 353746005.0, + "step": 13923 + }, + { + "epoch": 1.5291016911926203, + "grad_norm": 2.202578067779541, + "learning_rate": 1e-06, + "loss": 0.9814, + "mean_token_accuracy": 0.7003260254859924, + "num_tokens": 353774148.0, + "step": 13924 + }, + { + "epoch": 1.529211508895234, + "grad_norm": 2.653934955596924, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7492197155952454, + "num_tokens": 353792206.0, + "step": 13925 + }, + { + "epoch": 1.5293213265978476, + "grad_norm": 2.15509033203125, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7505762577056885, + "num_tokens": 353816542.0, + "step": 13926 + }, + { + "epoch": 1.5294311443004611, + "grad_norm": 2.326195001602173, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7205804586410522, + "num_tokens": 353840719.0, + "step": 13927 + }, + { + "epoch": 1.5295409620030749, + "grad_norm": 2.088170051574707, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7146751880645752, + "num_tokens": 353869296.0, + "step": 13928 + }, + { + "epoch": 1.5296507797056886, + "grad_norm": 2.4068455696105957, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7333250045776367, + "num_tokens": 353892809.0, + "step": 13929 + }, + { + "epoch": 1.5297605974083022, + "grad_norm": 2.1441681385040283, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7108259201049805, + "num_tokens": 353921277.0, + "step": 13930 + }, + { + "epoch": 1.529870415110916, + "grad_norm": 2.147808074951172, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7354073524475098, + "num_tokens": 353948246.0, + "step": 13931 + }, + { + "epoch": 1.5299802328135295, + "grad_norm": 2.3382227420806885, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7420885562896729, + "num_tokens": 353969879.0, + "step": 13932 + }, + { + "epoch": 1.5300900505161432, + "grad_norm": 2.449878215789795, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7339231967926025, + "num_tokens": 353990251.0, + "step": 13933 + }, + { + "epoch": 1.530199868218757, + "grad_norm": 2.5078277587890625, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7371871471405029, + "num_tokens": 354012798.0, + "step": 13934 + }, + { + "epoch": 1.5303096859213705, + "grad_norm": 2.6398532390594482, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7288496494293213, + "num_tokens": 354034522.0, + "step": 13935 + }, + { + "epoch": 1.530419503623984, + "grad_norm": 2.2229197025299072, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7112677097320557, + "num_tokens": 354063366.0, + "step": 13936 + }, + { + "epoch": 1.5305293213265978, + "grad_norm": 2.2629754543304443, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7098925113677979, + "num_tokens": 354087838.0, + "step": 13937 + }, + { + "epoch": 1.5306391390292116, + "grad_norm": 2.156977653503418, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7396153807640076, + "num_tokens": 354113606.0, + "step": 13938 + }, + { + "epoch": 1.5307489567318253, + "grad_norm": 2.25327730178833, + "learning_rate": 1e-06, + "loss": 0.8896, + "mean_token_accuracy": 0.7209675908088684, + "num_tokens": 354138262.0, + "step": 13939 + }, + { + "epoch": 1.5308587744344389, + "grad_norm": 1.9719599485397339, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7141355276107788, + "num_tokens": 354170155.0, + "step": 13940 + }, + { + "epoch": 1.5309685921370524, + "grad_norm": 2.095097064971924, + "learning_rate": 1e-06, + "loss": 1.0608, + "mean_token_accuracy": 0.6782222986221313, + "num_tokens": 354204257.0, + "step": 13941 + }, + { + "epoch": 1.5310784098396661, + "grad_norm": 2.5120418071746826, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7086862325668335, + "num_tokens": 354225232.0, + "step": 13942 + }, + { + "epoch": 1.53118822754228, + "grad_norm": 2.197406530380249, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7310387492179871, + "num_tokens": 354251554.0, + "step": 13943 + }, + { + "epoch": 1.5312980452448934, + "grad_norm": 2.097513198852539, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.705532431602478, + "num_tokens": 354281058.0, + "step": 13944 + }, + { + "epoch": 1.5314078629475072, + "grad_norm": 2.1542041301727295, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7152688503265381, + "num_tokens": 354309760.0, + "step": 13945 + }, + { + "epoch": 1.5315176806501207, + "grad_norm": 2.405710458755493, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7150100469589233, + "num_tokens": 354331580.0, + "step": 13946 + }, + { + "epoch": 1.5316274983527345, + "grad_norm": 2.0308659076690674, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7292399406433105, + "num_tokens": 354361590.0, + "step": 13947 + }, + { + "epoch": 1.5317373160553482, + "grad_norm": 2.0297563076019287, + "learning_rate": 1e-06, + "loss": 0.8829, + "mean_token_accuracy": 0.7224509716033936, + "num_tokens": 354394858.0, + "step": 13948 + }, + { + "epoch": 1.5318471337579618, + "grad_norm": 2.1677873134613037, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7006334662437439, + "num_tokens": 354421769.0, + "step": 13949 + }, + { + "epoch": 1.5319569514605753, + "grad_norm": 2.281567335128784, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7159126996994019, + "num_tokens": 354446779.0, + "step": 13950 + }, + { + "epoch": 1.532066769163189, + "grad_norm": 2.35955810546875, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7204715013504028, + "num_tokens": 354470358.0, + "step": 13951 + }, + { + "epoch": 1.5321765868658028, + "grad_norm": 1.983559250831604, + "learning_rate": 1e-06, + "loss": 1.0344, + "mean_token_accuracy": 0.680683970451355, + "num_tokens": 354505081.0, + "step": 13952 + }, + { + "epoch": 1.5322864045684166, + "grad_norm": 2.314051866531372, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7110779285430908, + "num_tokens": 354532098.0, + "step": 13953 + }, + { + "epoch": 1.5323962222710301, + "grad_norm": 2.2447688579559326, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7401094436645508, + "num_tokens": 354557064.0, + "step": 13954 + }, + { + "epoch": 1.5325060399736437, + "grad_norm": 2.5275118350982666, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7093790769577026, + "num_tokens": 354580650.0, + "step": 13955 + }, + { + "epoch": 1.5326158576762574, + "grad_norm": 2.223310708999634, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.71598219871521, + "num_tokens": 354606371.0, + "step": 13956 + }, + { + "epoch": 1.5327256753788712, + "grad_norm": 2.1627800464630127, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7127466201782227, + "num_tokens": 354634276.0, + "step": 13957 + }, + { + "epoch": 1.5328354930814847, + "grad_norm": 2.0402579307556152, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7110289335250854, + "num_tokens": 354664081.0, + "step": 13958 + }, + { + "epoch": 1.5329453107840982, + "grad_norm": 2.3620848655700684, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7240089774131775, + "num_tokens": 354687627.0, + "step": 13959 + }, + { + "epoch": 1.533055128486712, + "grad_norm": 2.145207166671753, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.726913332939148, + "num_tokens": 354715496.0, + "step": 13960 + }, + { + "epoch": 1.5331649461893258, + "grad_norm": 2.1748430728912354, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7242208123207092, + "num_tokens": 354743116.0, + "step": 13961 + }, + { + "epoch": 1.5332747638919395, + "grad_norm": 2.109754800796509, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6972286701202393, + "num_tokens": 354770925.0, + "step": 13962 + }, + { + "epoch": 1.533384581594553, + "grad_norm": 2.588266372680664, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7347432374954224, + "num_tokens": 354791778.0, + "step": 13963 + }, + { + "epoch": 1.5334943992971666, + "grad_norm": 2.132997989654541, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.6969132423400879, + "num_tokens": 354823343.0, + "step": 13964 + }, + { + "epoch": 1.5336042169997803, + "grad_norm": 2.558108329772949, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.71109938621521, + "num_tokens": 354843106.0, + "step": 13965 + }, + { + "epoch": 1.533714034702394, + "grad_norm": 2.435675859451294, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7163903713226318, + "num_tokens": 354866802.0, + "step": 13966 + }, + { + "epoch": 1.5338238524050078, + "grad_norm": 2.3847060203552246, + "learning_rate": 1e-06, + "loss": 1.0241, + "mean_token_accuracy": 0.6917336583137512, + "num_tokens": 354892611.0, + "step": 13967 + }, + { + "epoch": 1.5339336701076214, + "grad_norm": 2.5343990325927734, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7229466438293457, + "num_tokens": 354912593.0, + "step": 13968 + }, + { + "epoch": 1.534043487810235, + "grad_norm": 2.3906900882720947, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7252705693244934, + "num_tokens": 354933646.0, + "step": 13969 + }, + { + "epoch": 1.5341533055128487, + "grad_norm": 2.064764976501465, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.7027409076690674, + "num_tokens": 354963536.0, + "step": 13970 + }, + { + "epoch": 1.5342631232154624, + "grad_norm": 2.7970571517944336, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7249779105186462, + "num_tokens": 354982233.0, + "step": 13971 + }, + { + "epoch": 1.534372940918076, + "grad_norm": 2.0204334259033203, + "learning_rate": 1e-06, + "loss": 0.9941, + "mean_token_accuracy": 0.6973079442977905, + "num_tokens": 355014330.0, + "step": 13972 + }, + { + "epoch": 1.5344827586206895, + "grad_norm": 2.8479409217834473, + "learning_rate": 1e-06, + "loss": 0.7769, + "mean_token_accuracy": 0.7503591179847717, + "num_tokens": 355031485.0, + "step": 13973 + }, + { + "epoch": 1.5345925763233033, + "grad_norm": 2.003993272781372, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7022521495819092, + "num_tokens": 355061997.0, + "step": 13974 + }, + { + "epoch": 1.534702394025917, + "grad_norm": 2.5484235286712646, + "learning_rate": 1e-06, + "loss": 0.8645, + "mean_token_accuracy": 0.7322355508804321, + "num_tokens": 355082111.0, + "step": 13975 + }, + { + "epoch": 1.5348122117285308, + "grad_norm": 2.233980894088745, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7222779989242554, + "num_tokens": 355106565.0, + "step": 13976 + }, + { + "epoch": 1.5349220294311443, + "grad_norm": 1.9678382873535156, + "learning_rate": 1e-06, + "loss": 0.9976, + "mean_token_accuracy": 0.6922239661216736, + "num_tokens": 355138227.0, + "step": 13977 + }, + { + "epoch": 1.5350318471337578, + "grad_norm": 2.312547206878662, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.729068398475647, + "num_tokens": 355161712.0, + "step": 13978 + }, + { + "epoch": 1.5351416648363716, + "grad_norm": 2.352646827697754, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7117631435394287, + "num_tokens": 355187562.0, + "step": 13979 + }, + { + "epoch": 1.5352514825389854, + "grad_norm": 2.46488881111145, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7208698987960815, + "num_tokens": 355210242.0, + "step": 13980 + }, + { + "epoch": 1.535361300241599, + "grad_norm": 2.7679693698883057, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7278448939323425, + "num_tokens": 355230166.0, + "step": 13981 + }, + { + "epoch": 1.5354711179442126, + "grad_norm": 2.2490525245666504, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7209764719009399, + "num_tokens": 355255403.0, + "step": 13982 + }, + { + "epoch": 1.5355809356468262, + "grad_norm": 2.304121494293213, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7093659043312073, + "num_tokens": 355282223.0, + "step": 13983 + }, + { + "epoch": 1.53569075334944, + "grad_norm": 2.171013116836548, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7086547017097473, + "num_tokens": 355310830.0, + "step": 13984 + }, + { + "epoch": 1.5358005710520537, + "grad_norm": 2.3549082279205322, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.722795844078064, + "num_tokens": 355334354.0, + "step": 13985 + }, + { + "epoch": 1.5359103887546672, + "grad_norm": 2.345987558364868, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.7005749940872192, + "num_tokens": 355358968.0, + "step": 13986 + }, + { + "epoch": 1.5360202064572808, + "grad_norm": 2.252737522125244, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7115190625190735, + "num_tokens": 355383920.0, + "step": 13987 + }, + { + "epoch": 1.5361300241598945, + "grad_norm": 2.44468355178833, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7338705062866211, + "num_tokens": 355405589.0, + "step": 13988 + }, + { + "epoch": 1.5362398418625083, + "grad_norm": 2.0929927825927734, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7291051745414734, + "num_tokens": 355434547.0, + "step": 13989 + }, + { + "epoch": 1.536349659565122, + "grad_norm": 2.222227096557617, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.7058553695678711, + "num_tokens": 355461476.0, + "step": 13990 + }, + { + "epoch": 1.5364594772677356, + "grad_norm": 2.159153699874878, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7159836292266846, + "num_tokens": 355489317.0, + "step": 13991 + }, + { + "epoch": 1.536569294970349, + "grad_norm": 2.2475152015686035, + "learning_rate": 1e-06, + "loss": 0.9842, + "mean_token_accuracy": 0.6969811916351318, + "num_tokens": 355514976.0, + "step": 13992 + }, + { + "epoch": 1.5366791126729629, + "grad_norm": 2.453183889389038, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7099260091781616, + "num_tokens": 355538031.0, + "step": 13993 + }, + { + "epoch": 1.5367889303755766, + "grad_norm": 2.451319932937622, + "learning_rate": 1e-06, + "loss": 0.8659, + "mean_token_accuracy": 0.7302272319793701, + "num_tokens": 355559906.0, + "step": 13994 + }, + { + "epoch": 1.5368987480781902, + "grad_norm": 2.2535481452941895, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.6980644464492798, + "num_tokens": 355585529.0, + "step": 13995 + }, + { + "epoch": 1.537008565780804, + "grad_norm": 2.2172513008117676, + "learning_rate": 1e-06, + "loss": 0.8904, + "mean_token_accuracy": 0.724134087562561, + "num_tokens": 355610969.0, + "step": 13996 + }, + { + "epoch": 1.5371183834834174, + "grad_norm": 2.354114055633545, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7535287141799927, + "num_tokens": 355632774.0, + "step": 13997 + }, + { + "epoch": 1.5372282011860312, + "grad_norm": 1.9334731101989746, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7423765659332275, + "num_tokens": 355665098.0, + "step": 13998 + }, + { + "epoch": 1.537338018888645, + "grad_norm": 2.246027946472168, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.725258469581604, + "num_tokens": 355690229.0, + "step": 13999 + }, + { + "epoch": 1.5374478365912585, + "grad_norm": 2.2918319702148438, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.700502872467041, + "num_tokens": 355713619.0, + "step": 14000 + }, + { + "epoch": 1.537557654293872, + "grad_norm": 3.093822956085205, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7202482223510742, + "num_tokens": 355733289.0, + "step": 14001 + }, + { + "epoch": 1.5376674719964858, + "grad_norm": 1.990682601928711, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7153120636940002, + "num_tokens": 355763795.0, + "step": 14002 + }, + { + "epoch": 1.5377772896990995, + "grad_norm": 2.27421236038208, + "learning_rate": 1e-06, + "loss": 0.9436, + "mean_token_accuracy": 0.7203010320663452, + "num_tokens": 355789728.0, + "step": 14003 + }, + { + "epoch": 1.5378871074017133, + "grad_norm": 2.5459461212158203, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7060365676879883, + "num_tokens": 355810928.0, + "step": 14004 + }, + { + "epoch": 1.5379969251043268, + "grad_norm": 2.265808343887329, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7253730893135071, + "num_tokens": 355836374.0, + "step": 14005 + }, + { + "epoch": 1.5381067428069404, + "grad_norm": 2.230698347091675, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.7015490531921387, + "num_tokens": 355864359.0, + "step": 14006 + }, + { + "epoch": 1.5382165605095541, + "grad_norm": 2.1036150455474854, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7485837340354919, + "num_tokens": 355889873.0, + "step": 14007 + }, + { + "epoch": 1.5383263782121679, + "grad_norm": 2.4138379096984863, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7245224118232727, + "num_tokens": 355912097.0, + "step": 14008 + }, + { + "epoch": 1.5384361959147814, + "grad_norm": 2.1401312351226807, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7386347055435181, + "num_tokens": 355940355.0, + "step": 14009 + }, + { + "epoch": 1.538546013617395, + "grad_norm": 1.960645079612732, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.687785804271698, + "num_tokens": 355973307.0, + "step": 14010 + }, + { + "epoch": 1.5386558313200087, + "grad_norm": 2.209329843521118, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7307246327400208, + "num_tokens": 355997894.0, + "step": 14011 + }, + { + "epoch": 1.5387656490226225, + "grad_norm": 2.4426193237304688, + "learning_rate": 1e-06, + "loss": 0.852, + "mean_token_accuracy": 0.7374016642570496, + "num_tokens": 356019069.0, + "step": 14012 + }, + { + "epoch": 1.5388754667252362, + "grad_norm": 2.318326473236084, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7056052088737488, + "num_tokens": 356043239.0, + "step": 14013 + }, + { + "epoch": 1.5389852844278498, + "grad_norm": 2.0802783966064453, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.7139161825180054, + "num_tokens": 356072918.0, + "step": 14014 + }, + { + "epoch": 1.5390951021304633, + "grad_norm": 2.4990735054016113, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7524775266647339, + "num_tokens": 356095454.0, + "step": 14015 + }, + { + "epoch": 1.539204919833077, + "grad_norm": 2.5046584606170654, + "learning_rate": 1e-06, + "loss": 0.8303, + "mean_token_accuracy": 0.7458967566490173, + "num_tokens": 356117328.0, + "step": 14016 + }, + { + "epoch": 1.5393147375356908, + "grad_norm": 2.3363397121429443, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7252981066703796, + "num_tokens": 356140626.0, + "step": 14017 + }, + { + "epoch": 1.5394245552383046, + "grad_norm": 2.169262170791626, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7161909341812134, + "num_tokens": 356167000.0, + "step": 14018 + }, + { + "epoch": 1.539534372940918, + "grad_norm": 2.5930943489074707, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7412275671958923, + "num_tokens": 356187962.0, + "step": 14019 + }, + { + "epoch": 1.5396441906435316, + "grad_norm": 2.2003915309906006, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7181509733200073, + "num_tokens": 356215017.0, + "step": 14020 + }, + { + "epoch": 1.5397540083461454, + "grad_norm": 2.343475341796875, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.6998905539512634, + "num_tokens": 356240208.0, + "step": 14021 + }, + { + "epoch": 1.5398638260487592, + "grad_norm": 2.136648654937744, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7130163908004761, + "num_tokens": 356266431.0, + "step": 14022 + }, + { + "epoch": 1.5399736437513727, + "grad_norm": 2.144176483154297, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7198052406311035, + "num_tokens": 356294034.0, + "step": 14023 + }, + { + "epoch": 1.5400834614539862, + "grad_norm": 2.7754130363464355, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7430380582809448, + "num_tokens": 356312561.0, + "step": 14024 + }, + { + "epoch": 1.5401932791566, + "grad_norm": 2.3599703311920166, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.730955958366394, + "num_tokens": 356334630.0, + "step": 14025 + }, + { + "epoch": 1.5403030968592137, + "grad_norm": 1.8090338706970215, + "learning_rate": 1e-06, + "loss": 1.0469, + "mean_token_accuracy": 0.6847662329673767, + "num_tokens": 356370171.0, + "step": 14026 + }, + { + "epoch": 1.5404129145618275, + "grad_norm": 2.0704236030578613, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7127230167388916, + "num_tokens": 356400366.0, + "step": 14027 + }, + { + "epoch": 1.540522732264441, + "grad_norm": 2.4375314712524414, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7239052653312683, + "num_tokens": 356421768.0, + "step": 14028 + }, + { + "epoch": 1.5406325499670546, + "grad_norm": 1.879895567893982, + "learning_rate": 1e-06, + "loss": 0.9525, + "mean_token_accuracy": 0.7170923352241516, + "num_tokens": 356455754.0, + "step": 14029 + }, + { + "epoch": 1.5407423676696683, + "grad_norm": 2.3354744911193848, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7004474997520447, + "num_tokens": 356478859.0, + "step": 14030 + }, + { + "epoch": 1.540852185372282, + "grad_norm": 2.0391743183135986, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7069922685623169, + "num_tokens": 356508919.0, + "step": 14031 + }, + { + "epoch": 1.5409620030748958, + "grad_norm": 2.2985799312591553, + "learning_rate": 1e-06, + "loss": 0.8971, + "mean_token_accuracy": 0.7285209894180298, + "num_tokens": 356532413.0, + "step": 14032 + }, + { + "epoch": 1.5410718207775094, + "grad_norm": 2.380694627761841, + "learning_rate": 1e-06, + "loss": 0.8314, + "mean_token_accuracy": 0.7298771739006042, + "num_tokens": 356554962.0, + "step": 14033 + }, + { + "epoch": 1.541181638480123, + "grad_norm": 2.479739189147949, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7109719514846802, + "num_tokens": 356577076.0, + "step": 14034 + }, + { + "epoch": 1.5412914561827367, + "grad_norm": 2.356936454772949, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7051858901977539, + "num_tokens": 356602330.0, + "step": 14035 + }, + { + "epoch": 1.5414012738853504, + "grad_norm": 2.334251880645752, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7274959087371826, + "num_tokens": 356626225.0, + "step": 14036 + }, + { + "epoch": 1.541511091587964, + "grad_norm": 1.9828412532806396, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7038087844848633, + "num_tokens": 356658901.0, + "step": 14037 + }, + { + "epoch": 1.5416209092905775, + "grad_norm": 2.529672384262085, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7428188323974609, + "num_tokens": 356678858.0, + "step": 14038 + }, + { + "epoch": 1.5417307269931912, + "grad_norm": 2.5141549110412598, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.707375168800354, + "num_tokens": 356700353.0, + "step": 14039 + }, + { + "epoch": 1.541840544695805, + "grad_norm": 2.704658031463623, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7212985157966614, + "num_tokens": 356719738.0, + "step": 14040 + }, + { + "epoch": 1.5419503623984188, + "grad_norm": 2.0564792156219482, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.726917564868927, + "num_tokens": 356749330.0, + "step": 14041 + }, + { + "epoch": 1.5420601801010323, + "grad_norm": 2.6120102405548096, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7142435312271118, + "num_tokens": 356771450.0, + "step": 14042 + }, + { + "epoch": 1.5421699978036458, + "grad_norm": 2.3170998096466064, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7327970862388611, + "num_tokens": 356794455.0, + "step": 14043 + }, + { + "epoch": 1.5422798155062596, + "grad_norm": 2.0686159133911133, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7049343585968018, + "num_tokens": 356824129.0, + "step": 14044 + }, + { + "epoch": 1.5423896332088733, + "grad_norm": 2.512681484222412, + "learning_rate": 1e-06, + "loss": 0.9659, + "mean_token_accuracy": 0.7074482440948486, + "num_tokens": 356845744.0, + "step": 14045 + }, + { + "epoch": 1.5424994509114869, + "grad_norm": 2.313507556915283, + "learning_rate": 1e-06, + "loss": 0.9945, + "mean_token_accuracy": 0.7001742124557495, + "num_tokens": 356871185.0, + "step": 14046 + }, + { + "epoch": 1.5426092686141006, + "grad_norm": 2.0921292304992676, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7151985168457031, + "num_tokens": 356898892.0, + "step": 14047 + }, + { + "epoch": 1.5427190863167142, + "grad_norm": 2.253788948059082, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7161548733711243, + "num_tokens": 356924845.0, + "step": 14048 + }, + { + "epoch": 1.542828904019328, + "grad_norm": 2.2608485221862793, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.705646812915802, + "num_tokens": 356951429.0, + "step": 14049 + }, + { + "epoch": 1.5429387217219417, + "grad_norm": 2.410818576812744, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.7308858633041382, + "num_tokens": 356975409.0, + "step": 14050 + }, + { + "epoch": 1.5430485394245552, + "grad_norm": 2.026170015335083, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6833745241165161, + "num_tokens": 357008630.0, + "step": 14051 + }, + { + "epoch": 1.5431583571271688, + "grad_norm": 2.2703356742858887, + "learning_rate": 1e-06, + "loss": 0.9614, + "mean_token_accuracy": 0.7024253010749817, + "num_tokens": 357034828.0, + "step": 14052 + }, + { + "epoch": 1.5432681748297825, + "grad_norm": 2.241497755050659, + "learning_rate": 1e-06, + "loss": 0.9909, + "mean_token_accuracy": 0.6927189826965332, + "num_tokens": 357062481.0, + "step": 14053 + }, + { + "epoch": 1.5433779925323963, + "grad_norm": 2.245086669921875, + "learning_rate": 1e-06, + "loss": 0.9486, + "mean_token_accuracy": 0.7049930095672607, + "num_tokens": 357088085.0, + "step": 14054 + }, + { + "epoch": 1.54348781023501, + "grad_norm": 2.421520233154297, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7421654462814331, + "num_tokens": 357108747.0, + "step": 14055 + }, + { + "epoch": 1.5435976279376236, + "grad_norm": 2.205120801925659, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7052212953567505, + "num_tokens": 357134797.0, + "step": 14056 + }, + { + "epoch": 1.543707445640237, + "grad_norm": 2.2198071479797363, + "learning_rate": 1e-06, + "loss": 0.927, + "mean_token_accuracy": 0.7134608626365662, + "num_tokens": 357163429.0, + "step": 14057 + }, + { + "epoch": 1.5438172633428509, + "grad_norm": 1.8833132982254028, + "learning_rate": 1e-06, + "loss": 0.9852, + "mean_token_accuracy": 0.7024612426757812, + "num_tokens": 357197343.0, + "step": 14058 + }, + { + "epoch": 1.5439270810454646, + "grad_norm": 2.5443317890167236, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.707452118396759, + "num_tokens": 357219408.0, + "step": 14059 + }, + { + "epoch": 1.5440368987480781, + "grad_norm": 2.302377939224243, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7105099558830261, + "num_tokens": 357244134.0, + "step": 14060 + }, + { + "epoch": 1.544146716450692, + "grad_norm": 2.200106382369995, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7075283527374268, + "num_tokens": 357271133.0, + "step": 14061 + }, + { + "epoch": 1.5442565341533054, + "grad_norm": 2.346231460571289, + "learning_rate": 1e-06, + "loss": 0.7127, + "mean_token_accuracy": 0.7726131081581116, + "num_tokens": 357292845.0, + "step": 14062 + }, + { + "epoch": 1.5443663518559192, + "grad_norm": 2.347788095474243, + "learning_rate": 1e-06, + "loss": 0.8248, + "mean_token_accuracy": 0.7414542436599731, + "num_tokens": 357316528.0, + "step": 14063 + }, + { + "epoch": 1.544476169558533, + "grad_norm": 2.258563756942749, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7181373834609985, + "num_tokens": 357343420.0, + "step": 14064 + }, + { + "epoch": 1.5445859872611465, + "grad_norm": 2.3150670528411865, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7413754463195801, + "num_tokens": 357366485.0, + "step": 14065 + }, + { + "epoch": 1.54469580496376, + "grad_norm": 2.506146192550659, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7242604494094849, + "num_tokens": 357386747.0, + "step": 14066 + }, + { + "epoch": 1.5448056226663738, + "grad_norm": 2.1360292434692383, + "learning_rate": 1e-06, + "loss": 0.769, + "mean_token_accuracy": 0.7521714568138123, + "num_tokens": 357411010.0, + "step": 14067 + }, + { + "epoch": 1.5449154403689875, + "grad_norm": 2.323854684829712, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7285340428352356, + "num_tokens": 357435399.0, + "step": 14068 + }, + { + "epoch": 1.5450252580716013, + "grad_norm": 2.3761916160583496, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.702153205871582, + "num_tokens": 357459016.0, + "step": 14069 + }, + { + "epoch": 1.5451350757742148, + "grad_norm": 2.0541417598724365, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.6943731904029846, + "num_tokens": 357489753.0, + "step": 14070 + }, + { + "epoch": 1.5452448934768284, + "grad_norm": 2.429837703704834, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.7273232340812683, + "num_tokens": 357511531.0, + "step": 14071 + }, + { + "epoch": 1.5453547111794421, + "grad_norm": 1.9577447175979614, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7115175724029541, + "num_tokens": 357544724.0, + "step": 14072 + }, + { + "epoch": 1.5454645288820559, + "grad_norm": 2.2290570735931396, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.6990950703620911, + "num_tokens": 357573309.0, + "step": 14073 + }, + { + "epoch": 1.5455743465846694, + "grad_norm": 2.4242916107177734, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7306606769561768, + "num_tokens": 357594129.0, + "step": 14074 + }, + { + "epoch": 1.545684164287283, + "grad_norm": 2.455458879470825, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7343737483024597, + "num_tokens": 357615124.0, + "step": 14075 + }, + { + "epoch": 1.5457939819898967, + "grad_norm": 2.1448960304260254, + "learning_rate": 1e-06, + "loss": 0.9645, + "mean_token_accuracy": 0.7079739570617676, + "num_tokens": 357641526.0, + "step": 14076 + }, + { + "epoch": 1.5459037996925105, + "grad_norm": 2.466460943222046, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.7070673704147339, + "num_tokens": 357665738.0, + "step": 14077 + }, + { + "epoch": 1.5460136173951242, + "grad_norm": 2.3783695697784424, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7117435336112976, + "num_tokens": 357688763.0, + "step": 14078 + }, + { + "epoch": 1.5461234350977378, + "grad_norm": 2.1620373725891113, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.6956583857536316, + "num_tokens": 357719715.0, + "step": 14079 + }, + { + "epoch": 1.5462332528003513, + "grad_norm": 2.444824695587158, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7328755855560303, + "num_tokens": 357742269.0, + "step": 14080 + }, + { + "epoch": 1.546343070502965, + "grad_norm": 2.3216681480407715, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7195758819580078, + "num_tokens": 357767962.0, + "step": 14081 + }, + { + "epoch": 1.5464528882055788, + "grad_norm": 2.4600303173065186, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7149760723114014, + "num_tokens": 357789997.0, + "step": 14082 + }, + { + "epoch": 1.5465627059081926, + "grad_norm": 2.4025368690490723, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.705268144607544, + "num_tokens": 357818717.0, + "step": 14083 + }, + { + "epoch": 1.546672523610806, + "grad_norm": 2.5157315731048584, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7176417112350464, + "num_tokens": 357841372.0, + "step": 14084 + }, + { + "epoch": 1.5467823413134196, + "grad_norm": 2.088545322418213, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7208969593048096, + "num_tokens": 357868666.0, + "step": 14085 + }, + { + "epoch": 1.5468921590160334, + "grad_norm": 2.2470896244049072, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7146146297454834, + "num_tokens": 357893249.0, + "step": 14086 + }, + { + "epoch": 1.5470019767186471, + "grad_norm": 2.2149651050567627, + "learning_rate": 1e-06, + "loss": 0.8569, + "mean_token_accuracy": 0.7257472276687622, + "num_tokens": 357919503.0, + "step": 14087 + }, + { + "epoch": 1.5471117944212607, + "grad_norm": 2.49731707572937, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7209261655807495, + "num_tokens": 357942180.0, + "step": 14088 + }, + { + "epoch": 1.5472216121238742, + "grad_norm": 2.788569211959839, + "learning_rate": 1e-06, + "loss": 0.8878, + "mean_token_accuracy": 0.7295058965682983, + "num_tokens": 357961477.0, + "step": 14089 + }, + { + "epoch": 1.547331429826488, + "grad_norm": 2.481196403503418, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7340009212493896, + "num_tokens": 357982520.0, + "step": 14090 + }, + { + "epoch": 1.5474412475291017, + "grad_norm": 2.273380756378174, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7092957496643066, + "num_tokens": 358008135.0, + "step": 14091 + }, + { + "epoch": 1.5475510652317155, + "grad_norm": 2.5205576419830322, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7283515930175781, + "num_tokens": 358029008.0, + "step": 14092 + }, + { + "epoch": 1.547660882934329, + "grad_norm": 2.415498733520508, + "learning_rate": 1e-06, + "loss": 0.8749, + "mean_token_accuracy": 0.7342690229415894, + "num_tokens": 358051324.0, + "step": 14093 + }, + { + "epoch": 1.5477707006369426, + "grad_norm": 2.400679349899292, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7244399189949036, + "num_tokens": 358073686.0, + "step": 14094 + }, + { + "epoch": 1.5478805183395563, + "grad_norm": 2.2920548915863037, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7244841456413269, + "num_tokens": 358096925.0, + "step": 14095 + }, + { + "epoch": 1.54799033604217, + "grad_norm": 2.324918031692505, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7240646481513977, + "num_tokens": 358122719.0, + "step": 14096 + }, + { + "epoch": 1.5481001537447838, + "grad_norm": 2.0835793018341064, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7311128973960876, + "num_tokens": 358150456.0, + "step": 14097 + }, + { + "epoch": 1.5482099714473974, + "grad_norm": 2.4800961017608643, + "learning_rate": 1e-06, + "loss": 0.9339, + "mean_token_accuracy": 0.715484619140625, + "num_tokens": 358172916.0, + "step": 14098 + }, + { + "epoch": 1.548319789150011, + "grad_norm": 2.100088119506836, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7246600389480591, + "num_tokens": 358201350.0, + "step": 14099 + }, + { + "epoch": 1.5484296068526247, + "grad_norm": 1.9700541496276855, + "learning_rate": 1e-06, + "loss": 0.9807, + "mean_token_accuracy": 0.7051064372062683, + "num_tokens": 358233622.0, + "step": 14100 + }, + { + "epoch": 1.5485394245552384, + "grad_norm": 2.371609926223755, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.725351095199585, + "num_tokens": 358256413.0, + "step": 14101 + }, + { + "epoch": 1.548649242257852, + "grad_norm": 2.1640799045562744, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6978060603141785, + "num_tokens": 358285102.0, + "step": 14102 + }, + { + "epoch": 1.5487590599604655, + "grad_norm": 2.1858737468719482, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7298452854156494, + "num_tokens": 358310481.0, + "step": 14103 + }, + { + "epoch": 1.5488688776630792, + "grad_norm": 2.1447489261627197, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7186057567596436, + "num_tokens": 358337665.0, + "step": 14104 + }, + { + "epoch": 1.548978695365693, + "grad_norm": 2.043813943862915, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.6985312104225159, + "num_tokens": 358368130.0, + "step": 14105 + }, + { + "epoch": 1.5490885130683067, + "grad_norm": 2.1679980754852295, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.6972497701644897, + "num_tokens": 358395899.0, + "step": 14106 + }, + { + "epoch": 1.5491983307709203, + "grad_norm": 2.276686668395996, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7220908403396606, + "num_tokens": 358420002.0, + "step": 14107 + }, + { + "epoch": 1.5493081484735338, + "grad_norm": 2.440929412841797, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7211581468582153, + "num_tokens": 358443289.0, + "step": 14108 + }, + { + "epoch": 1.5494179661761476, + "grad_norm": 2.401721715927124, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7192916870117188, + "num_tokens": 358464399.0, + "step": 14109 + }, + { + "epoch": 1.5495277838787613, + "grad_norm": 2.1793291568756104, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7133342027664185, + "num_tokens": 358493202.0, + "step": 14110 + }, + { + "epoch": 1.5496376015813749, + "grad_norm": 2.6649742126464844, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7289950847625732, + "num_tokens": 358513630.0, + "step": 14111 + }, + { + "epoch": 1.5497474192839886, + "grad_norm": 2.2042324542999268, + "learning_rate": 1e-06, + "loss": 0.999, + "mean_token_accuracy": 0.6943785548210144, + "num_tokens": 358541907.0, + "step": 14112 + }, + { + "epoch": 1.5498572369866022, + "grad_norm": 1.9679784774780273, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.725143551826477, + "num_tokens": 358574647.0, + "step": 14113 + }, + { + "epoch": 1.549967054689216, + "grad_norm": 2.5192601680755615, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.6998616456985474, + "num_tokens": 358596823.0, + "step": 14114 + }, + { + "epoch": 1.5500768723918297, + "grad_norm": 2.4062769412994385, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7308441996574402, + "num_tokens": 358618548.0, + "step": 14115 + }, + { + "epoch": 1.5501866900944432, + "grad_norm": 2.262411594390869, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7207843065261841, + "num_tokens": 358643982.0, + "step": 14116 + }, + { + "epoch": 1.5502965077970567, + "grad_norm": 2.7288522720336914, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7218936681747437, + "num_tokens": 358663370.0, + "step": 14117 + }, + { + "epoch": 1.5504063254996705, + "grad_norm": 2.2176644802093506, + "learning_rate": 1e-06, + "loss": 0.794, + "mean_token_accuracy": 0.7452100515365601, + "num_tokens": 358687115.0, + "step": 14118 + }, + { + "epoch": 1.5505161432022843, + "grad_norm": 2.468764543533325, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7349028587341309, + "num_tokens": 358707095.0, + "step": 14119 + }, + { + "epoch": 1.550625960904898, + "grad_norm": 2.2824246883392334, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7342537045478821, + "num_tokens": 358731173.0, + "step": 14120 + }, + { + "epoch": 1.5507357786075116, + "grad_norm": 2.4315614700317383, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7211461067199707, + "num_tokens": 358753912.0, + "step": 14121 + }, + { + "epoch": 1.550845596310125, + "grad_norm": 2.0061914920806885, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7228673696517944, + "num_tokens": 358785183.0, + "step": 14122 + }, + { + "epoch": 1.5509554140127388, + "grad_norm": 2.194979429244995, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7069858312606812, + "num_tokens": 358812580.0, + "step": 14123 + }, + { + "epoch": 1.5510652317153526, + "grad_norm": 2.3534927368164062, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7209963202476501, + "num_tokens": 358837267.0, + "step": 14124 + }, + { + "epoch": 1.5511750494179661, + "grad_norm": 2.098757266998291, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.72474604845047, + "num_tokens": 358865858.0, + "step": 14125 + }, + { + "epoch": 1.55128486712058, + "grad_norm": 2.1915342807769775, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7070174217224121, + "num_tokens": 358893051.0, + "step": 14126 + }, + { + "epoch": 1.5513946848231934, + "grad_norm": 2.398489236831665, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7475656270980835, + "num_tokens": 358914228.0, + "step": 14127 + }, + { + "epoch": 1.5515045025258072, + "grad_norm": 2.249922752380371, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7381025552749634, + "num_tokens": 358938252.0, + "step": 14128 + }, + { + "epoch": 1.551614320228421, + "grad_norm": 2.303720474243164, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.714932918548584, + "num_tokens": 358963156.0, + "step": 14129 + }, + { + "epoch": 1.5517241379310345, + "grad_norm": 2.4798552989959717, + "learning_rate": 1e-06, + "loss": 0.7969, + "mean_token_accuracy": 0.7401437759399414, + "num_tokens": 358986180.0, + "step": 14130 + }, + { + "epoch": 1.551833955633648, + "grad_norm": 2.0973598957061768, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7250742316246033, + "num_tokens": 359012385.0, + "step": 14131 + }, + { + "epoch": 1.5519437733362618, + "grad_norm": 2.316579580307007, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7225945591926575, + "num_tokens": 359039591.0, + "step": 14132 + }, + { + "epoch": 1.5520535910388755, + "grad_norm": 2.2844462394714355, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7290585041046143, + "num_tokens": 359064193.0, + "step": 14133 + }, + { + "epoch": 1.5521634087414893, + "grad_norm": 2.7149407863616943, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7444075345993042, + "num_tokens": 359083535.0, + "step": 14134 + }, + { + "epoch": 1.5522732264441028, + "grad_norm": 2.040766954421997, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7297765016555786, + "num_tokens": 359114670.0, + "step": 14135 + }, + { + "epoch": 1.5523830441467164, + "grad_norm": 2.19685435295105, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7178764343261719, + "num_tokens": 359140800.0, + "step": 14136 + }, + { + "epoch": 1.55249286184933, + "grad_norm": 1.9838999509811401, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7379878759384155, + "num_tokens": 359170311.0, + "step": 14137 + }, + { + "epoch": 1.5526026795519439, + "grad_norm": 2.3954172134399414, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.6981741189956665, + "num_tokens": 359194462.0, + "step": 14138 + }, + { + "epoch": 1.5527124972545574, + "grad_norm": 2.315549373626709, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7095934152603149, + "num_tokens": 359218983.0, + "step": 14139 + }, + { + "epoch": 1.552822314957171, + "grad_norm": 2.4705774784088135, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7202537059783936, + "num_tokens": 359242909.0, + "step": 14140 + }, + { + "epoch": 1.5529321326597847, + "grad_norm": 2.3155405521392822, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7160433530807495, + "num_tokens": 359266189.0, + "step": 14141 + }, + { + "epoch": 1.5530419503623984, + "grad_norm": 2.0377237796783447, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7056071758270264, + "num_tokens": 359298705.0, + "step": 14142 + }, + { + "epoch": 1.5531517680650122, + "grad_norm": 2.3938963413238525, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7221615314483643, + "num_tokens": 359321085.0, + "step": 14143 + }, + { + "epoch": 1.5532615857676257, + "grad_norm": 2.544278860092163, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7193307876586914, + "num_tokens": 359341136.0, + "step": 14144 + }, + { + "epoch": 1.5533714034702393, + "grad_norm": 2.10099196434021, + "learning_rate": 1e-06, + "loss": 0.9204, + "mean_token_accuracy": 0.7109680771827698, + "num_tokens": 359367691.0, + "step": 14145 + }, + { + "epoch": 1.553481221172853, + "grad_norm": 2.1549301147460938, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7311094403266907, + "num_tokens": 359393762.0, + "step": 14146 + }, + { + "epoch": 1.5535910388754668, + "grad_norm": 2.362780809402466, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7105191349983215, + "num_tokens": 359418858.0, + "step": 14147 + }, + { + "epoch": 1.5537008565780805, + "grad_norm": 2.4793882369995117, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7374230623245239, + "num_tokens": 359441864.0, + "step": 14148 + }, + { + "epoch": 1.553810674280694, + "grad_norm": 2.36053204536438, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7256971597671509, + "num_tokens": 359464836.0, + "step": 14149 + }, + { + "epoch": 1.5539204919833076, + "grad_norm": 2.294409990310669, + "learning_rate": 1e-06, + "loss": 0.9874, + "mean_token_accuracy": 0.7045227289199829, + "num_tokens": 359487959.0, + "step": 14150 + }, + { + "epoch": 1.5540303096859214, + "grad_norm": 2.4328861236572266, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7487424612045288, + "num_tokens": 359509360.0, + "step": 14151 + }, + { + "epoch": 1.5541401273885351, + "grad_norm": 2.267594814300537, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7219300270080566, + "num_tokens": 359535019.0, + "step": 14152 + }, + { + "epoch": 1.5542499450911487, + "grad_norm": 2.074812650680542, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7198504209518433, + "num_tokens": 359563610.0, + "step": 14153 + }, + { + "epoch": 1.5543597627937622, + "grad_norm": 2.458364725112915, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.720180869102478, + "num_tokens": 359586480.0, + "step": 14154 + }, + { + "epoch": 1.554469580496376, + "grad_norm": 2.1783854961395264, + "learning_rate": 1e-06, + "loss": 0.8467, + "mean_token_accuracy": 0.7337064743041992, + "num_tokens": 359613765.0, + "step": 14155 + }, + { + "epoch": 1.5545793981989897, + "grad_norm": 2.0831780433654785, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6971498727798462, + "num_tokens": 359644916.0, + "step": 14156 + }, + { + "epoch": 1.5546892159016035, + "grad_norm": 2.231109619140625, + "learning_rate": 1e-06, + "loss": 0.8913, + "mean_token_accuracy": 0.7293363213539124, + "num_tokens": 359673230.0, + "step": 14157 + }, + { + "epoch": 1.554799033604217, + "grad_norm": 2.075275421142578, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7089903354644775, + "num_tokens": 359703590.0, + "step": 14158 + }, + { + "epoch": 1.5549088513068305, + "grad_norm": 2.0082077980041504, + "learning_rate": 1e-06, + "loss": 0.953, + "mean_token_accuracy": 0.7075161933898926, + "num_tokens": 359733492.0, + "step": 14159 + }, + { + "epoch": 1.5550186690094443, + "grad_norm": 2.1759912967681885, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.701221227645874, + "num_tokens": 359763524.0, + "step": 14160 + }, + { + "epoch": 1.555128486712058, + "grad_norm": 2.173553943634033, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7184785604476929, + "num_tokens": 359791832.0, + "step": 14161 + }, + { + "epoch": 1.5552383044146716, + "grad_norm": 2.4717206954956055, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6981774568557739, + "num_tokens": 359815378.0, + "step": 14162 + }, + { + "epoch": 1.5553481221172853, + "grad_norm": 2.2740209102630615, + "learning_rate": 1e-06, + "loss": 0.9831, + "mean_token_accuracy": 0.699385404586792, + "num_tokens": 359838523.0, + "step": 14163 + }, + { + "epoch": 1.5554579398198989, + "grad_norm": 2.5448811054229736, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7328585386276245, + "num_tokens": 359858888.0, + "step": 14164 + }, + { + "epoch": 1.5555677575225126, + "grad_norm": 2.239522933959961, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7167530059814453, + "num_tokens": 359884052.0, + "step": 14165 + }, + { + "epoch": 1.5556775752251264, + "grad_norm": 2.289506435394287, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7060996294021606, + "num_tokens": 359909389.0, + "step": 14166 + }, + { + "epoch": 1.55578739292774, + "grad_norm": 2.1070916652679443, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7112038135528564, + "num_tokens": 359938146.0, + "step": 14167 + }, + { + "epoch": 1.5558972106303535, + "grad_norm": 2.2084012031555176, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7067452669143677, + "num_tokens": 359964663.0, + "step": 14168 + }, + { + "epoch": 1.5560070283329672, + "grad_norm": 2.2664103507995605, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7216609120368958, + "num_tokens": 359989160.0, + "step": 14169 + }, + { + "epoch": 1.556116846035581, + "grad_norm": 2.164181709289551, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7178826332092285, + "num_tokens": 360017187.0, + "step": 14170 + }, + { + "epoch": 1.5562266637381947, + "grad_norm": 2.2705252170562744, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7167336344718933, + "num_tokens": 360042676.0, + "step": 14171 + }, + { + "epoch": 1.5563364814408083, + "grad_norm": 2.5709919929504395, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7417818307876587, + "num_tokens": 360062462.0, + "step": 14172 + }, + { + "epoch": 1.5564462991434218, + "grad_norm": 2.376509189605713, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7135730385780334, + "num_tokens": 360086861.0, + "step": 14173 + }, + { + "epoch": 1.5565561168460356, + "grad_norm": 2.5398595333099365, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7256389856338501, + "num_tokens": 360109125.0, + "step": 14174 + }, + { + "epoch": 1.5566659345486493, + "grad_norm": 2.2010550498962402, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7055352330207825, + "num_tokens": 360137316.0, + "step": 14175 + }, + { + "epoch": 1.5567757522512629, + "grad_norm": 2.763502597808838, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7055367827415466, + "num_tokens": 360156125.0, + "step": 14176 + }, + { + "epoch": 1.5568855699538766, + "grad_norm": 2.5456297397613525, + "learning_rate": 1e-06, + "loss": 0.8496, + "mean_token_accuracy": 0.7352996468544006, + "num_tokens": 360177429.0, + "step": 14177 + }, + { + "epoch": 1.5569953876564901, + "grad_norm": 2.520488977432251, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7207338213920593, + "num_tokens": 360197336.0, + "step": 14178 + }, + { + "epoch": 1.557105205359104, + "grad_norm": 2.292020082473755, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7203788757324219, + "num_tokens": 360221805.0, + "step": 14179 + }, + { + "epoch": 1.5572150230617177, + "grad_norm": 2.098963975906372, + "learning_rate": 1e-06, + "loss": 0.9326, + "mean_token_accuracy": 0.707319974899292, + "num_tokens": 360247969.0, + "step": 14180 + }, + { + "epoch": 1.5573248407643312, + "grad_norm": 2.25602126121521, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.735203742980957, + "num_tokens": 360271359.0, + "step": 14181 + }, + { + "epoch": 1.5574346584669447, + "grad_norm": 2.2332019805908203, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7216449975967407, + "num_tokens": 360295618.0, + "step": 14182 + }, + { + "epoch": 1.5575444761695585, + "grad_norm": 2.3272342681884766, + "learning_rate": 1e-06, + "loss": 0.77, + "mean_token_accuracy": 0.7539538145065308, + "num_tokens": 360316381.0, + "step": 14183 + }, + { + "epoch": 1.5576542938721722, + "grad_norm": 2.4821889400482178, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.72175133228302, + "num_tokens": 360338635.0, + "step": 14184 + }, + { + "epoch": 1.557764111574786, + "grad_norm": 2.165830612182617, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7267054915428162, + "num_tokens": 360365968.0, + "step": 14185 + }, + { + "epoch": 1.5578739292773995, + "grad_norm": 2.2818710803985596, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.7228373885154724, + "num_tokens": 360389805.0, + "step": 14186 + }, + { + "epoch": 1.557983746980013, + "grad_norm": 2.38666033744812, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.7047305107116699, + "num_tokens": 360414422.0, + "step": 14187 + }, + { + "epoch": 1.5580935646826268, + "grad_norm": 2.1788442134857178, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7292321920394897, + "num_tokens": 360444186.0, + "step": 14188 + }, + { + "epoch": 1.5582033823852406, + "grad_norm": 1.9402568340301514, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.704698920249939, + "num_tokens": 360476282.0, + "step": 14189 + }, + { + "epoch": 1.5583132000878541, + "grad_norm": 2.332617998123169, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.698421835899353, + "num_tokens": 360502673.0, + "step": 14190 + }, + { + "epoch": 1.5584230177904677, + "grad_norm": 2.3284947872161865, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7206835746765137, + "num_tokens": 360529206.0, + "step": 14191 + }, + { + "epoch": 1.5585328354930814, + "grad_norm": 2.1729419231414795, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7142208814620972, + "num_tokens": 360558321.0, + "step": 14192 + }, + { + "epoch": 1.5586426531956952, + "grad_norm": 2.331434726715088, + "learning_rate": 1e-06, + "loss": 0.8218, + "mean_token_accuracy": 0.7382648587226868, + "num_tokens": 360580669.0, + "step": 14193 + }, + { + "epoch": 1.558752470898309, + "grad_norm": 2.1302907466888428, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7325677275657654, + "num_tokens": 360605560.0, + "step": 14194 + }, + { + "epoch": 1.5588622886009225, + "grad_norm": 2.384382963180542, + "learning_rate": 1e-06, + "loss": 0.761, + "mean_token_accuracy": 0.7512640953063965, + "num_tokens": 360628847.0, + "step": 14195 + }, + { + "epoch": 1.558972106303536, + "grad_norm": 2.783674716949463, + "learning_rate": 1e-06, + "loss": 0.8497, + "mean_token_accuracy": 0.7313945293426514, + "num_tokens": 360648125.0, + "step": 14196 + }, + { + "epoch": 1.5590819240061498, + "grad_norm": 2.3472490310668945, + "learning_rate": 1e-06, + "loss": 0.9884, + "mean_token_accuracy": 0.711616039276123, + "num_tokens": 360673658.0, + "step": 14197 + }, + { + "epoch": 1.5591917417087635, + "grad_norm": 2.1926660537719727, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.7345736622810364, + "num_tokens": 360700425.0, + "step": 14198 + }, + { + "epoch": 1.5593015594113773, + "grad_norm": 2.178683280944824, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7318533658981323, + "num_tokens": 360725535.0, + "step": 14199 + }, + { + "epoch": 1.5594113771139908, + "grad_norm": 2.030219078063965, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7196623682975769, + "num_tokens": 360754342.0, + "step": 14200 + }, + { + "epoch": 1.5595211948166043, + "grad_norm": 2.2096667289733887, + "learning_rate": 1e-06, + "loss": 1.0, + "mean_token_accuracy": 0.6893920302391052, + "num_tokens": 360782825.0, + "step": 14201 + }, + { + "epoch": 1.559631012519218, + "grad_norm": 2.120333433151245, + "learning_rate": 1e-06, + "loss": 0.8883, + "mean_token_accuracy": 0.7217341661453247, + "num_tokens": 360809710.0, + "step": 14202 + }, + { + "epoch": 1.5597408302218319, + "grad_norm": 2.5139319896698, + "learning_rate": 1e-06, + "loss": 0.8173, + "mean_token_accuracy": 0.7501876354217529, + "num_tokens": 360831216.0, + "step": 14203 + }, + { + "epoch": 1.5598506479244454, + "grad_norm": 2.392235040664673, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.711322009563446, + "num_tokens": 360856180.0, + "step": 14204 + }, + { + "epoch": 1.559960465627059, + "grad_norm": 2.135991096496582, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7288863658905029, + "num_tokens": 360883481.0, + "step": 14205 + }, + { + "epoch": 1.5600702833296727, + "grad_norm": 2.35345458984375, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.698127806186676, + "num_tokens": 360908551.0, + "step": 14206 + }, + { + "epoch": 1.5601801010322864, + "grad_norm": 2.330251455307007, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7089439034461975, + "num_tokens": 360932336.0, + "step": 14207 + }, + { + "epoch": 1.5602899187349002, + "grad_norm": 2.270087480545044, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7368256449699402, + "num_tokens": 360956778.0, + "step": 14208 + }, + { + "epoch": 1.5603997364375137, + "grad_norm": 2.0272376537323, + "learning_rate": 1e-06, + "loss": 0.9616, + "mean_token_accuracy": 0.7051114439964294, + "num_tokens": 360987989.0, + "step": 14209 + }, + { + "epoch": 1.5605095541401273, + "grad_norm": 2.220721483230591, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7207211256027222, + "num_tokens": 361012108.0, + "step": 14210 + }, + { + "epoch": 1.560619371842741, + "grad_norm": 2.5114824771881104, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7306166887283325, + "num_tokens": 361031668.0, + "step": 14211 + }, + { + "epoch": 1.5607291895453548, + "grad_norm": 2.1085267066955566, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7260968685150146, + "num_tokens": 361059530.0, + "step": 14212 + }, + { + "epoch": 1.5608390072479685, + "grad_norm": 2.197669744491577, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7271506786346436, + "num_tokens": 361086389.0, + "step": 14213 + }, + { + "epoch": 1.560948824950582, + "grad_norm": 2.135859251022339, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7203421592712402, + "num_tokens": 361113066.0, + "step": 14214 + }, + { + "epoch": 1.5610586426531956, + "grad_norm": 2.060178756713867, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7162392139434814, + "num_tokens": 361143455.0, + "step": 14215 + }, + { + "epoch": 1.5611684603558094, + "grad_norm": 2.123600482940674, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.7032958269119263, + "num_tokens": 361173423.0, + "step": 14216 + }, + { + "epoch": 1.5612782780584231, + "grad_norm": 2.3928115367889404, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7194067239761353, + "num_tokens": 361195952.0, + "step": 14217 + }, + { + "epoch": 1.5613880957610367, + "grad_norm": 2.008165121078491, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7261568903923035, + "num_tokens": 361227918.0, + "step": 14218 + }, + { + "epoch": 1.5614979134636502, + "grad_norm": 2.498844861984253, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7355632781982422, + "num_tokens": 361249820.0, + "step": 14219 + }, + { + "epoch": 1.561607731166264, + "grad_norm": 2.643772840499878, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7252805233001709, + "num_tokens": 361269587.0, + "step": 14220 + }, + { + "epoch": 1.5617175488688777, + "grad_norm": 2.5571374893188477, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7313554286956787, + "num_tokens": 361291221.0, + "step": 14221 + }, + { + "epoch": 1.5618273665714915, + "grad_norm": 2.383391857147217, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7219986915588379, + "num_tokens": 361316431.0, + "step": 14222 + }, + { + "epoch": 1.561937184274105, + "grad_norm": 2.3839948177337646, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.7116042971611023, + "num_tokens": 361341215.0, + "step": 14223 + }, + { + "epoch": 1.5620470019767185, + "grad_norm": 2.17264723777771, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7316635251045227, + "num_tokens": 361368030.0, + "step": 14224 + }, + { + "epoch": 1.5621568196793323, + "grad_norm": 2.5052571296691895, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7118209600448608, + "num_tokens": 361388229.0, + "step": 14225 + }, + { + "epoch": 1.562266637381946, + "grad_norm": 2.2949681282043457, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.700381875038147, + "num_tokens": 361413669.0, + "step": 14226 + }, + { + "epoch": 1.5623764550845596, + "grad_norm": 2.226276159286499, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7249213457107544, + "num_tokens": 361441670.0, + "step": 14227 + }, + { + "epoch": 1.5624862727871733, + "grad_norm": 2.5510330200195312, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7226728796958923, + "num_tokens": 361461493.0, + "step": 14228 + }, + { + "epoch": 1.5625960904897869, + "grad_norm": 2.308211326599121, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.719714879989624, + "num_tokens": 361487292.0, + "step": 14229 + }, + { + "epoch": 1.5627059081924006, + "grad_norm": 2.2106359004974365, + "learning_rate": 1e-06, + "loss": 1.0076, + "mean_token_accuracy": 0.695124089717865, + "num_tokens": 361514982.0, + "step": 14230 + }, + { + "epoch": 1.5628157258950144, + "grad_norm": 2.314725160598755, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7217438220977783, + "num_tokens": 361540405.0, + "step": 14231 + }, + { + "epoch": 1.562925543597628, + "grad_norm": 2.3315932750701904, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7297777533531189, + "num_tokens": 361563437.0, + "step": 14232 + }, + { + "epoch": 1.5630353613002415, + "grad_norm": 2.257518768310547, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7232049703598022, + "num_tokens": 361587508.0, + "step": 14233 + }, + { + "epoch": 1.5631451790028552, + "grad_norm": 2.205341339111328, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7035831212997437, + "num_tokens": 361613584.0, + "step": 14234 + }, + { + "epoch": 1.563254996705469, + "grad_norm": 2.65122127532959, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7445262670516968, + "num_tokens": 361631961.0, + "step": 14235 + }, + { + "epoch": 1.5633648144080827, + "grad_norm": 2.0757362842559814, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7122164964675903, + "num_tokens": 361661501.0, + "step": 14236 + }, + { + "epoch": 1.5634746321106963, + "grad_norm": 2.143458604812622, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7120528221130371, + "num_tokens": 361686556.0, + "step": 14237 + }, + { + "epoch": 1.5635844498133098, + "grad_norm": 2.1421351432800293, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7245791554450989, + "num_tokens": 361715003.0, + "step": 14238 + }, + { + "epoch": 1.5636942675159236, + "grad_norm": 2.3089311122894287, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.701583981513977, + "num_tokens": 361742093.0, + "step": 14239 + }, + { + "epoch": 1.5638040852185373, + "grad_norm": 2.4265191555023193, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7356939315795898, + "num_tokens": 361762823.0, + "step": 14240 + }, + { + "epoch": 1.5639139029211508, + "grad_norm": 2.138579845428467, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.724994957447052, + "num_tokens": 361790556.0, + "step": 14241 + }, + { + "epoch": 1.5640237206237646, + "grad_norm": 2.0436885356903076, + "learning_rate": 1e-06, + "loss": 0.9673, + "mean_token_accuracy": 0.7012060880661011, + "num_tokens": 361821528.0, + "step": 14242 + }, + { + "epoch": 1.5641335383263781, + "grad_norm": 2.5491623878479004, + "learning_rate": 1e-06, + "loss": 0.888, + "mean_token_accuracy": 0.7257508039474487, + "num_tokens": 361843472.0, + "step": 14243 + }, + { + "epoch": 1.564243356028992, + "grad_norm": 2.0271222591400146, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7001762390136719, + "num_tokens": 361873786.0, + "step": 14244 + }, + { + "epoch": 1.5643531737316057, + "grad_norm": 2.0259897708892822, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7103430032730103, + "num_tokens": 361904144.0, + "step": 14245 + }, + { + "epoch": 1.5644629914342192, + "grad_norm": 2.0441648960113525, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7189623117446899, + "num_tokens": 361933047.0, + "step": 14246 + }, + { + "epoch": 1.5645728091368327, + "grad_norm": 2.1906938552856445, + "learning_rate": 1e-06, + "loss": 0.7975, + "mean_token_accuracy": 0.7429263591766357, + "num_tokens": 361958214.0, + "step": 14247 + }, + { + "epoch": 1.5646826268394465, + "grad_norm": 2.4048986434936523, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7113669514656067, + "num_tokens": 361982723.0, + "step": 14248 + }, + { + "epoch": 1.5647924445420602, + "grad_norm": 2.269054412841797, + "learning_rate": 1e-06, + "loss": 0.8675, + "mean_token_accuracy": 0.7287383079528809, + "num_tokens": 362008226.0, + "step": 14249 + }, + { + "epoch": 1.564902262244674, + "grad_norm": 2.13140869140625, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7166314721107483, + "num_tokens": 362034477.0, + "step": 14250 + }, + { + "epoch": 1.5650120799472875, + "grad_norm": 2.5854861736297607, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7252731323242188, + "num_tokens": 362054191.0, + "step": 14251 + }, + { + "epoch": 1.565121897649901, + "grad_norm": 2.379260778427124, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7087481617927551, + "num_tokens": 362077513.0, + "step": 14252 + }, + { + "epoch": 1.5652317153525148, + "grad_norm": 2.1291444301605225, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7216370105743408, + "num_tokens": 362105328.0, + "step": 14253 + }, + { + "epoch": 1.5653415330551286, + "grad_norm": 2.141622304916382, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7113457918167114, + "num_tokens": 362133172.0, + "step": 14254 + }, + { + "epoch": 1.565451350757742, + "grad_norm": 2.794506549835205, + "learning_rate": 1e-06, + "loss": 0.7962, + "mean_token_accuracy": 0.7457549571990967, + "num_tokens": 362150684.0, + "step": 14255 + }, + { + "epoch": 1.5655611684603556, + "grad_norm": 2.605431079864502, + "learning_rate": 1e-06, + "loss": 0.9507, + "mean_token_accuracy": 0.6987149119377136, + "num_tokens": 362172273.0, + "step": 14256 + }, + { + "epoch": 1.5656709861629694, + "grad_norm": 2.0136749744415283, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.717689037322998, + "num_tokens": 362206662.0, + "step": 14257 + }, + { + "epoch": 1.5657808038655832, + "grad_norm": 2.5050253868103027, + "learning_rate": 1e-06, + "loss": 0.8109, + "mean_token_accuracy": 0.7443522214889526, + "num_tokens": 362228907.0, + "step": 14258 + }, + { + "epoch": 1.565890621568197, + "grad_norm": 2.097074508666992, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7143728137016296, + "num_tokens": 362260464.0, + "step": 14259 + }, + { + "epoch": 1.5660004392708105, + "grad_norm": 2.2375407218933105, + "learning_rate": 1e-06, + "loss": 0.8296, + "mean_token_accuracy": 0.736365795135498, + "num_tokens": 362282781.0, + "step": 14260 + }, + { + "epoch": 1.566110256973424, + "grad_norm": 2.1296536922454834, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7099628448486328, + "num_tokens": 362312514.0, + "step": 14261 + }, + { + "epoch": 1.5662200746760377, + "grad_norm": 2.099079132080078, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7044456005096436, + "num_tokens": 362342004.0, + "step": 14262 + }, + { + "epoch": 1.5663298923786515, + "grad_norm": 2.3204174041748047, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7091057896614075, + "num_tokens": 362366278.0, + "step": 14263 + }, + { + "epoch": 1.5664397100812653, + "grad_norm": 2.2209346294403076, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7103740572929382, + "num_tokens": 362391692.0, + "step": 14264 + }, + { + "epoch": 1.5665495277838788, + "grad_norm": 2.16157865524292, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7130683660507202, + "num_tokens": 362419873.0, + "step": 14265 + }, + { + "epoch": 1.5666593454864923, + "grad_norm": 2.147308349609375, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.710207462310791, + "num_tokens": 362448076.0, + "step": 14266 + }, + { + "epoch": 1.566769163189106, + "grad_norm": 1.9099668264389038, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.6996051073074341, + "num_tokens": 362482275.0, + "step": 14267 + }, + { + "epoch": 1.5668789808917198, + "grad_norm": 2.2172162532806396, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7212003469467163, + "num_tokens": 362508345.0, + "step": 14268 + }, + { + "epoch": 1.5669887985943334, + "grad_norm": 2.2507505416870117, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7081208229064941, + "num_tokens": 362533079.0, + "step": 14269 + }, + { + "epoch": 1.567098616296947, + "grad_norm": 2.2698781490325928, + "learning_rate": 1e-06, + "loss": 0.8505, + "mean_token_accuracy": 0.729638934135437, + "num_tokens": 362558190.0, + "step": 14270 + }, + { + "epoch": 1.5672084339995607, + "grad_norm": 2.0438873767852783, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7250336408615112, + "num_tokens": 362589645.0, + "step": 14271 + }, + { + "epoch": 1.5673182517021744, + "grad_norm": 2.2039592266082764, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7162392139434814, + "num_tokens": 362614183.0, + "step": 14272 + }, + { + "epoch": 1.5674280694047882, + "grad_norm": 2.3052594661712646, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7283598184585571, + "num_tokens": 362639481.0, + "step": 14273 + }, + { + "epoch": 1.5675378871074017, + "grad_norm": 2.3258185386657715, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7154353857040405, + "num_tokens": 362664336.0, + "step": 14274 + }, + { + "epoch": 1.5676477048100153, + "grad_norm": 2.3799407482147217, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7144652605056763, + "num_tokens": 362689520.0, + "step": 14275 + }, + { + "epoch": 1.567757522512629, + "grad_norm": 2.43304705619812, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7352317571640015, + "num_tokens": 362710693.0, + "step": 14276 + }, + { + "epoch": 1.5678673402152428, + "grad_norm": 2.25288724899292, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7347727417945862, + "num_tokens": 362735966.0, + "step": 14277 + }, + { + "epoch": 1.5679771579178565, + "grad_norm": 2.0270819664001465, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7131079435348511, + "num_tokens": 362766496.0, + "step": 14278 + }, + { + "epoch": 1.56808697562047, + "grad_norm": 2.422084331512451, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7204614281654358, + "num_tokens": 362790029.0, + "step": 14279 + }, + { + "epoch": 1.5681967933230836, + "grad_norm": 2.1563639640808105, + "learning_rate": 1e-06, + "loss": 1.0488, + "mean_token_accuracy": 0.6856481432914734, + "num_tokens": 362820351.0, + "step": 14280 + }, + { + "epoch": 1.5683066110256974, + "grad_norm": 2.1448984146118164, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7241971492767334, + "num_tokens": 362847769.0, + "step": 14281 + }, + { + "epoch": 1.568416428728311, + "grad_norm": 2.3315420150756836, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7104223966598511, + "num_tokens": 362873087.0, + "step": 14282 + }, + { + "epoch": 1.5685262464309246, + "grad_norm": 1.919510841369629, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.718339204788208, + "num_tokens": 362905180.0, + "step": 14283 + }, + { + "epoch": 1.5686360641335382, + "grad_norm": 2.0531890392303467, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7099391222000122, + "num_tokens": 362933505.0, + "step": 14284 + }, + { + "epoch": 1.568745881836152, + "grad_norm": 2.36112642288208, + "learning_rate": 1e-06, + "loss": 0.8722, + "mean_token_accuracy": 0.7311782836914062, + "num_tokens": 362956389.0, + "step": 14285 + }, + { + "epoch": 1.5688556995387657, + "grad_norm": 2.263153314590454, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7251627445220947, + "num_tokens": 362979280.0, + "step": 14286 + }, + { + "epoch": 1.5689655172413794, + "grad_norm": 2.273104429244995, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.7539452314376831, + "num_tokens": 363001911.0, + "step": 14287 + }, + { + "epoch": 1.569075334943993, + "grad_norm": 2.2380921840667725, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7271561026573181, + "num_tokens": 363026308.0, + "step": 14288 + }, + { + "epoch": 1.5691851526466065, + "grad_norm": 2.091902732849121, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7105482816696167, + "num_tokens": 363054122.0, + "step": 14289 + }, + { + "epoch": 1.5692949703492203, + "grad_norm": 2.2189133167266846, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7326191067695618, + "num_tokens": 363078673.0, + "step": 14290 + }, + { + "epoch": 1.569404788051834, + "grad_norm": 1.8779935836791992, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7145683765411377, + "num_tokens": 363113257.0, + "step": 14291 + }, + { + "epoch": 1.5695146057544476, + "grad_norm": 2.1976144313812256, + "learning_rate": 1e-06, + "loss": 0.8609, + "mean_token_accuracy": 0.7299149632453918, + "num_tokens": 363138630.0, + "step": 14292 + }, + { + "epoch": 1.5696244234570613, + "grad_norm": 2.395820379257202, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7367599010467529, + "num_tokens": 363161685.0, + "step": 14293 + }, + { + "epoch": 1.5697342411596749, + "grad_norm": 2.054527521133423, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7178549766540527, + "num_tokens": 363191051.0, + "step": 14294 + }, + { + "epoch": 1.5698440588622886, + "grad_norm": 2.273357629776001, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7234783172607422, + "num_tokens": 363215889.0, + "step": 14295 + }, + { + "epoch": 1.5699538765649024, + "grad_norm": 2.3083720207214355, + "learning_rate": 1e-06, + "loss": 0.8187, + "mean_token_accuracy": 0.7436708211898804, + "num_tokens": 363238714.0, + "step": 14296 + }, + { + "epoch": 1.570063694267516, + "grad_norm": 2.0571889877319336, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7010947465896606, + "num_tokens": 363269894.0, + "step": 14297 + }, + { + "epoch": 1.5701735119701294, + "grad_norm": 2.31457257270813, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.7332032918930054, + "num_tokens": 363292828.0, + "step": 14298 + }, + { + "epoch": 1.5702833296727432, + "grad_norm": 2.154921770095825, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7082282304763794, + "num_tokens": 363321780.0, + "step": 14299 + }, + { + "epoch": 1.570393147375357, + "grad_norm": 2.216132402420044, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.7043373584747314, + "num_tokens": 363350436.0, + "step": 14300 + }, + { + "epoch": 1.5705029650779707, + "grad_norm": 2.2095162868499756, + "learning_rate": 1e-06, + "loss": 1.0217, + "mean_token_accuracy": 0.69279944896698, + "num_tokens": 363377506.0, + "step": 14301 + }, + { + "epoch": 1.5706127827805842, + "grad_norm": 2.4246349334716797, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.7101570963859558, + "num_tokens": 363400817.0, + "step": 14302 + }, + { + "epoch": 1.5707226004831978, + "grad_norm": 2.383608818054199, + "learning_rate": 1e-06, + "loss": 0.7977, + "mean_token_accuracy": 0.7426961064338684, + "num_tokens": 363422966.0, + "step": 14303 + }, + { + "epoch": 1.5708324181858115, + "grad_norm": 2.021862268447876, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7095227241516113, + "num_tokens": 363452990.0, + "step": 14304 + }, + { + "epoch": 1.5709422358884253, + "grad_norm": 2.254406452178955, + "learning_rate": 1e-06, + "loss": 0.8061, + "mean_token_accuracy": 0.7421460151672363, + "num_tokens": 363475893.0, + "step": 14305 + }, + { + "epoch": 1.5710520535910388, + "grad_norm": 2.112740993499756, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7065775990486145, + "num_tokens": 363505515.0, + "step": 14306 + }, + { + "epoch": 1.5711618712936526, + "grad_norm": 2.2094106674194336, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7009505033493042, + "num_tokens": 363531365.0, + "step": 14307 + }, + { + "epoch": 1.5712716889962661, + "grad_norm": 2.1129989624023438, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.744620680809021, + "num_tokens": 363555807.0, + "step": 14308 + }, + { + "epoch": 1.5713815066988799, + "grad_norm": 2.2965071201324463, + "learning_rate": 1e-06, + "loss": 0.8054, + "mean_token_accuracy": 0.7417055368423462, + "num_tokens": 363578616.0, + "step": 14309 + }, + { + "epoch": 1.5714913244014936, + "grad_norm": 2.191952705383301, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.731577455997467, + "num_tokens": 363605123.0, + "step": 14310 + }, + { + "epoch": 1.5716011421041072, + "grad_norm": 1.985663890838623, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7090932130813599, + "num_tokens": 363636176.0, + "step": 14311 + }, + { + "epoch": 1.5717109598067207, + "grad_norm": 2.1499545574188232, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7146049737930298, + "num_tokens": 363663091.0, + "step": 14312 + }, + { + "epoch": 1.5718207775093345, + "grad_norm": 2.3557136058807373, + "learning_rate": 1e-06, + "loss": 0.8794, + "mean_token_accuracy": 0.72004234790802, + "num_tokens": 363685505.0, + "step": 14313 + }, + { + "epoch": 1.5719305952119482, + "grad_norm": 2.488429069519043, + "learning_rate": 1e-06, + "loss": 0.8098, + "mean_token_accuracy": 0.7440756559371948, + "num_tokens": 363706968.0, + "step": 14314 + }, + { + "epoch": 1.572040412914562, + "grad_norm": 2.3264238834381104, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7448632717132568, + "num_tokens": 363730133.0, + "step": 14315 + }, + { + "epoch": 1.5721502306171755, + "grad_norm": 2.0898170471191406, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7169109582901001, + "num_tokens": 363760835.0, + "step": 14316 + }, + { + "epoch": 1.572260048319789, + "grad_norm": 2.2689945697784424, + "learning_rate": 1e-06, + "loss": 1.0192, + "mean_token_accuracy": 0.6954496502876282, + "num_tokens": 363786696.0, + "step": 14317 + }, + { + "epoch": 1.5723698660224028, + "grad_norm": 2.2030153274536133, + "learning_rate": 1e-06, + "loss": 0.9681, + "mean_token_accuracy": 0.7011455297470093, + "num_tokens": 363813080.0, + "step": 14318 + }, + { + "epoch": 1.5724796837250166, + "grad_norm": 2.2098731994628906, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7061365842819214, + "num_tokens": 363840692.0, + "step": 14319 + }, + { + "epoch": 1.57258950142763, + "grad_norm": 2.241781711578369, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7220212817192078, + "num_tokens": 363866807.0, + "step": 14320 + }, + { + "epoch": 1.5726993191302436, + "grad_norm": 2.0621602535247803, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7018442153930664, + "num_tokens": 363897811.0, + "step": 14321 + }, + { + "epoch": 1.5728091368328574, + "grad_norm": 2.5344858169555664, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7177976965904236, + "num_tokens": 363918729.0, + "step": 14322 + }, + { + "epoch": 1.5729189545354711, + "grad_norm": 2.0815272331237793, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7160364389419556, + "num_tokens": 363947471.0, + "step": 14323 + }, + { + "epoch": 1.573028772238085, + "grad_norm": 2.1948158740997314, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.7148021459579468, + "num_tokens": 363974206.0, + "step": 14324 + }, + { + "epoch": 1.5731385899406984, + "grad_norm": 2.4989829063415527, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7179760932922363, + "num_tokens": 363996000.0, + "step": 14325 + }, + { + "epoch": 1.573248407643312, + "grad_norm": 2.1749536991119385, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7217142581939697, + "num_tokens": 364021584.0, + "step": 14326 + }, + { + "epoch": 1.5733582253459257, + "grad_norm": 1.9421643018722534, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7022618055343628, + "num_tokens": 364053931.0, + "step": 14327 + }, + { + "epoch": 1.5734680430485395, + "grad_norm": 2.388105630874634, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7379955649375916, + "num_tokens": 364076206.0, + "step": 14328 + }, + { + "epoch": 1.5735778607511532, + "grad_norm": 2.3082258701324463, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.6982421278953552, + "num_tokens": 364100614.0, + "step": 14329 + }, + { + "epoch": 1.5736876784537668, + "grad_norm": 2.1898324489593506, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7037125825881958, + "num_tokens": 364126764.0, + "step": 14330 + }, + { + "epoch": 1.5737974961563803, + "grad_norm": 2.532803535461426, + "learning_rate": 1e-06, + "loss": 0.8083, + "mean_token_accuracy": 0.7427127957344055, + "num_tokens": 364146148.0, + "step": 14331 + }, + { + "epoch": 1.573907313858994, + "grad_norm": 2.152050018310547, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7201359272003174, + "num_tokens": 364171176.0, + "step": 14332 + }, + { + "epoch": 1.5740171315616078, + "grad_norm": 2.281904458999634, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7099511623382568, + "num_tokens": 364196507.0, + "step": 14333 + }, + { + "epoch": 1.5741269492642214, + "grad_norm": 2.243239164352417, + "learning_rate": 1e-06, + "loss": 1.0183, + "mean_token_accuracy": 0.6865977048873901, + "num_tokens": 364223939.0, + "step": 14334 + }, + { + "epoch": 1.574236766966835, + "grad_norm": 2.1599748134613037, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.6988783478736877, + "num_tokens": 364250116.0, + "step": 14335 + }, + { + "epoch": 1.5743465846694487, + "grad_norm": 2.2198500633239746, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7205002307891846, + "num_tokens": 364275796.0, + "step": 14336 + }, + { + "epoch": 1.5744564023720624, + "grad_norm": 2.3174901008605957, + "learning_rate": 1e-06, + "loss": 0.9691, + "mean_token_accuracy": 0.701208233833313, + "num_tokens": 364299744.0, + "step": 14337 + }, + { + "epoch": 1.5745662200746762, + "grad_norm": 2.6128594875335693, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7255216836929321, + "num_tokens": 364319233.0, + "step": 14338 + }, + { + "epoch": 1.5746760377772897, + "grad_norm": 2.2171342372894287, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7234278917312622, + "num_tokens": 364343817.0, + "step": 14339 + }, + { + "epoch": 1.5747858554799032, + "grad_norm": 2.14290714263916, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7065761685371399, + "num_tokens": 364371110.0, + "step": 14340 + }, + { + "epoch": 1.574895673182517, + "grad_norm": 2.148730754852295, + "learning_rate": 1e-06, + "loss": 0.8374, + "mean_token_accuracy": 0.7313417196273804, + "num_tokens": 364398457.0, + "step": 14341 + }, + { + "epoch": 1.5750054908851308, + "grad_norm": 2.4776461124420166, + "learning_rate": 1e-06, + "loss": 0.8578, + "mean_token_accuracy": 0.7375028133392334, + "num_tokens": 364421266.0, + "step": 14342 + }, + { + "epoch": 1.5751153085877443, + "grad_norm": 2.5463814735412598, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7204113006591797, + "num_tokens": 364442124.0, + "step": 14343 + }, + { + "epoch": 1.575225126290358, + "grad_norm": 1.9753409624099731, + "learning_rate": 1e-06, + "loss": 0.9985, + "mean_token_accuracy": 0.6950569152832031, + "num_tokens": 364476244.0, + "step": 14344 + }, + { + "epoch": 1.5753349439929716, + "grad_norm": 2.1163008213043213, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7356541156768799, + "num_tokens": 364503611.0, + "step": 14345 + }, + { + "epoch": 1.5754447616955853, + "grad_norm": 2.266801357269287, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7131984233856201, + "num_tokens": 364528144.0, + "step": 14346 + }, + { + "epoch": 1.575554579398199, + "grad_norm": 2.202023506164551, + "learning_rate": 1e-06, + "loss": 0.993, + "mean_token_accuracy": 0.7014878988265991, + "num_tokens": 364555017.0, + "step": 14347 + }, + { + "epoch": 1.5756643971008126, + "grad_norm": 2.414313793182373, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7300547361373901, + "num_tokens": 364577329.0, + "step": 14348 + }, + { + "epoch": 1.5757742148034262, + "grad_norm": 2.073936700820923, + "learning_rate": 1e-06, + "loss": 1.0334, + "mean_token_accuracy": 0.6998363137245178, + "num_tokens": 364611840.0, + "step": 14349 + }, + { + "epoch": 1.57588403250604, + "grad_norm": 2.423975944519043, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7267102003097534, + "num_tokens": 364633903.0, + "step": 14350 + }, + { + "epoch": 1.5759938502086537, + "grad_norm": 2.220311403274536, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7389492988586426, + "num_tokens": 364658169.0, + "step": 14351 + }, + { + "epoch": 1.5761036679112674, + "grad_norm": 2.218182325363159, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7027336359024048, + "num_tokens": 364685606.0, + "step": 14352 + }, + { + "epoch": 1.576213485613881, + "grad_norm": 2.4518604278564453, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7220385670661926, + "num_tokens": 364708139.0, + "step": 14353 + }, + { + "epoch": 1.5763233033164945, + "grad_norm": 2.255913019180298, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7174416184425354, + "num_tokens": 364732604.0, + "step": 14354 + }, + { + "epoch": 1.5764331210191083, + "grad_norm": 2.437283515930176, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7412793636322021, + "num_tokens": 364754905.0, + "step": 14355 + }, + { + "epoch": 1.576542938721722, + "grad_norm": 2.1776700019836426, + "learning_rate": 1e-06, + "loss": 1.0044, + "mean_token_accuracy": 0.6908145546913147, + "num_tokens": 364784551.0, + "step": 14356 + }, + { + "epoch": 1.5766527564243356, + "grad_norm": 2.300819158554077, + "learning_rate": 1e-06, + "loss": 0.8284, + "mean_token_accuracy": 0.7434552907943726, + "num_tokens": 364807437.0, + "step": 14357 + }, + { + "epoch": 1.5767625741269493, + "grad_norm": 2.2063040733337402, + "learning_rate": 1e-06, + "loss": 0.842, + "mean_token_accuracy": 0.7384710311889648, + "num_tokens": 364833582.0, + "step": 14358 + }, + { + "epoch": 1.5768723918295628, + "grad_norm": 1.979742169380188, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7242177724838257, + "num_tokens": 364864674.0, + "step": 14359 + }, + { + "epoch": 1.5769822095321766, + "grad_norm": 2.2543792724609375, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7019519805908203, + "num_tokens": 364890190.0, + "step": 14360 + }, + { + "epoch": 1.5770920272347904, + "grad_norm": 2.2272093296051025, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7115788459777832, + "num_tokens": 364917077.0, + "step": 14361 + }, + { + "epoch": 1.577201844937404, + "grad_norm": 2.282383680343628, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7194855213165283, + "num_tokens": 364942415.0, + "step": 14362 + }, + { + "epoch": 1.5773116626400174, + "grad_norm": 1.9976048469543457, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7021001577377319, + "num_tokens": 364974317.0, + "step": 14363 + }, + { + "epoch": 1.5774214803426312, + "grad_norm": 2.3343567848205566, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7373344898223877, + "num_tokens": 364999864.0, + "step": 14364 + }, + { + "epoch": 1.577531298045245, + "grad_norm": 2.0819711685180664, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6984530687332153, + "num_tokens": 365028502.0, + "step": 14365 + }, + { + "epoch": 1.5776411157478587, + "grad_norm": 1.9290790557861328, + "learning_rate": 1e-06, + "loss": 0.983, + "mean_token_accuracy": 0.7006053924560547, + "num_tokens": 365061501.0, + "step": 14366 + }, + { + "epoch": 1.5777509334504722, + "grad_norm": 2.246397018432617, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7133135795593262, + "num_tokens": 365088530.0, + "step": 14367 + }, + { + "epoch": 1.5778607511530858, + "grad_norm": 2.4192795753479004, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7299100160598755, + "num_tokens": 365109473.0, + "step": 14368 + }, + { + "epoch": 1.5779705688556995, + "grad_norm": 1.9938926696777344, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7161872386932373, + "num_tokens": 365140569.0, + "step": 14369 + }, + { + "epoch": 1.5780803865583133, + "grad_norm": 2.3212592601776123, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7085281014442444, + "num_tokens": 365165113.0, + "step": 14370 + }, + { + "epoch": 1.5781902042609268, + "grad_norm": 2.188206911087036, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7094221115112305, + "num_tokens": 365192322.0, + "step": 14371 + }, + { + "epoch": 1.5783000219635406, + "grad_norm": 2.249579429626465, + "learning_rate": 1e-06, + "loss": 0.8551, + "mean_token_accuracy": 0.7355080842971802, + "num_tokens": 365216560.0, + "step": 14372 + }, + { + "epoch": 1.5784098396661541, + "grad_norm": 2.3380656242370605, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7186383008956909, + "num_tokens": 365239432.0, + "step": 14373 + }, + { + "epoch": 1.5785196573687679, + "grad_norm": 2.176877021789551, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7078762650489807, + "num_tokens": 365265700.0, + "step": 14374 + }, + { + "epoch": 1.5786294750713816, + "grad_norm": 2.5619585514068604, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.7451808452606201, + "num_tokens": 365285237.0, + "step": 14375 + }, + { + "epoch": 1.5787392927739952, + "grad_norm": 2.1735241413116455, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7440176606178284, + "num_tokens": 365311435.0, + "step": 14376 + }, + { + "epoch": 1.5788491104766087, + "grad_norm": 2.35880446434021, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.7062270641326904, + "num_tokens": 365334043.0, + "step": 14377 + }, + { + "epoch": 1.5789589281792225, + "grad_norm": 2.1938486099243164, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.702501654624939, + "num_tokens": 365360292.0, + "step": 14378 + }, + { + "epoch": 1.5790687458818362, + "grad_norm": 2.125718832015991, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7120110988616943, + "num_tokens": 365388442.0, + "step": 14379 + }, + { + "epoch": 1.57917856358445, + "grad_norm": 2.282594919204712, + "learning_rate": 1e-06, + "loss": 0.7806, + "mean_token_accuracy": 0.7480692267417908, + "num_tokens": 365410569.0, + "step": 14380 + }, + { + "epoch": 1.5792883812870635, + "grad_norm": 2.1242549419403076, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7143275737762451, + "num_tokens": 365437757.0, + "step": 14381 + }, + { + "epoch": 1.579398198989677, + "grad_norm": 2.3655800819396973, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7258358001708984, + "num_tokens": 365459425.0, + "step": 14382 + }, + { + "epoch": 1.5795080166922908, + "grad_norm": 2.2637407779693604, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7126436233520508, + "num_tokens": 365484484.0, + "step": 14383 + }, + { + "epoch": 1.5796178343949046, + "grad_norm": 2.085941791534424, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7247638702392578, + "num_tokens": 365511607.0, + "step": 14384 + }, + { + "epoch": 1.579727652097518, + "grad_norm": 2.191941738128662, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7213984727859497, + "num_tokens": 365537286.0, + "step": 14385 + }, + { + "epoch": 1.5798374698001316, + "grad_norm": 2.411699056625366, + "learning_rate": 1e-06, + "loss": 1.0352, + "mean_token_accuracy": 0.6813074350357056, + "num_tokens": 365562989.0, + "step": 14386 + }, + { + "epoch": 1.5799472875027454, + "grad_norm": 2.248633861541748, + "learning_rate": 1e-06, + "loss": 0.9228, + "mean_token_accuracy": 0.7138012647628784, + "num_tokens": 365588089.0, + "step": 14387 + }, + { + "epoch": 1.5800571052053591, + "grad_norm": 2.5186026096343994, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7303780317306519, + "num_tokens": 365608524.0, + "step": 14388 + }, + { + "epoch": 1.580166922907973, + "grad_norm": 2.138056755065918, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.7288510799407959, + "num_tokens": 365633087.0, + "step": 14389 + }, + { + "epoch": 1.5802767406105864, + "grad_norm": 2.177372932434082, + "learning_rate": 1e-06, + "loss": 0.965, + "mean_token_accuracy": 0.7027741074562073, + "num_tokens": 365660987.0, + "step": 14390 + }, + { + "epoch": 1.5803865583132, + "grad_norm": 2.0844335556030273, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7070765495300293, + "num_tokens": 365690556.0, + "step": 14391 + }, + { + "epoch": 1.5804963760158137, + "grad_norm": 2.1778621673583984, + "learning_rate": 1e-06, + "loss": 1.0071, + "mean_token_accuracy": 0.684721827507019, + "num_tokens": 365720207.0, + "step": 14392 + }, + { + "epoch": 1.5806061937184275, + "grad_norm": 2.1273927688598633, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7404956817626953, + "num_tokens": 365746871.0, + "step": 14393 + }, + { + "epoch": 1.5807160114210412, + "grad_norm": 2.6788535118103027, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7203727960586548, + "num_tokens": 365767316.0, + "step": 14394 + }, + { + "epoch": 1.5808258291236548, + "grad_norm": 2.7774760723114014, + "learning_rate": 1e-06, + "loss": 0.8103, + "mean_token_accuracy": 0.7375447750091553, + "num_tokens": 365785973.0, + "step": 14395 + }, + { + "epoch": 1.5809356468262683, + "grad_norm": 2.3304803371429443, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7145683765411377, + "num_tokens": 365809991.0, + "step": 14396 + }, + { + "epoch": 1.581045464528882, + "grad_norm": 2.449512243270874, + "learning_rate": 1e-06, + "loss": 0.9772, + "mean_token_accuracy": 0.7162346243858337, + "num_tokens": 365833995.0, + "step": 14397 + }, + { + "epoch": 1.5811552822314958, + "grad_norm": 2.303548812866211, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7227010726928711, + "num_tokens": 365859476.0, + "step": 14398 + }, + { + "epoch": 1.5812650999341094, + "grad_norm": 2.165874719619751, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7107944488525391, + "num_tokens": 365884591.0, + "step": 14399 + }, + { + "epoch": 1.5813749176367229, + "grad_norm": 2.406149387359619, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.725030243396759, + "num_tokens": 365905403.0, + "step": 14400 + }, + { + "epoch": 1.5814847353393366, + "grad_norm": 2.2012434005737305, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7342310547828674, + "num_tokens": 365930020.0, + "step": 14401 + }, + { + "epoch": 1.5815945530419504, + "grad_norm": 2.3305728435516357, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.7259259819984436, + "num_tokens": 365955414.0, + "step": 14402 + }, + { + "epoch": 1.5817043707445642, + "grad_norm": 2.6324431896209717, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7171541452407837, + "num_tokens": 365975828.0, + "step": 14403 + }, + { + "epoch": 1.5818141884471777, + "grad_norm": 2.046022891998291, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7235480546951294, + "num_tokens": 366004750.0, + "step": 14404 + }, + { + "epoch": 1.5819240061497912, + "grad_norm": 2.086132287979126, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6979151964187622, + "num_tokens": 366034185.0, + "step": 14405 + }, + { + "epoch": 1.582033823852405, + "grad_norm": 2.1813812255859375, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7262137532234192, + "num_tokens": 366059171.0, + "step": 14406 + }, + { + "epoch": 1.5821436415550187, + "grad_norm": 2.0340118408203125, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7241583466529846, + "num_tokens": 366087944.0, + "step": 14407 + }, + { + "epoch": 1.5822534592576323, + "grad_norm": 2.0770246982574463, + "learning_rate": 1e-06, + "loss": 0.7944, + "mean_token_accuracy": 0.747620165348053, + "num_tokens": 366115659.0, + "step": 14408 + }, + { + "epoch": 1.582363276960246, + "grad_norm": 2.0968124866485596, + "learning_rate": 1e-06, + "loss": 0.8184, + "mean_token_accuracy": 0.7396163940429688, + "num_tokens": 366144791.0, + "step": 14409 + }, + { + "epoch": 1.5824730946628596, + "grad_norm": 2.094949722290039, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7034969329833984, + "num_tokens": 366176187.0, + "step": 14410 + }, + { + "epoch": 1.5825829123654733, + "grad_norm": 2.2451632022857666, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7177845239639282, + "num_tokens": 366201250.0, + "step": 14411 + }, + { + "epoch": 1.582692730068087, + "grad_norm": 2.1637206077575684, + "learning_rate": 1e-06, + "loss": 0.7983, + "mean_token_accuracy": 0.7515910863876343, + "num_tokens": 366227342.0, + "step": 14412 + }, + { + "epoch": 1.5828025477707006, + "grad_norm": 2.329084873199463, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7224560379981995, + "num_tokens": 366250504.0, + "step": 14413 + }, + { + "epoch": 1.5829123654733142, + "grad_norm": 2.041048526763916, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7209875583648682, + "num_tokens": 366279083.0, + "step": 14414 + }, + { + "epoch": 1.583022183175928, + "grad_norm": 2.1647934913635254, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7119088172912598, + "num_tokens": 366307027.0, + "step": 14415 + }, + { + "epoch": 1.5831320008785417, + "grad_norm": 2.529407501220703, + "learning_rate": 1e-06, + "loss": 0.8853, + "mean_token_accuracy": 0.7223114967346191, + "num_tokens": 366328712.0, + "step": 14416 + }, + { + "epoch": 1.5832418185811554, + "grad_norm": 2.2974109649658203, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7182115316390991, + "num_tokens": 366353472.0, + "step": 14417 + }, + { + "epoch": 1.583351636283769, + "grad_norm": 2.1525399684906006, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.713450014591217, + "num_tokens": 366380554.0, + "step": 14418 + }, + { + "epoch": 1.5834614539863825, + "grad_norm": 2.6307036876678467, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7130203247070312, + "num_tokens": 366401694.0, + "step": 14419 + }, + { + "epoch": 1.5835712716889963, + "grad_norm": 2.190739870071411, + "learning_rate": 1e-06, + "loss": 0.9739, + "mean_token_accuracy": 0.7019134759902954, + "num_tokens": 366431064.0, + "step": 14420 + }, + { + "epoch": 1.58368108939161, + "grad_norm": 2.633366346359253, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7459482550621033, + "num_tokens": 366449871.0, + "step": 14421 + }, + { + "epoch": 1.5837909070942235, + "grad_norm": 2.349825382232666, + "learning_rate": 1e-06, + "loss": 0.8137, + "mean_token_accuracy": 0.7404914498329163, + "num_tokens": 366472308.0, + "step": 14422 + }, + { + "epoch": 1.5839007247968373, + "grad_norm": 2.582719326019287, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7203913331031799, + "num_tokens": 366493479.0, + "step": 14423 + }, + { + "epoch": 1.5840105424994508, + "grad_norm": 2.218045711517334, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7195279598236084, + "num_tokens": 366517889.0, + "step": 14424 + }, + { + "epoch": 1.5841203602020646, + "grad_norm": 2.346670627593994, + "learning_rate": 1e-06, + "loss": 0.9087, + "mean_token_accuracy": 0.7209254503250122, + "num_tokens": 366540049.0, + "step": 14425 + }, + { + "epoch": 1.5842301779046783, + "grad_norm": 2.222836494445801, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7202272415161133, + "num_tokens": 366565391.0, + "step": 14426 + }, + { + "epoch": 1.5843399956072919, + "grad_norm": 2.0540995597839355, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7136640548706055, + "num_tokens": 366596207.0, + "step": 14427 + }, + { + "epoch": 1.5844498133099054, + "grad_norm": 2.6017918586730957, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7189359664916992, + "num_tokens": 366615803.0, + "step": 14428 + }, + { + "epoch": 1.5845596310125192, + "grad_norm": 1.868054986000061, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7035517692565918, + "num_tokens": 366649558.0, + "step": 14429 + }, + { + "epoch": 1.584669448715133, + "grad_norm": 2.307875394821167, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7235966920852661, + "num_tokens": 366673107.0, + "step": 14430 + }, + { + "epoch": 1.5847792664177467, + "grad_norm": 2.5285675525665283, + "learning_rate": 1e-06, + "loss": 0.9973, + "mean_token_accuracy": 0.6886283755302429, + "num_tokens": 366695093.0, + "step": 14431 + }, + { + "epoch": 1.5848890841203602, + "grad_norm": 2.234879970550537, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.6967378854751587, + "num_tokens": 366724444.0, + "step": 14432 + }, + { + "epoch": 1.5849989018229738, + "grad_norm": 2.3319098949432373, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7149478793144226, + "num_tokens": 366748780.0, + "step": 14433 + }, + { + "epoch": 1.5851087195255875, + "grad_norm": 2.6246390342712402, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.7487923502922058, + "num_tokens": 366768132.0, + "step": 14434 + }, + { + "epoch": 1.5852185372282013, + "grad_norm": 2.1992204189300537, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7353382706642151, + "num_tokens": 366793100.0, + "step": 14435 + }, + { + "epoch": 1.5853283549308148, + "grad_norm": 2.410860776901245, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7237389087677002, + "num_tokens": 366815816.0, + "step": 14436 + }, + { + "epoch": 1.5854381726334283, + "grad_norm": 2.5537049770355225, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7326714992523193, + "num_tokens": 366838850.0, + "step": 14437 + }, + { + "epoch": 1.585547990336042, + "grad_norm": 2.2145867347717285, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.726965069770813, + "num_tokens": 366867060.0, + "step": 14438 + }, + { + "epoch": 1.5856578080386559, + "grad_norm": 2.1517128944396973, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7367662787437439, + "num_tokens": 366893890.0, + "step": 14439 + }, + { + "epoch": 1.5857676257412696, + "grad_norm": 2.3520114421844482, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7117671966552734, + "num_tokens": 366919584.0, + "step": 14440 + }, + { + "epoch": 1.5858774434438832, + "grad_norm": 2.379345417022705, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7202496528625488, + "num_tokens": 366943624.0, + "step": 14441 + }, + { + "epoch": 1.5859872611464967, + "grad_norm": 2.1321911811828613, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7159196138381958, + "num_tokens": 366971122.0, + "step": 14442 + }, + { + "epoch": 1.5860970788491104, + "grad_norm": 2.198396682739258, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7184197902679443, + "num_tokens": 366997289.0, + "step": 14443 + }, + { + "epoch": 1.5862068965517242, + "grad_norm": 2.389667272567749, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7098058462142944, + "num_tokens": 367019912.0, + "step": 14444 + }, + { + "epoch": 1.586316714254338, + "grad_norm": 2.138432025909424, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7219324707984924, + "num_tokens": 367046385.0, + "step": 14445 + }, + { + "epoch": 1.5864265319569515, + "grad_norm": 2.067370891571045, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7230252027511597, + "num_tokens": 367076298.0, + "step": 14446 + }, + { + "epoch": 1.586536349659565, + "grad_norm": 2.386110305786133, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7075995206832886, + "num_tokens": 367100520.0, + "step": 14447 + }, + { + "epoch": 1.5866461673621788, + "grad_norm": 2.492879629135132, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7159909009933472, + "num_tokens": 367122021.0, + "step": 14448 + }, + { + "epoch": 1.5867559850647925, + "grad_norm": 2.319471836090088, + "learning_rate": 1e-06, + "loss": 0.8927, + "mean_token_accuracy": 0.7242425084114075, + "num_tokens": 367146071.0, + "step": 14449 + }, + { + "epoch": 1.586865802767406, + "grad_norm": 2.3641648292541504, + "learning_rate": 1e-06, + "loss": 0.8451, + "mean_token_accuracy": 0.7466843128204346, + "num_tokens": 367167942.0, + "step": 14450 + }, + { + "epoch": 1.5869756204700196, + "grad_norm": 2.367086410522461, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7421544790267944, + "num_tokens": 367190777.0, + "step": 14451 + }, + { + "epoch": 1.5870854381726334, + "grad_norm": 2.496778726577759, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7279372215270996, + "num_tokens": 367211275.0, + "step": 14452 + }, + { + "epoch": 1.5871952558752471, + "grad_norm": 2.411943197250366, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7119133472442627, + "num_tokens": 367234778.0, + "step": 14453 + }, + { + "epoch": 1.5873050735778609, + "grad_norm": 2.3970730304718018, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7162822484970093, + "num_tokens": 367259714.0, + "step": 14454 + }, + { + "epoch": 1.5874148912804744, + "grad_norm": 2.303645610809326, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7350074052810669, + "num_tokens": 367283081.0, + "step": 14455 + }, + { + "epoch": 1.587524708983088, + "grad_norm": 2.2230477333068848, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7023539543151855, + "num_tokens": 367308918.0, + "step": 14456 + }, + { + "epoch": 1.5876345266857017, + "grad_norm": 2.1877808570861816, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7333815097808838, + "num_tokens": 367335764.0, + "step": 14457 + }, + { + "epoch": 1.5877443443883155, + "grad_norm": 2.215697765350342, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7022528648376465, + "num_tokens": 367361817.0, + "step": 14458 + }, + { + "epoch": 1.5878541620909292, + "grad_norm": 2.1527793407440186, + "learning_rate": 1e-06, + "loss": 0.9961, + "mean_token_accuracy": 0.6991561055183411, + "num_tokens": 367390717.0, + "step": 14459 + }, + { + "epoch": 1.5879639797935428, + "grad_norm": 2.2584333419799805, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7194575667381287, + "num_tokens": 367416533.0, + "step": 14460 + }, + { + "epoch": 1.5880737974961563, + "grad_norm": 2.1751043796539307, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7108554244041443, + "num_tokens": 367442539.0, + "step": 14461 + }, + { + "epoch": 1.58818361519877, + "grad_norm": 2.2372655868530273, + "learning_rate": 1e-06, + "loss": 0.9917, + "mean_token_accuracy": 0.694833517074585, + "num_tokens": 367469294.0, + "step": 14462 + }, + { + "epoch": 1.5882934329013838, + "grad_norm": 2.158679246902466, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7196340560913086, + "num_tokens": 367496962.0, + "step": 14463 + }, + { + "epoch": 1.5884032506039973, + "grad_norm": 2.6362991333007812, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.734752893447876, + "num_tokens": 367516706.0, + "step": 14464 + }, + { + "epoch": 1.5885130683066109, + "grad_norm": 2.219097137451172, + "learning_rate": 1e-06, + "loss": 0.7961, + "mean_token_accuracy": 0.7510506510734558, + "num_tokens": 367541494.0, + "step": 14465 + }, + { + "epoch": 1.5886228860092246, + "grad_norm": 2.2264761924743652, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7099315524101257, + "num_tokens": 367567714.0, + "step": 14466 + }, + { + "epoch": 1.5887327037118384, + "grad_norm": 2.1055355072021484, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7122918367385864, + "num_tokens": 367595473.0, + "step": 14467 + }, + { + "epoch": 1.5888425214144521, + "grad_norm": 2.190727949142456, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7348666787147522, + "num_tokens": 367620507.0, + "step": 14468 + }, + { + "epoch": 1.5889523391170657, + "grad_norm": 2.2660725116729736, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7213930487632751, + "num_tokens": 367643041.0, + "step": 14469 + }, + { + "epoch": 1.5890621568196792, + "grad_norm": 2.1615023612976074, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7246930599212646, + "num_tokens": 367670209.0, + "step": 14470 + }, + { + "epoch": 1.589171974522293, + "grad_norm": 1.9649441242218018, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7181144952774048, + "num_tokens": 367702521.0, + "step": 14471 + }, + { + "epoch": 1.5892817922249067, + "grad_norm": 2.2744669914245605, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7352601289749146, + "num_tokens": 367725768.0, + "step": 14472 + }, + { + "epoch": 1.5893916099275203, + "grad_norm": 2.523069143295288, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7307056784629822, + "num_tokens": 367744602.0, + "step": 14473 + }, + { + "epoch": 1.589501427630134, + "grad_norm": 1.9871066808700562, + "learning_rate": 1e-06, + "loss": 0.9335, + "mean_token_accuracy": 0.7162572145462036, + "num_tokens": 367774450.0, + "step": 14474 + }, + { + "epoch": 1.5896112453327476, + "grad_norm": 2.3329453468322754, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7249852418899536, + "num_tokens": 367798682.0, + "step": 14475 + }, + { + "epoch": 1.5897210630353613, + "grad_norm": 2.217215061187744, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.7120627164840698, + "num_tokens": 367826374.0, + "step": 14476 + }, + { + "epoch": 1.589830880737975, + "grad_norm": 1.9542585611343384, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7259536981582642, + "num_tokens": 367860514.0, + "step": 14477 + }, + { + "epoch": 1.5899406984405886, + "grad_norm": 2.105674982070923, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.720630407333374, + "num_tokens": 367889277.0, + "step": 14478 + }, + { + "epoch": 1.5900505161432021, + "grad_norm": 2.046412944793701, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7242625951766968, + "num_tokens": 367916827.0, + "step": 14479 + }, + { + "epoch": 1.590160333845816, + "grad_norm": 2.3545031547546387, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7065762281417847, + "num_tokens": 367942319.0, + "step": 14480 + }, + { + "epoch": 1.5902701515484297, + "grad_norm": 2.321770429611206, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7211835980415344, + "num_tokens": 367965877.0, + "step": 14481 + }, + { + "epoch": 1.5903799692510434, + "grad_norm": 2.398148536682129, + "learning_rate": 1e-06, + "loss": 0.7875, + "mean_token_accuracy": 0.753120481967926, + "num_tokens": 367987617.0, + "step": 14482 + }, + { + "epoch": 1.590489786953657, + "grad_norm": 2.352029323577881, + "learning_rate": 1e-06, + "loss": 0.8105, + "mean_token_accuracy": 0.7430579662322998, + "num_tokens": 368009493.0, + "step": 14483 + }, + { + "epoch": 1.5905996046562705, + "grad_norm": 2.048449754714966, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7385163903236389, + "num_tokens": 368038367.0, + "step": 14484 + }, + { + "epoch": 1.5907094223588842, + "grad_norm": 2.3311703205108643, + "learning_rate": 1e-06, + "loss": 0.984, + "mean_token_accuracy": 0.700223445892334, + "num_tokens": 368064202.0, + "step": 14485 + }, + { + "epoch": 1.590819240061498, + "grad_norm": 2.114124059677124, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7200340032577515, + "num_tokens": 368092401.0, + "step": 14486 + }, + { + "epoch": 1.5909290577641115, + "grad_norm": 2.3427414894104004, + "learning_rate": 1e-06, + "loss": 0.8437, + "mean_token_accuracy": 0.7341724634170532, + "num_tokens": 368116373.0, + "step": 14487 + }, + { + "epoch": 1.5910388754667253, + "grad_norm": 2.207430362701416, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7211761474609375, + "num_tokens": 368141970.0, + "step": 14488 + }, + { + "epoch": 1.5911486931693388, + "grad_norm": 2.5185205936431885, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7111204862594604, + "num_tokens": 368163844.0, + "step": 14489 + }, + { + "epoch": 1.5912585108719526, + "grad_norm": 2.0391438007354736, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7316837310791016, + "num_tokens": 368192013.0, + "step": 14490 + }, + { + "epoch": 1.5913683285745663, + "grad_norm": 1.9491780996322632, + "learning_rate": 1e-06, + "loss": 1.0197, + "mean_token_accuracy": 0.6866796016693115, + "num_tokens": 368226311.0, + "step": 14491 + }, + { + "epoch": 1.5914781462771799, + "grad_norm": 1.935670018196106, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.6998206377029419, + "num_tokens": 368259530.0, + "step": 14492 + }, + { + "epoch": 1.5915879639797934, + "grad_norm": 2.01735520362854, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.7019081115722656, + "num_tokens": 368291342.0, + "step": 14493 + }, + { + "epoch": 1.5916977816824072, + "grad_norm": 2.5316951274871826, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.720202624797821, + "num_tokens": 368311634.0, + "step": 14494 + }, + { + "epoch": 1.591807599385021, + "grad_norm": 2.4780049324035645, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7324093580245972, + "num_tokens": 368334152.0, + "step": 14495 + }, + { + "epoch": 1.5919174170876347, + "grad_norm": 2.1947054862976074, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7630302906036377, + "num_tokens": 368359369.0, + "step": 14496 + }, + { + "epoch": 1.5920272347902482, + "grad_norm": 1.8976233005523682, + "learning_rate": 1e-06, + "loss": 0.8144, + "mean_token_accuracy": 0.7400621771812439, + "num_tokens": 368391854.0, + "step": 14497 + }, + { + "epoch": 1.5921370524928617, + "grad_norm": 1.9488129615783691, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.6999367475509644, + "num_tokens": 368425721.0, + "step": 14498 + }, + { + "epoch": 1.5922468701954755, + "grad_norm": 2.0658013820648193, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7052820920944214, + "num_tokens": 368456885.0, + "step": 14499 + }, + { + "epoch": 1.5923566878980893, + "grad_norm": 2.283719539642334, + "learning_rate": 1e-06, + "loss": 0.8085, + "mean_token_accuracy": 0.7436318397521973, + "num_tokens": 368482133.0, + "step": 14500 + }, + { + "epoch": 1.5924665056007028, + "grad_norm": 2.7514028549194336, + "learning_rate": 1e-06, + "loss": 0.7558, + "mean_token_accuracy": 0.7548370361328125, + "num_tokens": 368499658.0, + "step": 14501 + }, + { + "epoch": 1.5925763233033163, + "grad_norm": 2.0747604370117188, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7189669609069824, + "num_tokens": 368529157.0, + "step": 14502 + }, + { + "epoch": 1.59268614100593, + "grad_norm": 2.4031050205230713, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7065319418907166, + "num_tokens": 368555531.0, + "step": 14503 + }, + { + "epoch": 1.5927959587085438, + "grad_norm": 2.4504988193511963, + "learning_rate": 1e-06, + "loss": 1.0112, + "mean_token_accuracy": 0.694025993347168, + "num_tokens": 368579340.0, + "step": 14504 + }, + { + "epoch": 1.5929057764111576, + "grad_norm": 2.529860734939575, + "learning_rate": 1e-06, + "loss": 0.905, + "mean_token_accuracy": 0.7171971201896667, + "num_tokens": 368599221.0, + "step": 14505 + }, + { + "epoch": 1.5930155941137711, + "grad_norm": 2.376732110977173, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7222995758056641, + "num_tokens": 368623024.0, + "step": 14506 + }, + { + "epoch": 1.5931254118163847, + "grad_norm": 2.140942335128784, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7229203581809998, + "num_tokens": 368649957.0, + "step": 14507 + }, + { + "epoch": 1.5932352295189984, + "grad_norm": 2.051706075668335, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7131391167640686, + "num_tokens": 368679284.0, + "step": 14508 + }, + { + "epoch": 1.5933450472216122, + "grad_norm": 2.468050479888916, + "learning_rate": 1e-06, + "loss": 0.8764, + "mean_token_accuracy": 0.7213149070739746, + "num_tokens": 368700210.0, + "step": 14509 + }, + { + "epoch": 1.593454864924226, + "grad_norm": 2.2885324954986572, + "learning_rate": 1e-06, + "loss": 0.9094, + "mean_token_accuracy": 0.7159611582756042, + "num_tokens": 368724848.0, + "step": 14510 + }, + { + "epoch": 1.5935646826268395, + "grad_norm": 2.161404848098755, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7116727828979492, + "num_tokens": 368751337.0, + "step": 14511 + }, + { + "epoch": 1.593674500329453, + "grad_norm": 2.208306074142456, + "learning_rate": 1e-06, + "loss": 0.862, + "mean_token_accuracy": 0.726932942867279, + "num_tokens": 368777381.0, + "step": 14512 + }, + { + "epoch": 1.5937843180320668, + "grad_norm": 2.409393310546875, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7098138332366943, + "num_tokens": 368799795.0, + "step": 14513 + }, + { + "epoch": 1.5938941357346805, + "grad_norm": 2.337921142578125, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7066853642463684, + "num_tokens": 368824981.0, + "step": 14514 + }, + { + "epoch": 1.594003953437294, + "grad_norm": 2.2352707386016846, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7146925330162048, + "num_tokens": 368850541.0, + "step": 14515 + }, + { + "epoch": 1.5941137711399076, + "grad_norm": 2.637401580810547, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7277354598045349, + "num_tokens": 368869926.0, + "step": 14516 + }, + { + "epoch": 1.5942235888425214, + "grad_norm": 2.355081558227539, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.71048903465271, + "num_tokens": 368894096.0, + "step": 14517 + }, + { + "epoch": 1.5943334065451351, + "grad_norm": 2.101595401763916, + "learning_rate": 1e-06, + "loss": 0.9956, + "mean_token_accuracy": 0.6953974962234497, + "num_tokens": 368925324.0, + "step": 14518 + }, + { + "epoch": 1.5944432242477489, + "grad_norm": 2.240598440170288, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7378925681114197, + "num_tokens": 368949710.0, + "step": 14519 + }, + { + "epoch": 1.5945530419503624, + "grad_norm": 2.4500069618225098, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.723388671875, + "num_tokens": 368970699.0, + "step": 14520 + }, + { + "epoch": 1.594662859652976, + "grad_norm": 2.302464008331299, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7260724306106567, + "num_tokens": 368996334.0, + "step": 14521 + }, + { + "epoch": 1.5947726773555897, + "grad_norm": 2.3699750900268555, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.7112149000167847, + "num_tokens": 369019993.0, + "step": 14522 + }, + { + "epoch": 1.5948824950582035, + "grad_norm": 2.1236982345581055, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6888071894645691, + "num_tokens": 369049511.0, + "step": 14523 + }, + { + "epoch": 1.5949923127608172, + "grad_norm": 2.0762839317321777, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7323616743087769, + "num_tokens": 369078947.0, + "step": 14524 + }, + { + "epoch": 1.5951021304634307, + "grad_norm": 2.322247266769409, + "learning_rate": 1e-06, + "loss": 0.941, + "mean_token_accuracy": 0.7120896577835083, + "num_tokens": 369102896.0, + "step": 14525 + }, + { + "epoch": 1.5952119481660443, + "grad_norm": 2.2729616165161133, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.7022891640663147, + "num_tokens": 369127917.0, + "step": 14526 + }, + { + "epoch": 1.595321765868658, + "grad_norm": 2.3521575927734375, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7168710231781006, + "num_tokens": 369151995.0, + "step": 14527 + }, + { + "epoch": 1.5954315835712718, + "grad_norm": 2.1996796131134033, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7254375219345093, + "num_tokens": 369178876.0, + "step": 14528 + }, + { + "epoch": 1.5955414012738853, + "grad_norm": 2.6388862133026123, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7217085361480713, + "num_tokens": 369200486.0, + "step": 14529 + }, + { + "epoch": 1.5956512189764989, + "grad_norm": 2.0793590545654297, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7210963368415833, + "num_tokens": 369229162.0, + "step": 14530 + }, + { + "epoch": 1.5957610366791126, + "grad_norm": 2.448967933654785, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7286644577980042, + "num_tokens": 369252116.0, + "step": 14531 + }, + { + "epoch": 1.5958708543817264, + "grad_norm": 2.5998733043670654, + "learning_rate": 1e-06, + "loss": 0.8018, + "mean_token_accuracy": 0.7526443004608154, + "num_tokens": 369270915.0, + "step": 14532 + }, + { + "epoch": 1.5959806720843401, + "grad_norm": 2.1895453929901123, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7096642255783081, + "num_tokens": 369299036.0, + "step": 14533 + }, + { + "epoch": 1.5960904897869537, + "grad_norm": 2.531782627105713, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7369136810302734, + "num_tokens": 369319327.0, + "step": 14534 + }, + { + "epoch": 1.5962003074895672, + "grad_norm": 2.432666540145874, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.701797604560852, + "num_tokens": 369343433.0, + "step": 14535 + }, + { + "epoch": 1.596310125192181, + "grad_norm": 2.258301258087158, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7174298763275146, + "num_tokens": 369368142.0, + "step": 14536 + }, + { + "epoch": 1.5964199428947947, + "grad_norm": 2.3053746223449707, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.6998682022094727, + "num_tokens": 369394752.0, + "step": 14537 + }, + { + "epoch": 1.5965297605974083, + "grad_norm": 2.237880229949951, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7380666732788086, + "num_tokens": 369420490.0, + "step": 14538 + }, + { + "epoch": 1.596639578300022, + "grad_norm": 2.1479074954986572, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7245744466781616, + "num_tokens": 369447713.0, + "step": 14539 + }, + { + "epoch": 1.5967493960026355, + "grad_norm": 2.174600124359131, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7133796811103821, + "num_tokens": 369475184.0, + "step": 14540 + }, + { + "epoch": 1.5968592137052493, + "grad_norm": 2.160557985305786, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7107113003730774, + "num_tokens": 369502230.0, + "step": 14541 + }, + { + "epoch": 1.596969031407863, + "grad_norm": 2.5248663425445557, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7244483232498169, + "num_tokens": 369522981.0, + "step": 14542 + }, + { + "epoch": 1.5970788491104766, + "grad_norm": 2.4034411907196045, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.72677081823349, + "num_tokens": 369543797.0, + "step": 14543 + }, + { + "epoch": 1.5971886668130901, + "grad_norm": 2.39005184173584, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7254104018211365, + "num_tokens": 369566051.0, + "step": 14544 + }, + { + "epoch": 1.5972984845157039, + "grad_norm": 2.4878458976745605, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7238026857376099, + "num_tokens": 369586643.0, + "step": 14545 + }, + { + "epoch": 1.5974083022183176, + "grad_norm": 2.1569900512695312, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.7012296319007874, + "num_tokens": 369614533.0, + "step": 14546 + }, + { + "epoch": 1.5975181199209314, + "grad_norm": 2.5063841342926025, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7248080372810364, + "num_tokens": 369635715.0, + "step": 14547 + }, + { + "epoch": 1.597627937623545, + "grad_norm": 2.1053860187530518, + "learning_rate": 1e-06, + "loss": 0.8352, + "mean_token_accuracy": 0.7395514249801636, + "num_tokens": 369664193.0, + "step": 14548 + }, + { + "epoch": 1.5977377553261585, + "grad_norm": 2.242879629135132, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7327760457992554, + "num_tokens": 369690085.0, + "step": 14549 + }, + { + "epoch": 1.5978475730287722, + "grad_norm": 2.1883463859558105, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7003666162490845, + "num_tokens": 369717363.0, + "step": 14550 + }, + { + "epoch": 1.597957390731386, + "grad_norm": 2.3792760372161865, + "learning_rate": 1e-06, + "loss": 0.8023, + "mean_token_accuracy": 0.7406681776046753, + "num_tokens": 369739894.0, + "step": 14551 + }, + { + "epoch": 1.5980672084339995, + "grad_norm": 2.3011488914489746, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.718690037727356, + "num_tokens": 369764838.0, + "step": 14552 + }, + { + "epoch": 1.5981770261366133, + "grad_norm": 2.2572200298309326, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7195305228233337, + "num_tokens": 369789670.0, + "step": 14553 + }, + { + "epoch": 1.5982868438392268, + "grad_norm": 2.2067863941192627, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7305879592895508, + "num_tokens": 369816385.0, + "step": 14554 + }, + { + "epoch": 1.5983966615418406, + "grad_norm": 2.764406681060791, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7464287877082825, + "num_tokens": 369834584.0, + "step": 14555 + }, + { + "epoch": 1.5985064792444543, + "grad_norm": 2.6868948936462402, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7194069027900696, + "num_tokens": 369854283.0, + "step": 14556 + }, + { + "epoch": 1.5986162969470679, + "grad_norm": 2.22912859916687, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7096691727638245, + "num_tokens": 369883923.0, + "step": 14557 + }, + { + "epoch": 1.5987261146496814, + "grad_norm": 2.7337276935577393, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7216444611549377, + "num_tokens": 369901211.0, + "step": 14558 + }, + { + "epoch": 1.5988359323522952, + "grad_norm": 2.207860231399536, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7257804870605469, + "num_tokens": 369927500.0, + "step": 14559 + }, + { + "epoch": 1.598945750054909, + "grad_norm": 2.322993516921997, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7241937518119812, + "num_tokens": 369951697.0, + "step": 14560 + }, + { + "epoch": 1.5990555677575227, + "grad_norm": 2.2505311965942383, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7163171768188477, + "num_tokens": 369978067.0, + "step": 14561 + }, + { + "epoch": 1.5991653854601362, + "grad_norm": 2.373445510864258, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.7173258066177368, + "num_tokens": 370003853.0, + "step": 14562 + }, + { + "epoch": 1.5992752031627497, + "grad_norm": 2.4769814014434814, + "learning_rate": 1e-06, + "loss": 0.8873, + "mean_token_accuracy": 0.7240839004516602, + "num_tokens": 370025464.0, + "step": 14563 + }, + { + "epoch": 1.5993850208653635, + "grad_norm": 2.1682393550872803, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7124793529510498, + "num_tokens": 370052970.0, + "step": 14564 + }, + { + "epoch": 1.5994948385679773, + "grad_norm": 2.263617753982544, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7215198278427124, + "num_tokens": 370079641.0, + "step": 14565 + }, + { + "epoch": 1.5996046562705908, + "grad_norm": 2.1536219120025635, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.729874312877655, + "num_tokens": 370104783.0, + "step": 14566 + }, + { + "epoch": 1.5997144739732043, + "grad_norm": 2.327118396759033, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7114794254302979, + "num_tokens": 370131794.0, + "step": 14567 + }, + { + "epoch": 1.599824291675818, + "grad_norm": 2.2062926292419434, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7091248035430908, + "num_tokens": 370158250.0, + "step": 14568 + }, + { + "epoch": 1.5999341093784318, + "grad_norm": 2.1711668968200684, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7337292432785034, + "num_tokens": 370183837.0, + "step": 14569 + }, + { + "epoch": 1.6000439270810456, + "grad_norm": 2.244750499725342, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7104858756065369, + "num_tokens": 370210052.0, + "step": 14570 + }, + { + "epoch": 1.6001537447836591, + "grad_norm": 2.239032506942749, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.719372034072876, + "num_tokens": 370235031.0, + "step": 14571 + }, + { + "epoch": 1.6002635624862727, + "grad_norm": 2.0843706130981445, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7229264378547668, + "num_tokens": 370262393.0, + "step": 14572 + }, + { + "epoch": 1.6003733801888864, + "grad_norm": 2.379204034805298, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.700019359588623, + "num_tokens": 370287730.0, + "step": 14573 + }, + { + "epoch": 1.6004831978915002, + "grad_norm": 2.1858532428741455, + "learning_rate": 1e-06, + "loss": 0.8267, + "mean_token_accuracy": 0.7393149733543396, + "num_tokens": 370314585.0, + "step": 14574 + }, + { + "epoch": 1.600593015594114, + "grad_norm": 2.5930161476135254, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7299378514289856, + "num_tokens": 370334932.0, + "step": 14575 + }, + { + "epoch": 1.6007028332967275, + "grad_norm": 2.1651270389556885, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.70884108543396, + "num_tokens": 370361285.0, + "step": 14576 + }, + { + "epoch": 1.600812650999341, + "grad_norm": 2.3441197872161865, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7039763927459717, + "num_tokens": 370384529.0, + "step": 14577 + }, + { + "epoch": 1.6009224687019548, + "grad_norm": 2.7118425369262695, + "learning_rate": 1e-06, + "loss": 0.7809, + "mean_token_accuracy": 0.7493133544921875, + "num_tokens": 370401203.0, + "step": 14578 + }, + { + "epoch": 1.6010322864045685, + "grad_norm": 2.6109790802001953, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.7349236607551575, + "num_tokens": 370421285.0, + "step": 14579 + }, + { + "epoch": 1.601142104107182, + "grad_norm": 2.0869431495666504, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7224427461624146, + "num_tokens": 370450749.0, + "step": 14580 + }, + { + "epoch": 1.6012519218097956, + "grad_norm": 1.867262840270996, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7329976558685303, + "num_tokens": 370485978.0, + "step": 14581 + }, + { + "epoch": 1.6013617395124093, + "grad_norm": 2.2628560066223145, + "learning_rate": 1e-06, + "loss": 0.9595, + "mean_token_accuracy": 0.7084125280380249, + "num_tokens": 370514791.0, + "step": 14582 + }, + { + "epoch": 1.601471557215023, + "grad_norm": 2.5676238536834717, + "learning_rate": 1e-06, + "loss": 0.8805, + "mean_token_accuracy": 0.7197759747505188, + "num_tokens": 370535991.0, + "step": 14583 + }, + { + "epoch": 1.6015813749176369, + "grad_norm": 2.1808197498321533, + "learning_rate": 1e-06, + "loss": 1.0245, + "mean_token_accuracy": 0.6918572187423706, + "num_tokens": 370565919.0, + "step": 14584 + }, + { + "epoch": 1.6016911926202504, + "grad_norm": 2.271320343017578, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.730995774269104, + "num_tokens": 370589695.0, + "step": 14585 + }, + { + "epoch": 1.601801010322864, + "grad_norm": 2.309997081756592, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7153127193450928, + "num_tokens": 370615756.0, + "step": 14586 + }, + { + "epoch": 1.6019108280254777, + "grad_norm": 2.2452898025512695, + "learning_rate": 1e-06, + "loss": 0.9764, + "mean_token_accuracy": 0.6942145824432373, + "num_tokens": 370642430.0, + "step": 14587 + }, + { + "epoch": 1.6020206457280914, + "grad_norm": 2.447183609008789, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7117934226989746, + "num_tokens": 370664737.0, + "step": 14588 + }, + { + "epoch": 1.602130463430705, + "grad_norm": 2.2486629486083984, + "learning_rate": 1e-06, + "loss": 1.0185, + "mean_token_accuracy": 0.6886606216430664, + "num_tokens": 370691751.0, + "step": 14589 + }, + { + "epoch": 1.6022402811333187, + "grad_norm": 2.068765163421631, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.7189223170280457, + "num_tokens": 370719262.0, + "step": 14590 + }, + { + "epoch": 1.6023500988359323, + "grad_norm": 2.0377073287963867, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6986640691757202, + "num_tokens": 370750346.0, + "step": 14591 + }, + { + "epoch": 1.602459916538546, + "grad_norm": 2.7937002182006836, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7261821031570435, + "num_tokens": 370769241.0, + "step": 14592 + }, + { + "epoch": 1.6025697342411598, + "grad_norm": 2.193326950073242, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.710638701915741, + "num_tokens": 370798032.0, + "step": 14593 + }, + { + "epoch": 1.6026795519437733, + "grad_norm": 2.1313254833221436, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7222801446914673, + "num_tokens": 370826635.0, + "step": 14594 + }, + { + "epoch": 1.6027893696463869, + "grad_norm": 2.2815277576446533, + "learning_rate": 1e-06, + "loss": 1.0072, + "mean_token_accuracy": 0.6992200613021851, + "num_tokens": 370854410.0, + "step": 14595 + }, + { + "epoch": 1.6028991873490006, + "grad_norm": 2.0722477436065674, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7301185131072998, + "num_tokens": 370881614.0, + "step": 14596 + }, + { + "epoch": 1.6030090050516144, + "grad_norm": 2.0652565956115723, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7263011932373047, + "num_tokens": 370910260.0, + "step": 14597 + }, + { + "epoch": 1.6031188227542281, + "grad_norm": 2.1208221912384033, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7120436429977417, + "num_tokens": 370935586.0, + "step": 14598 + }, + { + "epoch": 1.6032286404568417, + "grad_norm": 2.2484867572784424, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7092795372009277, + "num_tokens": 370961772.0, + "step": 14599 + }, + { + "epoch": 1.6033384581594552, + "grad_norm": 2.093330144882202, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7036562561988831, + "num_tokens": 370992788.0, + "step": 14600 + }, + { + "epoch": 1.603448275862069, + "grad_norm": 2.09720778465271, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7242813110351562, + "num_tokens": 371019968.0, + "step": 14601 + }, + { + "epoch": 1.6035580935646827, + "grad_norm": 2.3474133014678955, + "learning_rate": 1e-06, + "loss": 0.7819, + "mean_token_accuracy": 0.7567179203033447, + "num_tokens": 371043501.0, + "step": 14602 + }, + { + "epoch": 1.6036679112672962, + "grad_norm": 2.152259111404419, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7267035841941833, + "num_tokens": 371070052.0, + "step": 14603 + }, + { + "epoch": 1.60377772896991, + "grad_norm": 2.3616364002227783, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7372449636459351, + "num_tokens": 371091838.0, + "step": 14604 + }, + { + "epoch": 1.6038875466725235, + "grad_norm": 2.3956639766693115, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7427616715431213, + "num_tokens": 371115964.0, + "step": 14605 + }, + { + "epoch": 1.6039973643751373, + "grad_norm": 2.104909658432007, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7234597206115723, + "num_tokens": 371143193.0, + "step": 14606 + }, + { + "epoch": 1.604107182077751, + "grad_norm": 2.1958417892456055, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7178324460983276, + "num_tokens": 371170236.0, + "step": 14607 + }, + { + "epoch": 1.6042169997803646, + "grad_norm": 2.3546042442321777, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7361760139465332, + "num_tokens": 371193791.0, + "step": 14608 + }, + { + "epoch": 1.6043268174829781, + "grad_norm": 2.525179624557495, + "learning_rate": 1e-06, + "loss": 0.7728, + "mean_token_accuracy": 0.7560862898826599, + "num_tokens": 371213109.0, + "step": 14609 + }, + { + "epoch": 1.6044366351855919, + "grad_norm": 1.973031997680664, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.7277354001998901, + "num_tokens": 371245664.0, + "step": 14610 + }, + { + "epoch": 1.6045464528882056, + "grad_norm": 2.1745569705963135, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7002875804901123, + "num_tokens": 371275028.0, + "step": 14611 + }, + { + "epoch": 1.6046562705908194, + "grad_norm": 2.104006052017212, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7025344967842102, + "num_tokens": 371306305.0, + "step": 14612 + }, + { + "epoch": 1.604766088293433, + "grad_norm": 2.2556264400482178, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7327973246574402, + "num_tokens": 371330444.0, + "step": 14613 + }, + { + "epoch": 1.6048759059960465, + "grad_norm": 2.2824454307556152, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7286955118179321, + "num_tokens": 371354129.0, + "step": 14614 + }, + { + "epoch": 1.6049857236986602, + "grad_norm": 2.0531561374664307, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7153195738792419, + "num_tokens": 371383542.0, + "step": 14615 + }, + { + "epoch": 1.605095541401274, + "grad_norm": 2.33133864402771, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7087345123291016, + "num_tokens": 371406921.0, + "step": 14616 + }, + { + "epoch": 1.6052053591038875, + "grad_norm": 2.1868579387664795, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7093002796173096, + "num_tokens": 371432845.0, + "step": 14617 + }, + { + "epoch": 1.605315176806501, + "grad_norm": 1.9855341911315918, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.718684196472168, + "num_tokens": 371463607.0, + "step": 14618 + }, + { + "epoch": 1.6054249945091148, + "grad_norm": 2.365532636642456, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7287605404853821, + "num_tokens": 371485904.0, + "step": 14619 + }, + { + "epoch": 1.6055348122117286, + "grad_norm": 2.057002544403076, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.6980152726173401, + "num_tokens": 371514605.0, + "step": 14620 + }, + { + "epoch": 1.6056446299143423, + "grad_norm": 2.144597291946411, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7150989770889282, + "num_tokens": 371541886.0, + "step": 14621 + }, + { + "epoch": 1.6057544476169558, + "grad_norm": 2.1664209365844727, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7143765687942505, + "num_tokens": 371569540.0, + "step": 14622 + }, + { + "epoch": 1.6058642653195694, + "grad_norm": 2.1329076290130615, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7454186677932739, + "num_tokens": 371595203.0, + "step": 14623 + }, + { + "epoch": 1.6059740830221831, + "grad_norm": 2.217270612716675, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.6975598335266113, + "num_tokens": 371621365.0, + "step": 14624 + }, + { + "epoch": 1.606083900724797, + "grad_norm": 2.7208642959594727, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7297547459602356, + "num_tokens": 371640339.0, + "step": 14625 + }, + { + "epoch": 1.6061937184274107, + "grad_norm": 2.611398220062256, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7329879403114319, + "num_tokens": 371659725.0, + "step": 14626 + }, + { + "epoch": 1.6063035361300242, + "grad_norm": 2.545397996902466, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7466065287590027, + "num_tokens": 371679535.0, + "step": 14627 + }, + { + "epoch": 1.6064133538326377, + "grad_norm": 2.1766014099121094, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7131175994873047, + "num_tokens": 371707125.0, + "step": 14628 + }, + { + "epoch": 1.6065231715352515, + "grad_norm": 2.3998665809631348, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7190978527069092, + "num_tokens": 371730853.0, + "step": 14629 + }, + { + "epoch": 1.6066329892378652, + "grad_norm": 2.2704544067382812, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7090705633163452, + "num_tokens": 371758113.0, + "step": 14630 + }, + { + "epoch": 1.6067428069404788, + "grad_norm": 2.1802501678466797, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6981668472290039, + "num_tokens": 371788362.0, + "step": 14631 + }, + { + "epoch": 1.6068526246430923, + "grad_norm": 2.224000930786133, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7207159399986267, + "num_tokens": 371813356.0, + "step": 14632 + }, + { + "epoch": 1.606962442345706, + "grad_norm": 2.422814130783081, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7120555639266968, + "num_tokens": 371836134.0, + "step": 14633 + }, + { + "epoch": 1.6070722600483198, + "grad_norm": 2.3224260807037354, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7166932821273804, + "num_tokens": 371862077.0, + "step": 14634 + }, + { + "epoch": 1.6071820777509336, + "grad_norm": 2.000347852706909, + "learning_rate": 1e-06, + "loss": 0.9185, + "mean_token_accuracy": 0.7157992124557495, + "num_tokens": 371894400.0, + "step": 14635 + }, + { + "epoch": 1.6072918954535471, + "grad_norm": 2.280360221862793, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.7092336416244507, + "num_tokens": 371920548.0, + "step": 14636 + }, + { + "epoch": 1.6074017131561606, + "grad_norm": 2.2010090351104736, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7223129272460938, + "num_tokens": 371946133.0, + "step": 14637 + }, + { + "epoch": 1.6075115308587744, + "grad_norm": 2.1115009784698486, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7265454530715942, + "num_tokens": 371974942.0, + "step": 14638 + }, + { + "epoch": 1.6076213485613882, + "grad_norm": 2.058051586151123, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.709064245223999, + "num_tokens": 372004949.0, + "step": 14639 + }, + { + "epoch": 1.607731166264002, + "grad_norm": 2.342672824859619, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7132634520530701, + "num_tokens": 372029187.0, + "step": 14640 + }, + { + "epoch": 1.6078409839666155, + "grad_norm": 2.4079596996307373, + "learning_rate": 1e-06, + "loss": 0.8441, + "mean_token_accuracy": 0.7337678670883179, + "num_tokens": 372050779.0, + "step": 14641 + }, + { + "epoch": 1.607950801669229, + "grad_norm": 1.930379867553711, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7305188775062561, + "num_tokens": 372082555.0, + "step": 14642 + }, + { + "epoch": 1.6080606193718427, + "grad_norm": 1.9867854118347168, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7171894907951355, + "num_tokens": 372114444.0, + "step": 14643 + }, + { + "epoch": 1.6081704370744565, + "grad_norm": 2.2370049953460693, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7212403416633606, + "num_tokens": 372140722.0, + "step": 14644 + }, + { + "epoch": 1.60828025477707, + "grad_norm": 2.476274251937866, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.7068617343902588, + "num_tokens": 372165126.0, + "step": 14645 + }, + { + "epoch": 1.6083900724796836, + "grad_norm": 2.302009344100952, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7300195693969727, + "num_tokens": 372188798.0, + "step": 14646 + }, + { + "epoch": 1.6084998901822973, + "grad_norm": 2.2911858558654785, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7245475053787231, + "num_tokens": 372213273.0, + "step": 14647 + }, + { + "epoch": 1.608609707884911, + "grad_norm": 2.3669331073760986, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7285192608833313, + "num_tokens": 372236340.0, + "step": 14648 + }, + { + "epoch": 1.6087195255875248, + "grad_norm": 2.1718926429748535, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7202587127685547, + "num_tokens": 372262405.0, + "step": 14649 + }, + { + "epoch": 1.6088293432901384, + "grad_norm": 2.1712074279785156, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7154190540313721, + "num_tokens": 372288718.0, + "step": 14650 + }, + { + "epoch": 1.608939160992752, + "grad_norm": 2.3348464965820312, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.716448187828064, + "num_tokens": 372311248.0, + "step": 14651 + }, + { + "epoch": 1.6090489786953657, + "grad_norm": 2.3887720108032227, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7222875356674194, + "num_tokens": 372335138.0, + "step": 14652 + }, + { + "epoch": 1.6091587963979794, + "grad_norm": 2.591414213180542, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7131401896476746, + "num_tokens": 372356172.0, + "step": 14653 + }, + { + "epoch": 1.609268614100593, + "grad_norm": 2.32321834564209, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7143983840942383, + "num_tokens": 372379941.0, + "step": 14654 + }, + { + "epoch": 1.6093784318032067, + "grad_norm": 2.1123862266540527, + "learning_rate": 1e-06, + "loss": 1.0062, + "mean_token_accuracy": 0.6952759027481079, + "num_tokens": 372410040.0, + "step": 14655 + }, + { + "epoch": 1.6094882495058203, + "grad_norm": 2.233400344848633, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7129517793655396, + "num_tokens": 372435424.0, + "step": 14656 + }, + { + "epoch": 1.609598067208434, + "grad_norm": 2.225978136062622, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.706784188747406, + "num_tokens": 372463046.0, + "step": 14657 + }, + { + "epoch": 1.6097078849110478, + "grad_norm": 2.6828248500823975, + "learning_rate": 1e-06, + "loss": 0.8129, + "mean_token_accuracy": 0.7355310916900635, + "num_tokens": 372481804.0, + "step": 14658 + }, + { + "epoch": 1.6098177026136613, + "grad_norm": 2.316179037094116, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.723770022392273, + "num_tokens": 372504870.0, + "step": 14659 + }, + { + "epoch": 1.6099275203162748, + "grad_norm": 2.2911999225616455, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7162511348724365, + "num_tokens": 372529273.0, + "step": 14660 + }, + { + "epoch": 1.6100373380188886, + "grad_norm": 1.987931489944458, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7315032482147217, + "num_tokens": 372560949.0, + "step": 14661 + }, + { + "epoch": 1.6101471557215024, + "grad_norm": 2.3280839920043945, + "learning_rate": 1e-06, + "loss": 0.8231, + "mean_token_accuracy": 0.7435683012008667, + "num_tokens": 372586221.0, + "step": 14662 + }, + { + "epoch": 1.6102569734241161, + "grad_norm": 2.106491804122925, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7061129212379456, + "num_tokens": 372617053.0, + "step": 14663 + }, + { + "epoch": 1.6103667911267296, + "grad_norm": 2.1061673164367676, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7045251131057739, + "num_tokens": 372646318.0, + "step": 14664 + }, + { + "epoch": 1.6104766088293432, + "grad_norm": 2.7299234867095947, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7201967835426331, + "num_tokens": 372665548.0, + "step": 14665 + }, + { + "epoch": 1.610586426531957, + "grad_norm": 2.7287073135375977, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7153968811035156, + "num_tokens": 372683856.0, + "step": 14666 + }, + { + "epoch": 1.6106962442345707, + "grad_norm": 2.1202194690704346, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.741192102432251, + "num_tokens": 372712119.0, + "step": 14667 + }, + { + "epoch": 1.6108060619371842, + "grad_norm": 2.2829222679138184, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7204177379608154, + "num_tokens": 372738489.0, + "step": 14668 + }, + { + "epoch": 1.610915879639798, + "grad_norm": 2.1158390045166016, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7014905214309692, + "num_tokens": 372767676.0, + "step": 14669 + }, + { + "epoch": 1.6110256973424115, + "grad_norm": 2.173494338989258, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7130420207977295, + "num_tokens": 372794995.0, + "step": 14670 + }, + { + "epoch": 1.6111355150450253, + "grad_norm": 2.269435405731201, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7088265419006348, + "num_tokens": 372821353.0, + "step": 14671 + }, + { + "epoch": 1.611245332747639, + "grad_norm": 1.993316888809204, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.6961699724197388, + "num_tokens": 372852757.0, + "step": 14672 + }, + { + "epoch": 1.6113551504502526, + "grad_norm": 2.266622543334961, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.6987491846084595, + "num_tokens": 372879893.0, + "step": 14673 + }, + { + "epoch": 1.611464968152866, + "grad_norm": 2.1622159481048584, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.71155846118927, + "num_tokens": 372908129.0, + "step": 14674 + }, + { + "epoch": 1.6115747858554799, + "grad_norm": 2.2427093982696533, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7295880317687988, + "num_tokens": 372934412.0, + "step": 14675 + }, + { + "epoch": 1.6116846035580936, + "grad_norm": 2.4522414207458496, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7157827615737915, + "num_tokens": 372955509.0, + "step": 14676 + }, + { + "epoch": 1.6117944212607074, + "grad_norm": 2.0146126747131348, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7126814126968384, + "num_tokens": 372985811.0, + "step": 14677 + }, + { + "epoch": 1.611904238963321, + "grad_norm": 2.5547828674316406, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7129086852073669, + "num_tokens": 373006608.0, + "step": 14678 + }, + { + "epoch": 1.6120140566659344, + "grad_norm": 2.269421100616455, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7295079231262207, + "num_tokens": 373030841.0, + "step": 14679 + }, + { + "epoch": 1.6121238743685482, + "grad_norm": 2.6487603187561035, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.741461992263794, + "num_tokens": 373049181.0, + "step": 14680 + }, + { + "epoch": 1.612233692071162, + "grad_norm": 2.4828245639801025, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7030068635940552, + "num_tokens": 373070242.0, + "step": 14681 + }, + { + "epoch": 1.6123435097737755, + "grad_norm": 2.530858039855957, + "learning_rate": 1e-06, + "loss": 0.8097, + "mean_token_accuracy": 0.7347844839096069, + "num_tokens": 373090096.0, + "step": 14682 + }, + { + "epoch": 1.612453327476389, + "grad_norm": 2.2234184741973877, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7317695021629333, + "num_tokens": 373119762.0, + "step": 14683 + }, + { + "epoch": 1.6125631451790028, + "grad_norm": 2.2604751586914062, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7144822478294373, + "num_tokens": 373146036.0, + "step": 14684 + }, + { + "epoch": 1.6126729628816165, + "grad_norm": 2.259087324142456, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7298405170440674, + "num_tokens": 373171967.0, + "step": 14685 + }, + { + "epoch": 1.6127827805842303, + "grad_norm": 2.3173725605010986, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7291105389595032, + "num_tokens": 373196567.0, + "step": 14686 + }, + { + "epoch": 1.6128925982868438, + "grad_norm": 2.096524238586426, + "learning_rate": 1e-06, + "loss": 0.9297, + "mean_token_accuracy": 0.7190688848495483, + "num_tokens": 373226374.0, + "step": 14687 + }, + { + "epoch": 1.6130024159894574, + "grad_norm": 2.331803560256958, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7189846038818359, + "num_tokens": 373250415.0, + "step": 14688 + }, + { + "epoch": 1.6131122336920711, + "grad_norm": 2.473529577255249, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7240341901779175, + "num_tokens": 373272142.0, + "step": 14689 + }, + { + "epoch": 1.6132220513946849, + "grad_norm": 2.5270230770111084, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7099561095237732, + "num_tokens": 373292809.0, + "step": 14690 + }, + { + "epoch": 1.6133318690972986, + "grad_norm": 2.446321725845337, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7272428870201111, + "num_tokens": 373314792.0, + "step": 14691 + }, + { + "epoch": 1.6134416867999122, + "grad_norm": 2.3089239597320557, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7137891054153442, + "num_tokens": 373340112.0, + "step": 14692 + }, + { + "epoch": 1.6135515045025257, + "grad_norm": 2.3580517768859863, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7148760557174683, + "num_tokens": 373364146.0, + "step": 14693 + }, + { + "epoch": 1.6136613222051395, + "grad_norm": 2.3063788414001465, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7160594463348389, + "num_tokens": 373389508.0, + "step": 14694 + }, + { + "epoch": 1.6137711399077532, + "grad_norm": 2.7831296920776367, + "learning_rate": 1e-06, + "loss": 0.8701, + "mean_token_accuracy": 0.7348384857177734, + "num_tokens": 373407412.0, + "step": 14695 + }, + { + "epoch": 1.6138809576103668, + "grad_norm": 2.3815715312957764, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7217270135879517, + "num_tokens": 373429286.0, + "step": 14696 + }, + { + "epoch": 1.6139907753129803, + "grad_norm": 2.19106125831604, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7192574739456177, + "num_tokens": 373456317.0, + "step": 14697 + }, + { + "epoch": 1.614100593015594, + "grad_norm": 2.2346251010894775, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7079855799674988, + "num_tokens": 373481414.0, + "step": 14698 + }, + { + "epoch": 1.6142104107182078, + "grad_norm": 2.7900993824005127, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7261114716529846, + "num_tokens": 373498321.0, + "step": 14699 + }, + { + "epoch": 1.6143202284208216, + "grad_norm": 2.2457993030548096, + "learning_rate": 1e-06, + "loss": 0.8917, + "mean_token_accuracy": 0.7197976112365723, + "num_tokens": 373525446.0, + "step": 14700 + }, + { + "epoch": 1.614430046123435, + "grad_norm": 2.28147292137146, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7090378403663635, + "num_tokens": 373551200.0, + "step": 14701 + }, + { + "epoch": 1.6145398638260486, + "grad_norm": 2.311494827270508, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7187368869781494, + "num_tokens": 373576921.0, + "step": 14702 + }, + { + "epoch": 1.6146496815286624, + "grad_norm": 2.3389205932617188, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7177940011024475, + "num_tokens": 373601235.0, + "step": 14703 + }, + { + "epoch": 1.6147594992312762, + "grad_norm": 2.444275379180908, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7219253182411194, + "num_tokens": 373621836.0, + "step": 14704 + }, + { + "epoch": 1.61486931693389, + "grad_norm": 2.282029867172241, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7100803256034851, + "num_tokens": 373648598.0, + "step": 14705 + }, + { + "epoch": 1.6149791346365034, + "grad_norm": 2.164804458618164, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7234843373298645, + "num_tokens": 373674960.0, + "step": 14706 + }, + { + "epoch": 1.615088952339117, + "grad_norm": 2.387223243713379, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7213595509529114, + "num_tokens": 373696705.0, + "step": 14707 + }, + { + "epoch": 1.6151987700417307, + "grad_norm": 2.363769054412842, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7093435525894165, + "num_tokens": 373720807.0, + "step": 14708 + }, + { + "epoch": 1.6153085877443445, + "grad_norm": 2.0944926738739014, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7355278134346008, + "num_tokens": 373749504.0, + "step": 14709 + }, + { + "epoch": 1.615418405446958, + "grad_norm": 1.9138243198394775, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.7018053531646729, + "num_tokens": 373783067.0, + "step": 14710 + }, + { + "epoch": 1.6155282231495716, + "grad_norm": 2.0092320442199707, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.6944726705551147, + "num_tokens": 373814771.0, + "step": 14711 + }, + { + "epoch": 1.6156380408521853, + "grad_norm": 2.215306282043457, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7154618501663208, + "num_tokens": 373839627.0, + "step": 14712 + }, + { + "epoch": 1.615747858554799, + "grad_norm": 2.2085704803466797, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7107183337211609, + "num_tokens": 373864789.0, + "step": 14713 + }, + { + "epoch": 1.6158576762574128, + "grad_norm": 2.388944387435913, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7214893698692322, + "num_tokens": 373889088.0, + "step": 14714 + }, + { + "epoch": 1.6159674939600264, + "grad_norm": 3.0819172859191895, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7444528341293335, + "num_tokens": 373906391.0, + "step": 14715 + }, + { + "epoch": 1.61607731166264, + "grad_norm": 2.4120090007781982, + "learning_rate": 1e-06, + "loss": 0.9122, + "mean_token_accuracy": 0.7127847671508789, + "num_tokens": 373928245.0, + "step": 14716 + }, + { + "epoch": 1.6161871293652537, + "grad_norm": 2.3347055912017822, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7280318737030029, + "num_tokens": 373950690.0, + "step": 14717 + }, + { + "epoch": 1.6162969470678674, + "grad_norm": 2.2020697593688965, + "learning_rate": 1e-06, + "loss": 0.8194, + "mean_token_accuracy": 0.7363526821136475, + "num_tokens": 373974642.0, + "step": 14718 + }, + { + "epoch": 1.616406764770481, + "grad_norm": 2.1314685344696045, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7139073014259338, + "num_tokens": 374002611.0, + "step": 14719 + }, + { + "epoch": 1.6165165824730947, + "grad_norm": 2.527681350708008, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7184475660324097, + "num_tokens": 374025129.0, + "step": 14720 + }, + { + "epoch": 1.6166264001757082, + "grad_norm": 2.366821765899658, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7102668285369873, + "num_tokens": 374049898.0, + "step": 14721 + }, + { + "epoch": 1.616736217878322, + "grad_norm": 2.395359516143799, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7311537265777588, + "num_tokens": 374071542.0, + "step": 14722 + }, + { + "epoch": 1.6168460355809358, + "grad_norm": 2.111180305480957, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7293990850448608, + "num_tokens": 374100003.0, + "step": 14723 + }, + { + "epoch": 1.6169558532835493, + "grad_norm": 2.0828354358673096, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.709473729133606, + "num_tokens": 374130913.0, + "step": 14724 + }, + { + "epoch": 1.6170656709861628, + "grad_norm": 2.2185122966766357, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6875298023223877, + "num_tokens": 374156937.0, + "step": 14725 + }, + { + "epoch": 1.6171754886887766, + "grad_norm": 1.9059847593307495, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7175523042678833, + "num_tokens": 374189461.0, + "step": 14726 + }, + { + "epoch": 1.6172853063913903, + "grad_norm": 2.346083641052246, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.718630313873291, + "num_tokens": 374214302.0, + "step": 14727 + }, + { + "epoch": 1.617395124094004, + "grad_norm": 2.3756842613220215, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7059975862503052, + "num_tokens": 374238705.0, + "step": 14728 + }, + { + "epoch": 1.6175049417966176, + "grad_norm": 2.021031141281128, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.7143551111221313, + "num_tokens": 374270887.0, + "step": 14729 + }, + { + "epoch": 1.6176147594992312, + "grad_norm": 2.220515727996826, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7059543132781982, + "num_tokens": 374299759.0, + "step": 14730 + }, + { + "epoch": 1.617724577201845, + "grad_norm": 2.1039836406707764, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.6994292736053467, + "num_tokens": 374328463.0, + "step": 14731 + }, + { + "epoch": 1.6178343949044587, + "grad_norm": 2.3224997520446777, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.7338454723358154, + "num_tokens": 374352175.0, + "step": 14732 + }, + { + "epoch": 1.6179442126070722, + "grad_norm": 1.8545571565628052, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.6972576975822449, + "num_tokens": 374390449.0, + "step": 14733 + }, + { + "epoch": 1.618054030309686, + "grad_norm": 2.4916186332702637, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7108098864555359, + "num_tokens": 374414542.0, + "step": 14734 + }, + { + "epoch": 1.6181638480122995, + "grad_norm": 2.3297739028930664, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.722933292388916, + "num_tokens": 374438147.0, + "step": 14735 + }, + { + "epoch": 1.6182736657149133, + "grad_norm": 2.594712018966675, + "learning_rate": 1e-06, + "loss": 0.7985, + "mean_token_accuracy": 0.7456281781196594, + "num_tokens": 374458053.0, + "step": 14736 + }, + { + "epoch": 1.618383483417527, + "grad_norm": 2.5678601264953613, + "learning_rate": 1e-06, + "loss": 1.0406, + "mean_token_accuracy": 0.7161806225776672, + "num_tokens": 374482991.0, + "step": 14737 + }, + { + "epoch": 1.6184933011201406, + "grad_norm": 2.33309006690979, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7202631235122681, + "num_tokens": 374509676.0, + "step": 14738 + }, + { + "epoch": 1.618603118822754, + "grad_norm": 2.19356107711792, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7011367082595825, + "num_tokens": 374536016.0, + "step": 14739 + }, + { + "epoch": 1.6187129365253679, + "grad_norm": 2.456071138381958, + "learning_rate": 1e-06, + "loss": 0.7597, + "mean_token_accuracy": 0.7551910877227783, + "num_tokens": 374556760.0, + "step": 14740 + }, + { + "epoch": 1.6188227542279816, + "grad_norm": 2.4007933139801025, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7226220965385437, + "num_tokens": 374579254.0, + "step": 14741 + }, + { + "epoch": 1.6189325719305954, + "grad_norm": 2.1538639068603516, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.714006781578064, + "num_tokens": 374608170.0, + "step": 14742 + }, + { + "epoch": 1.619042389633209, + "grad_norm": 1.9633151292800903, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7244723439216614, + "num_tokens": 374640429.0, + "step": 14743 + }, + { + "epoch": 1.6191522073358224, + "grad_norm": 2.5192718505859375, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7444864511489868, + "num_tokens": 374660699.0, + "step": 14744 + }, + { + "epoch": 1.6192620250384362, + "grad_norm": 2.2001585960388184, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.730586588382721, + "num_tokens": 374686313.0, + "step": 14745 + }, + { + "epoch": 1.61937184274105, + "grad_norm": 2.1960301399230957, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.715652346611023, + "num_tokens": 374711730.0, + "step": 14746 + }, + { + "epoch": 1.6194816604436635, + "grad_norm": 2.458887815475464, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7240885496139526, + "num_tokens": 374734420.0, + "step": 14747 + }, + { + "epoch": 1.619591478146277, + "grad_norm": 2.1073248386383057, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.6958785057067871, + "num_tokens": 374762556.0, + "step": 14748 + }, + { + "epoch": 1.6197012958488908, + "grad_norm": 2.0687203407287598, + "learning_rate": 1e-06, + "loss": 0.8555, + "mean_token_accuracy": 0.734880805015564, + "num_tokens": 374791693.0, + "step": 14749 + }, + { + "epoch": 1.6198111135515045, + "grad_norm": 2.2942616939544678, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7363971471786499, + "num_tokens": 374815604.0, + "step": 14750 + }, + { + "epoch": 1.6199209312541183, + "grad_norm": 2.227806806564331, + "learning_rate": 1e-06, + "loss": 0.7838, + "mean_token_accuracy": 0.7516905069351196, + "num_tokens": 374837631.0, + "step": 14751 + }, + { + "epoch": 1.6200307489567318, + "grad_norm": 2.090667247772217, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7254804372787476, + "num_tokens": 374865311.0, + "step": 14752 + }, + { + "epoch": 1.6201405666593454, + "grad_norm": 2.3187098503112793, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7344940304756165, + "num_tokens": 374888363.0, + "step": 14753 + }, + { + "epoch": 1.6202503843619591, + "grad_norm": 2.340315341949463, + "learning_rate": 1e-06, + "loss": 0.9032, + "mean_token_accuracy": 0.7385228872299194, + "num_tokens": 374909542.0, + "step": 14754 + }, + { + "epoch": 1.6203602020645729, + "grad_norm": 2.157522201538086, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7247453927993774, + "num_tokens": 374938136.0, + "step": 14755 + }, + { + "epoch": 1.6204700197671866, + "grad_norm": 2.287841320037842, + "learning_rate": 1e-06, + "loss": 0.9009, + "mean_token_accuracy": 0.7184695601463318, + "num_tokens": 374962833.0, + "step": 14756 + }, + { + "epoch": 1.6205798374698002, + "grad_norm": 2.1079158782958984, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.71436607837677, + "num_tokens": 374991555.0, + "step": 14757 + }, + { + "epoch": 1.6206896551724137, + "grad_norm": 2.134281873703003, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7133895754814148, + "num_tokens": 375021194.0, + "step": 14758 + }, + { + "epoch": 1.6207994728750275, + "grad_norm": 2.5181305408477783, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7228275537490845, + "num_tokens": 375041875.0, + "step": 14759 + }, + { + "epoch": 1.6209092905776412, + "grad_norm": 2.1169235706329346, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7364100217819214, + "num_tokens": 375070920.0, + "step": 14760 + }, + { + "epoch": 1.6210191082802548, + "grad_norm": 2.343657970428467, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7161847352981567, + "num_tokens": 375095626.0, + "step": 14761 + }, + { + "epoch": 1.6211289259828683, + "grad_norm": 2.216712713241577, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7220438718795776, + "num_tokens": 375123017.0, + "step": 14762 + }, + { + "epoch": 1.621238743685482, + "grad_norm": 2.053269863128662, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.706701397895813, + "num_tokens": 375155653.0, + "step": 14763 + }, + { + "epoch": 1.6213485613880958, + "grad_norm": 2.22815203666687, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7352321147918701, + "num_tokens": 375180901.0, + "step": 14764 + }, + { + "epoch": 1.6214583790907096, + "grad_norm": 2.1102616786956787, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.722446858882904, + "num_tokens": 375208197.0, + "step": 14765 + }, + { + "epoch": 1.621568196793323, + "grad_norm": 2.4710452556610107, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7106648683547974, + "num_tokens": 375230744.0, + "step": 14766 + }, + { + "epoch": 1.6216780144959366, + "grad_norm": 2.4153530597686768, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7154051661491394, + "num_tokens": 375254050.0, + "step": 14767 + }, + { + "epoch": 1.6217878321985504, + "grad_norm": 2.4815497398376465, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7394385933876038, + "num_tokens": 375277260.0, + "step": 14768 + }, + { + "epoch": 1.6218976499011641, + "grad_norm": 2.2053754329681396, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7146891355514526, + "num_tokens": 375303522.0, + "step": 14769 + }, + { + "epoch": 1.6220074676037777, + "grad_norm": 2.4403436183929443, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7256941199302673, + "num_tokens": 375326205.0, + "step": 14770 + }, + { + "epoch": 1.6221172853063914, + "grad_norm": 2.0532829761505127, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6967065930366516, + "num_tokens": 375357337.0, + "step": 14771 + }, + { + "epoch": 1.622227103009005, + "grad_norm": 2.2065675258636475, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.6915322542190552, + "num_tokens": 375386320.0, + "step": 14772 + }, + { + "epoch": 1.6223369207116187, + "grad_norm": 2.6122360229492188, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7102041244506836, + "num_tokens": 375406983.0, + "step": 14773 + }, + { + "epoch": 1.6224467384142325, + "grad_norm": 2.1894590854644775, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7083992958068848, + "num_tokens": 375433313.0, + "step": 14774 + }, + { + "epoch": 1.622556556116846, + "grad_norm": 2.122481107711792, + "learning_rate": 1e-06, + "loss": 0.8445, + "mean_token_accuracy": 0.7393766045570374, + "num_tokens": 375459985.0, + "step": 14775 + }, + { + "epoch": 1.6226663738194596, + "grad_norm": 2.3253586292266846, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7343162298202515, + "num_tokens": 375482940.0, + "step": 14776 + }, + { + "epoch": 1.6227761915220733, + "grad_norm": 2.3721184730529785, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7258476614952087, + "num_tokens": 375506934.0, + "step": 14777 + }, + { + "epoch": 1.622886009224687, + "grad_norm": 2.2745537757873535, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7001210451126099, + "num_tokens": 375533521.0, + "step": 14778 + }, + { + "epoch": 1.6229958269273008, + "grad_norm": 2.1120870113372803, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.7163454294204712, + "num_tokens": 375563975.0, + "step": 14779 + }, + { + "epoch": 1.6231056446299144, + "grad_norm": 2.2318501472473145, + "learning_rate": 1e-06, + "loss": 0.929, + "mean_token_accuracy": 0.7150042057037354, + "num_tokens": 375590779.0, + "step": 14780 + }, + { + "epoch": 1.623215462332528, + "grad_norm": 2.394805431365967, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7105889320373535, + "num_tokens": 375614373.0, + "step": 14781 + }, + { + "epoch": 1.6233252800351416, + "grad_norm": 2.057194709777832, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7249243259429932, + "num_tokens": 375643083.0, + "step": 14782 + }, + { + "epoch": 1.6234350977377554, + "grad_norm": 2.250883102416992, + "learning_rate": 1e-06, + "loss": 0.9093, + "mean_token_accuracy": 0.7150244116783142, + "num_tokens": 375667328.0, + "step": 14783 + }, + { + "epoch": 1.623544915440369, + "grad_norm": 2.106358766555786, + "learning_rate": 1e-06, + "loss": 0.8882, + "mean_token_accuracy": 0.7273424863815308, + "num_tokens": 375696387.0, + "step": 14784 + }, + { + "epoch": 1.6236547331429827, + "grad_norm": 2.479124069213867, + "learning_rate": 1e-06, + "loss": 0.8716, + "mean_token_accuracy": 0.7253715395927429, + "num_tokens": 375717366.0, + "step": 14785 + }, + { + "epoch": 1.6237645508455962, + "grad_norm": 1.9285998344421387, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7153955698013306, + "num_tokens": 375749235.0, + "step": 14786 + }, + { + "epoch": 1.62387436854821, + "grad_norm": 2.061330556869507, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7278486490249634, + "num_tokens": 375777131.0, + "step": 14787 + }, + { + "epoch": 1.6239841862508237, + "grad_norm": 2.3336594104766846, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7332074642181396, + "num_tokens": 375801018.0, + "step": 14788 + }, + { + "epoch": 1.6240940039534373, + "grad_norm": 2.18768310546875, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7109405994415283, + "num_tokens": 375828493.0, + "step": 14789 + }, + { + "epoch": 1.6242038216560508, + "grad_norm": 2.2864999771118164, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.708991289138794, + "num_tokens": 375855928.0, + "step": 14790 + }, + { + "epoch": 1.6243136393586646, + "grad_norm": 1.935982346534729, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7002496719360352, + "num_tokens": 375890413.0, + "step": 14791 + }, + { + "epoch": 1.6244234570612783, + "grad_norm": 2.0661189556121826, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.6997199654579163, + "num_tokens": 375921562.0, + "step": 14792 + }, + { + "epoch": 1.624533274763892, + "grad_norm": 1.9811066389083862, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7152887582778931, + "num_tokens": 375952191.0, + "step": 14793 + }, + { + "epoch": 1.6246430924665056, + "grad_norm": 2.261702537536621, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7346764206886292, + "num_tokens": 375977238.0, + "step": 14794 + }, + { + "epoch": 1.6247529101691192, + "grad_norm": 2.326904773712158, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7226205468177795, + "num_tokens": 376001095.0, + "step": 14795 + }, + { + "epoch": 1.624862727871733, + "grad_norm": 2.10239839553833, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.6973106265068054, + "num_tokens": 376031352.0, + "step": 14796 + }, + { + "epoch": 1.6249725455743467, + "grad_norm": 2.6114230155944824, + "learning_rate": 1e-06, + "loss": 0.788, + "mean_token_accuracy": 0.753135085105896, + "num_tokens": 376049594.0, + "step": 14797 + }, + { + "epoch": 1.6250823632769602, + "grad_norm": 2.2849643230438232, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7051440477371216, + "num_tokens": 376075364.0, + "step": 14798 + }, + { + "epoch": 1.6251921809795737, + "grad_norm": 2.3275249004364014, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7361080646514893, + "num_tokens": 376099023.0, + "step": 14799 + }, + { + "epoch": 1.6253019986821875, + "grad_norm": 2.257906436920166, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7230334281921387, + "num_tokens": 376125765.0, + "step": 14800 + }, + { + "epoch": 1.6254118163848013, + "grad_norm": 2.062419891357422, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7108173370361328, + "num_tokens": 376156445.0, + "step": 14801 + }, + { + "epoch": 1.625521634087415, + "grad_norm": 2.183490753173828, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7276831865310669, + "num_tokens": 376181666.0, + "step": 14802 + }, + { + "epoch": 1.6256314517900285, + "grad_norm": 2.3242228031158447, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.711580753326416, + "num_tokens": 376206200.0, + "step": 14803 + }, + { + "epoch": 1.625741269492642, + "grad_norm": 2.3388993740081787, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7298091650009155, + "num_tokens": 376228799.0, + "step": 14804 + }, + { + "epoch": 1.6258510871952558, + "grad_norm": 2.212158203125, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7444606423377991, + "num_tokens": 376253827.0, + "step": 14805 + }, + { + "epoch": 1.6259609048978696, + "grad_norm": 1.9820544719696045, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.7047573328018188, + "num_tokens": 376285236.0, + "step": 14806 + }, + { + "epoch": 1.6260707226004834, + "grad_norm": 2.0189261436462402, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7205387949943542, + "num_tokens": 376315651.0, + "step": 14807 + }, + { + "epoch": 1.6261805403030969, + "grad_norm": 2.304844379425049, + "learning_rate": 1e-06, + "loss": 0.9004, + "mean_token_accuracy": 0.7193455696105957, + "num_tokens": 376342323.0, + "step": 14808 + }, + { + "epoch": 1.6262903580057104, + "grad_norm": 2.206310272216797, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7184193730354309, + "num_tokens": 376368607.0, + "step": 14809 + }, + { + "epoch": 1.6264001757083242, + "grad_norm": 2.2252564430236816, + "learning_rate": 1e-06, + "loss": 0.8079, + "mean_token_accuracy": 0.743516206741333, + "num_tokens": 376392552.0, + "step": 14810 + }, + { + "epoch": 1.626509993410938, + "grad_norm": 2.712148427963257, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7114979028701782, + "num_tokens": 376411725.0, + "step": 14811 + }, + { + "epoch": 1.6266198111135515, + "grad_norm": 2.465350389480591, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7165255546569824, + "num_tokens": 376433620.0, + "step": 14812 + }, + { + "epoch": 1.626729628816165, + "grad_norm": 2.368314504623413, + "learning_rate": 1e-06, + "loss": 0.8485, + "mean_token_accuracy": 0.7378912568092346, + "num_tokens": 376457573.0, + "step": 14813 + }, + { + "epoch": 1.6268394465187788, + "grad_norm": 2.104959011077881, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7200080156326294, + "num_tokens": 376487914.0, + "step": 14814 + }, + { + "epoch": 1.6269492642213925, + "grad_norm": 2.300299644470215, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7198302745819092, + "num_tokens": 376511694.0, + "step": 14815 + }, + { + "epoch": 1.6270590819240063, + "grad_norm": 2.1371936798095703, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7252097129821777, + "num_tokens": 376538385.0, + "step": 14816 + }, + { + "epoch": 1.6271688996266198, + "grad_norm": 2.024470806121826, + "learning_rate": 1e-06, + "loss": 0.8826, + "mean_token_accuracy": 0.7189647555351257, + "num_tokens": 376569613.0, + "step": 14817 + }, + { + "epoch": 1.6272787173292333, + "grad_norm": 2.9512548446655273, + "learning_rate": 1e-06, + "loss": 0.8482, + "mean_token_accuracy": 0.7337254881858826, + "num_tokens": 376584849.0, + "step": 14818 + }, + { + "epoch": 1.627388535031847, + "grad_norm": 2.2923529148101807, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7114609479904175, + "num_tokens": 376610401.0, + "step": 14819 + }, + { + "epoch": 1.6274983527344609, + "grad_norm": 2.0242326259613037, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7168899774551392, + "num_tokens": 376639317.0, + "step": 14820 + }, + { + "epoch": 1.6276081704370746, + "grad_norm": 2.089529514312744, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7211325168609619, + "num_tokens": 376667575.0, + "step": 14821 + }, + { + "epoch": 1.6277179881396882, + "grad_norm": 2.156466007232666, + "learning_rate": 1e-06, + "loss": 0.8331, + "mean_token_accuracy": 0.739193320274353, + "num_tokens": 376692965.0, + "step": 14822 + }, + { + "epoch": 1.6278278058423017, + "grad_norm": 2.394700527191162, + "learning_rate": 1e-06, + "loss": 0.8435, + "mean_token_accuracy": 0.7357730269432068, + "num_tokens": 376715424.0, + "step": 14823 + }, + { + "epoch": 1.6279376235449154, + "grad_norm": 2.3742709159851074, + "learning_rate": 1e-06, + "loss": 0.8617, + "mean_token_accuracy": 0.731634795665741, + "num_tokens": 376739370.0, + "step": 14824 + }, + { + "epoch": 1.6280474412475292, + "grad_norm": 2.0730865001678467, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7172908186912537, + "num_tokens": 376769246.0, + "step": 14825 + }, + { + "epoch": 1.6281572589501427, + "grad_norm": 2.2992043495178223, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7045536041259766, + "num_tokens": 376794763.0, + "step": 14826 + }, + { + "epoch": 1.6282670766527563, + "grad_norm": 2.078193426132202, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7160554528236389, + "num_tokens": 376824501.0, + "step": 14827 + }, + { + "epoch": 1.62837689435537, + "grad_norm": 2.2042362689971924, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7175866365432739, + "num_tokens": 376851304.0, + "step": 14828 + }, + { + "epoch": 1.6284867120579838, + "grad_norm": 2.3082196712493896, + "learning_rate": 1e-06, + "loss": 0.9653, + "mean_token_accuracy": 0.701734185218811, + "num_tokens": 376877598.0, + "step": 14829 + }, + { + "epoch": 1.6285965297605975, + "grad_norm": 2.091783285140991, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7038030028343201, + "num_tokens": 376906537.0, + "step": 14830 + }, + { + "epoch": 1.628706347463211, + "grad_norm": 2.1643898487091064, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7221945524215698, + "num_tokens": 376934070.0, + "step": 14831 + }, + { + "epoch": 1.6288161651658246, + "grad_norm": 2.1999881267547607, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7061936855316162, + "num_tokens": 376960255.0, + "step": 14832 + }, + { + "epoch": 1.6289259828684384, + "grad_norm": 2.391260862350464, + "learning_rate": 1e-06, + "loss": 0.871, + "mean_token_accuracy": 0.7279739379882812, + "num_tokens": 376984212.0, + "step": 14833 + }, + { + "epoch": 1.6290358005710521, + "grad_norm": 2.3562264442443848, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.7141884565353394, + "num_tokens": 377006679.0, + "step": 14834 + }, + { + "epoch": 1.6291456182736657, + "grad_norm": 2.466308355331421, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7379069328308105, + "num_tokens": 377029060.0, + "step": 14835 + }, + { + "epoch": 1.6292554359762794, + "grad_norm": 2.140256404876709, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.7021862268447876, + "num_tokens": 377057101.0, + "step": 14836 + }, + { + "epoch": 1.629365253678893, + "grad_norm": 1.8775793313980103, + "learning_rate": 1e-06, + "loss": 1.0056, + "mean_token_accuracy": 0.689716637134552, + "num_tokens": 377093416.0, + "step": 14837 + }, + { + "epoch": 1.6294750713815067, + "grad_norm": 2.741396427154541, + "learning_rate": 1e-06, + "loss": 0.8256, + "mean_token_accuracy": 0.7342368364334106, + "num_tokens": 377110759.0, + "step": 14838 + }, + { + "epoch": 1.6295848890841205, + "grad_norm": 2.4160001277923584, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.6965912580490112, + "num_tokens": 377135353.0, + "step": 14839 + }, + { + "epoch": 1.629694706786734, + "grad_norm": 2.00052547454834, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.6990553140640259, + "num_tokens": 377166940.0, + "step": 14840 + }, + { + "epoch": 1.6298045244893475, + "grad_norm": 2.449408531188965, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7372940182685852, + "num_tokens": 377188404.0, + "step": 14841 + }, + { + "epoch": 1.6299143421919613, + "grad_norm": 2.3247268199920654, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7239497900009155, + "num_tokens": 377211942.0, + "step": 14842 + }, + { + "epoch": 1.630024159894575, + "grad_norm": 2.211582660675049, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.7001513242721558, + "num_tokens": 377238043.0, + "step": 14843 + }, + { + "epoch": 1.6301339775971888, + "grad_norm": 2.261465311050415, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.722920298576355, + "num_tokens": 377263094.0, + "step": 14844 + }, + { + "epoch": 1.6302437952998023, + "grad_norm": 2.2033932209014893, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7173140048980713, + "num_tokens": 377290915.0, + "step": 14845 + }, + { + "epoch": 1.6303536130024159, + "grad_norm": 2.2745513916015625, + "learning_rate": 1e-06, + "loss": 0.9166, + "mean_token_accuracy": 0.7170959711074829, + "num_tokens": 377320127.0, + "step": 14846 + }, + { + "epoch": 1.6304634307050296, + "grad_norm": 2.197892189025879, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7340026497840881, + "num_tokens": 377347558.0, + "step": 14847 + }, + { + "epoch": 1.6305732484076434, + "grad_norm": 2.263906240463257, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7102736234664917, + "num_tokens": 377372698.0, + "step": 14848 + }, + { + "epoch": 1.630683066110257, + "grad_norm": 2.4438202381134033, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7334639430046082, + "num_tokens": 377394018.0, + "step": 14849 + }, + { + "epoch": 1.6307928838128707, + "grad_norm": 2.4304580688476562, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7276653051376343, + "num_tokens": 377419082.0, + "step": 14850 + }, + { + "epoch": 1.6309027015154842, + "grad_norm": 2.355966091156006, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7332708239555359, + "num_tokens": 377442714.0, + "step": 14851 + }, + { + "epoch": 1.631012519218098, + "grad_norm": 2.202423572540283, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7188552618026733, + "num_tokens": 377469820.0, + "step": 14852 + }, + { + "epoch": 1.6311223369207117, + "grad_norm": 2.3027470111846924, + "learning_rate": 1e-06, + "loss": 0.7742, + "mean_token_accuracy": 0.7526979446411133, + "num_tokens": 377491610.0, + "step": 14853 + }, + { + "epoch": 1.6312321546233253, + "grad_norm": 2.613830327987671, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7249013781547546, + "num_tokens": 377511141.0, + "step": 14854 + }, + { + "epoch": 1.6313419723259388, + "grad_norm": 2.008431911468506, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7314566373825073, + "num_tokens": 377541129.0, + "step": 14855 + }, + { + "epoch": 1.6314517900285526, + "grad_norm": 2.0615384578704834, + "learning_rate": 1e-06, + "loss": 0.9835, + "mean_token_accuracy": 0.7166457176208496, + "num_tokens": 377571744.0, + "step": 14856 + }, + { + "epoch": 1.6315616077311663, + "grad_norm": 2.0673675537109375, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7655783891677856, + "num_tokens": 377598150.0, + "step": 14857 + }, + { + "epoch": 1.63167142543378, + "grad_norm": 2.5111117362976074, + "learning_rate": 1e-06, + "loss": 0.9648, + "mean_token_accuracy": 0.7038978338241577, + "num_tokens": 377620468.0, + "step": 14858 + }, + { + "epoch": 1.6317812431363936, + "grad_norm": 2.303544282913208, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7277542948722839, + "num_tokens": 377644739.0, + "step": 14859 + }, + { + "epoch": 1.6318910608390071, + "grad_norm": 2.226386070251465, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.708939790725708, + "num_tokens": 377673338.0, + "step": 14860 + }, + { + "epoch": 1.632000878541621, + "grad_norm": 2.419090509414673, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7116661071777344, + "num_tokens": 377694760.0, + "step": 14861 + }, + { + "epoch": 1.6321106962442347, + "grad_norm": 2.2627298831939697, + "learning_rate": 1e-06, + "loss": 0.7945, + "mean_token_accuracy": 0.7449311017990112, + "num_tokens": 377718710.0, + "step": 14862 + }, + { + "epoch": 1.6322205139468482, + "grad_norm": 2.114698886871338, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7262049913406372, + "num_tokens": 377747571.0, + "step": 14863 + }, + { + "epoch": 1.6323303316494617, + "grad_norm": 2.1346898078918457, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7108758687973022, + "num_tokens": 377774650.0, + "step": 14864 + }, + { + "epoch": 1.6324401493520755, + "grad_norm": 2.3903648853302, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7318114638328552, + "num_tokens": 377796212.0, + "step": 14865 + }, + { + "epoch": 1.6325499670546892, + "grad_norm": 2.2284393310546875, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.6990127563476562, + "num_tokens": 377822026.0, + "step": 14866 + }, + { + "epoch": 1.632659784757303, + "grad_norm": 2.258286476135254, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7258063554763794, + "num_tokens": 377847350.0, + "step": 14867 + }, + { + "epoch": 1.6327696024599165, + "grad_norm": 2.4610595703125, + "learning_rate": 1e-06, + "loss": 0.8117, + "mean_token_accuracy": 0.7398560047149658, + "num_tokens": 377869066.0, + "step": 14868 + }, + { + "epoch": 1.63287942016253, + "grad_norm": 2.098367691040039, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.6987025141716003, + "num_tokens": 377899461.0, + "step": 14869 + }, + { + "epoch": 1.6329892378651438, + "grad_norm": 2.1480391025543213, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7265114784240723, + "num_tokens": 377925581.0, + "step": 14870 + }, + { + "epoch": 1.6330990555677576, + "grad_norm": 2.4437501430511475, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6915954351425171, + "num_tokens": 377951441.0, + "step": 14871 + }, + { + "epoch": 1.6332088732703713, + "grad_norm": 2.4108474254608154, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7244697213172913, + "num_tokens": 377974463.0, + "step": 14872 + }, + { + "epoch": 1.6333186909729849, + "grad_norm": 2.432446241378784, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7071132063865662, + "num_tokens": 378000605.0, + "step": 14873 + }, + { + "epoch": 1.6334285086755984, + "grad_norm": 2.347111463546753, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7327033281326294, + "num_tokens": 378022395.0, + "step": 14874 + }, + { + "epoch": 1.6335383263782122, + "grad_norm": 2.3149702548980713, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7310202121734619, + "num_tokens": 378047402.0, + "step": 14875 + }, + { + "epoch": 1.633648144080826, + "grad_norm": 2.4525504112243652, + "learning_rate": 1e-06, + "loss": 0.7833, + "mean_token_accuracy": 0.7555328011512756, + "num_tokens": 378071978.0, + "step": 14876 + }, + { + "epoch": 1.6337579617834395, + "grad_norm": 2.009636878967285, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7061539888381958, + "num_tokens": 378104601.0, + "step": 14877 + }, + { + "epoch": 1.633867779486053, + "grad_norm": 2.5212666988372803, + "learning_rate": 1e-06, + "loss": 0.8639, + "mean_token_accuracy": 0.7290711402893066, + "num_tokens": 378125067.0, + "step": 14878 + }, + { + "epoch": 1.6339775971886668, + "grad_norm": 2.478959798812866, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.720363974571228, + "num_tokens": 378148534.0, + "step": 14879 + }, + { + "epoch": 1.6340874148912805, + "grad_norm": 2.2230753898620605, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7074533700942993, + "num_tokens": 378175056.0, + "step": 14880 + }, + { + "epoch": 1.6341972325938943, + "grad_norm": 2.5869381427764893, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7045934200286865, + "num_tokens": 378196576.0, + "step": 14881 + }, + { + "epoch": 1.6343070502965078, + "grad_norm": 2.4747486114501953, + "learning_rate": 1e-06, + "loss": 0.8574, + "mean_token_accuracy": 0.7315722703933716, + "num_tokens": 378217577.0, + "step": 14882 + }, + { + "epoch": 1.6344168679991213, + "grad_norm": 2.2878973484039307, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7236464619636536, + "num_tokens": 378243458.0, + "step": 14883 + }, + { + "epoch": 1.634526685701735, + "grad_norm": 2.05968976020813, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7299076318740845, + "num_tokens": 378272827.0, + "step": 14884 + }, + { + "epoch": 1.6346365034043489, + "grad_norm": 2.4395346641540527, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7287821173667908, + "num_tokens": 378296843.0, + "step": 14885 + }, + { + "epoch": 1.6347463211069626, + "grad_norm": 1.9965401887893677, + "learning_rate": 1e-06, + "loss": 0.8804, + "mean_token_accuracy": 0.7314459681510925, + "num_tokens": 378327683.0, + "step": 14886 + }, + { + "epoch": 1.6348561388095761, + "grad_norm": 2.40997052192688, + "learning_rate": 1e-06, + "loss": 0.8444, + "mean_token_accuracy": 0.7287610769271851, + "num_tokens": 378350140.0, + "step": 14887 + }, + { + "epoch": 1.6349659565121897, + "grad_norm": 2.4268856048583984, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7259626388549805, + "num_tokens": 378374432.0, + "step": 14888 + }, + { + "epoch": 1.6350757742148034, + "grad_norm": 2.7138819694519043, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7332192659378052, + "num_tokens": 378395526.0, + "step": 14889 + }, + { + "epoch": 1.6351855919174172, + "grad_norm": 2.8685874938964844, + "learning_rate": 1e-06, + "loss": 0.7982, + "mean_token_accuracy": 0.7490513324737549, + "num_tokens": 378412055.0, + "step": 14890 + }, + { + "epoch": 1.6352954096200307, + "grad_norm": 2.403189182281494, + "learning_rate": 1e-06, + "loss": 0.9707, + "mean_token_accuracy": 0.7008098363876343, + "num_tokens": 378437007.0, + "step": 14891 + }, + { + "epoch": 1.6354052273226443, + "grad_norm": 2.411578416824341, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7257406711578369, + "num_tokens": 378458829.0, + "step": 14892 + }, + { + "epoch": 1.635515045025258, + "grad_norm": 2.1235949993133545, + "learning_rate": 1e-06, + "loss": 0.9278, + "mean_token_accuracy": 0.7122159004211426, + "num_tokens": 378489753.0, + "step": 14893 + }, + { + "epoch": 1.6356248627278718, + "grad_norm": 2.1826772689819336, + "learning_rate": 1e-06, + "loss": 1.004, + "mean_token_accuracy": 0.6882331371307373, + "num_tokens": 378517205.0, + "step": 14894 + }, + { + "epoch": 1.6357346804304855, + "grad_norm": 2.469998836517334, + "learning_rate": 1e-06, + "loss": 0.8876, + "mean_token_accuracy": 0.7228202223777771, + "num_tokens": 378537984.0, + "step": 14895 + }, + { + "epoch": 1.635844498133099, + "grad_norm": 2.3245654106140137, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7363150715827942, + "num_tokens": 378561277.0, + "step": 14896 + }, + { + "epoch": 1.6359543158357126, + "grad_norm": 2.086743116378784, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7129753828048706, + "num_tokens": 378589404.0, + "step": 14897 + }, + { + "epoch": 1.6360641335383264, + "grad_norm": 2.164764881134033, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7318528890609741, + "num_tokens": 378617062.0, + "step": 14898 + }, + { + "epoch": 1.6361739512409401, + "grad_norm": 2.3390262126922607, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7063857316970825, + "num_tokens": 378641397.0, + "step": 14899 + }, + { + "epoch": 1.6362837689435537, + "grad_norm": 2.5675735473632812, + "learning_rate": 1e-06, + "loss": 0.9425, + "mean_token_accuracy": 0.7215749025344849, + "num_tokens": 378662528.0, + "step": 14900 + }, + { + "epoch": 1.6363935866461674, + "grad_norm": 2.8476717472076416, + "learning_rate": 1e-06, + "loss": 0.7664, + "mean_token_accuracy": 0.751758873462677, + "num_tokens": 378679498.0, + "step": 14901 + }, + { + "epoch": 1.636503404348781, + "grad_norm": 2.350954294204712, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7435808777809143, + "num_tokens": 378703420.0, + "step": 14902 + }, + { + "epoch": 1.6366132220513947, + "grad_norm": 2.0362918376922607, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7216392755508423, + "num_tokens": 378732077.0, + "step": 14903 + }, + { + "epoch": 1.6367230397540085, + "grad_norm": 2.2484500408172607, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7195684909820557, + "num_tokens": 378758042.0, + "step": 14904 + }, + { + "epoch": 1.636832857456622, + "grad_norm": 2.3381404876708984, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7203236222267151, + "num_tokens": 378787028.0, + "step": 14905 + }, + { + "epoch": 1.6369426751592355, + "grad_norm": 2.088263511657715, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7190011143684387, + "num_tokens": 378814703.0, + "step": 14906 + }, + { + "epoch": 1.6370524928618493, + "grad_norm": 2.742863655090332, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.730354905128479, + "num_tokens": 378835127.0, + "step": 14907 + }, + { + "epoch": 1.637162310564463, + "grad_norm": 2.027278184890747, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.730391263961792, + "num_tokens": 378863268.0, + "step": 14908 + }, + { + "epoch": 1.6372721282670768, + "grad_norm": 2.092054605484009, + "learning_rate": 1e-06, + "loss": 0.9588, + "mean_token_accuracy": 0.7004369497299194, + "num_tokens": 378893174.0, + "step": 14909 + }, + { + "epoch": 1.6373819459696903, + "grad_norm": 2.2058494091033936, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7256839871406555, + "num_tokens": 378917595.0, + "step": 14910 + }, + { + "epoch": 1.6374917636723039, + "grad_norm": 2.2172915935516357, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.6981073021888733, + "num_tokens": 378943091.0, + "step": 14911 + }, + { + "epoch": 1.6376015813749176, + "grad_norm": 2.3289291858673096, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7374533414840698, + "num_tokens": 378966730.0, + "step": 14912 + }, + { + "epoch": 1.6377113990775314, + "grad_norm": 2.3395626544952393, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.6966387033462524, + "num_tokens": 378990603.0, + "step": 14913 + }, + { + "epoch": 1.637821216780145, + "grad_norm": 2.10255765914917, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7341874837875366, + "num_tokens": 379019555.0, + "step": 14914 + }, + { + "epoch": 1.6379310344827587, + "grad_norm": 2.2447268962860107, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.703607976436615, + "num_tokens": 379045131.0, + "step": 14915 + }, + { + "epoch": 1.6380408521853722, + "grad_norm": 2.5312936305999756, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.712065577507019, + "num_tokens": 379065170.0, + "step": 14916 + }, + { + "epoch": 1.638150669887986, + "grad_norm": 2.273522138595581, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7175800204277039, + "num_tokens": 379090807.0, + "step": 14917 + }, + { + "epoch": 1.6382604875905997, + "grad_norm": 2.496610641479492, + "learning_rate": 1e-06, + "loss": 0.8781, + "mean_token_accuracy": 0.7254467606544495, + "num_tokens": 379112237.0, + "step": 14918 + }, + { + "epoch": 1.6383703052932133, + "grad_norm": 2.48380970954895, + "learning_rate": 1e-06, + "loss": 0.8276, + "mean_token_accuracy": 0.7410674095153809, + "num_tokens": 379132991.0, + "step": 14919 + }, + { + "epoch": 1.6384801229958268, + "grad_norm": 2.264777660369873, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.719265341758728, + "num_tokens": 379158859.0, + "step": 14920 + }, + { + "epoch": 1.6385899406984406, + "grad_norm": 2.229107618331909, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7061631679534912, + "num_tokens": 379186174.0, + "step": 14921 + }, + { + "epoch": 1.6386997584010543, + "grad_norm": 2.2138779163360596, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.7176381349563599, + "num_tokens": 379212599.0, + "step": 14922 + }, + { + "epoch": 1.638809576103668, + "grad_norm": 2.059983968734741, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7399333119392395, + "num_tokens": 379240182.0, + "step": 14923 + }, + { + "epoch": 1.6389193938062816, + "grad_norm": 1.9434894323349, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6878151893615723, + "num_tokens": 379273190.0, + "step": 14924 + }, + { + "epoch": 1.6390292115088951, + "grad_norm": 2.3037421703338623, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7135879993438721, + "num_tokens": 379298507.0, + "step": 14925 + }, + { + "epoch": 1.639139029211509, + "grad_norm": 2.346226930618286, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7184385061264038, + "num_tokens": 379322493.0, + "step": 14926 + }, + { + "epoch": 1.6392488469141226, + "grad_norm": 2.3626182079315186, + "learning_rate": 1e-06, + "loss": 0.7886, + "mean_token_accuracy": 0.7455943822860718, + "num_tokens": 379343610.0, + "step": 14927 + }, + { + "epoch": 1.6393586646167362, + "grad_norm": 2.13268780708313, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.702755331993103, + "num_tokens": 379374261.0, + "step": 14928 + }, + { + "epoch": 1.6394684823193497, + "grad_norm": 2.0865466594696045, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7317277789115906, + "num_tokens": 379402438.0, + "step": 14929 + }, + { + "epoch": 1.6395783000219635, + "grad_norm": 2.0464916229248047, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7215866446495056, + "num_tokens": 379432055.0, + "step": 14930 + }, + { + "epoch": 1.6396881177245772, + "grad_norm": 1.9152228832244873, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.715278685092926, + "num_tokens": 379465242.0, + "step": 14931 + }, + { + "epoch": 1.639797935427191, + "grad_norm": 2.236217498779297, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7372974157333374, + "num_tokens": 379488486.0, + "step": 14932 + }, + { + "epoch": 1.6399077531298045, + "grad_norm": 2.0631988048553467, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7057936191558838, + "num_tokens": 379519035.0, + "step": 14933 + }, + { + "epoch": 1.640017570832418, + "grad_norm": 2.333146095275879, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.708220899105072, + "num_tokens": 379543659.0, + "step": 14934 + }, + { + "epoch": 1.6401273885350318, + "grad_norm": 2.2821755409240723, + "learning_rate": 1e-06, + "loss": 0.8, + "mean_token_accuracy": 0.7433005571365356, + "num_tokens": 379568033.0, + "step": 14935 + }, + { + "epoch": 1.6402372062376456, + "grad_norm": 2.337026357650757, + "learning_rate": 1e-06, + "loss": 0.8851, + "mean_token_accuracy": 0.7221649885177612, + "num_tokens": 379591258.0, + "step": 14936 + }, + { + "epoch": 1.6403470239402593, + "grad_norm": 2.275960922241211, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.738146185874939, + "num_tokens": 379615606.0, + "step": 14937 + }, + { + "epoch": 1.6404568416428729, + "grad_norm": 2.223491907119751, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7163193225860596, + "num_tokens": 379641570.0, + "step": 14938 + }, + { + "epoch": 1.6405666593454864, + "grad_norm": 2.083108425140381, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7134552597999573, + "num_tokens": 379672181.0, + "step": 14939 + }, + { + "epoch": 1.6406764770481002, + "grad_norm": 2.192300796508789, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7143004536628723, + "num_tokens": 379698676.0, + "step": 14940 + }, + { + "epoch": 1.640786294750714, + "grad_norm": 2.3393001556396484, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.715417742729187, + "num_tokens": 379723261.0, + "step": 14941 + }, + { + "epoch": 1.6408961124533274, + "grad_norm": 2.3455209732055664, + "learning_rate": 1e-06, + "loss": 0.8955, + "mean_token_accuracy": 0.7213361859321594, + "num_tokens": 379746538.0, + "step": 14942 + }, + { + "epoch": 1.641005930155941, + "grad_norm": 2.18475604057312, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7280745506286621, + "num_tokens": 379772876.0, + "step": 14943 + }, + { + "epoch": 1.6411157478585547, + "grad_norm": 2.3617115020751953, + "learning_rate": 1e-06, + "loss": 0.9676, + "mean_token_accuracy": 0.7025537490844727, + "num_tokens": 379796708.0, + "step": 14944 + }, + { + "epoch": 1.6412255655611685, + "grad_norm": 2.3018641471862793, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7148444652557373, + "num_tokens": 379822563.0, + "step": 14945 + }, + { + "epoch": 1.6413353832637823, + "grad_norm": 2.5845584869384766, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7238798141479492, + "num_tokens": 379843259.0, + "step": 14946 + }, + { + "epoch": 1.6414452009663958, + "grad_norm": 2.231205940246582, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7264474630355835, + "num_tokens": 379867988.0, + "step": 14947 + }, + { + "epoch": 1.6415550186690093, + "grad_norm": 2.0073330402374268, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.7222410440444946, + "num_tokens": 379897836.0, + "step": 14948 + }, + { + "epoch": 1.641664836371623, + "grad_norm": 2.247107744216919, + "learning_rate": 1e-06, + "loss": 0.9104, + "mean_token_accuracy": 0.7162416577339172, + "num_tokens": 379924537.0, + "step": 14949 + }, + { + "epoch": 1.6417746540742368, + "grad_norm": 2.3847153186798096, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7105892300605774, + "num_tokens": 379947716.0, + "step": 14950 + }, + { + "epoch": 1.6418844717768504, + "grad_norm": 2.4129462242126465, + "learning_rate": 1e-06, + "loss": 0.8168, + "mean_token_accuracy": 0.7471028566360474, + "num_tokens": 379968788.0, + "step": 14951 + }, + { + "epoch": 1.6419942894794641, + "grad_norm": 1.985360026359558, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7039051651954651, + "num_tokens": 380001600.0, + "step": 14952 + }, + { + "epoch": 1.6421041071820777, + "grad_norm": 2.278813362121582, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7132025957107544, + "num_tokens": 380026132.0, + "step": 14953 + }, + { + "epoch": 1.6422139248846914, + "grad_norm": 2.642904281616211, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7278165221214294, + "num_tokens": 380045826.0, + "step": 14954 + }, + { + "epoch": 1.6423237425873052, + "grad_norm": 2.174844264984131, + "learning_rate": 1e-06, + "loss": 0.9873, + "mean_token_accuracy": 0.6940661668777466, + "num_tokens": 380074547.0, + "step": 14955 + }, + { + "epoch": 1.6424335602899187, + "grad_norm": 2.230898857116699, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7235077619552612, + "num_tokens": 380101148.0, + "step": 14956 + }, + { + "epoch": 1.6425433779925322, + "grad_norm": 2.891526222229004, + "learning_rate": 1e-06, + "loss": 0.8411, + "mean_token_accuracy": 0.7368520498275757, + "num_tokens": 380117529.0, + "step": 14957 + }, + { + "epoch": 1.642653195695146, + "grad_norm": 2.5209944248199463, + "learning_rate": 1e-06, + "loss": 0.8526, + "mean_token_accuracy": 0.7361210584640503, + "num_tokens": 380136951.0, + "step": 14958 + }, + { + "epoch": 1.6427630133977598, + "grad_norm": 2.2061445713043213, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7328025102615356, + "num_tokens": 380163707.0, + "step": 14959 + }, + { + "epoch": 1.6428728311003735, + "grad_norm": 2.2550859451293945, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7118987441062927, + "num_tokens": 380192289.0, + "step": 14960 + }, + { + "epoch": 1.642982648802987, + "grad_norm": 2.594863176345825, + "learning_rate": 1e-06, + "loss": 0.8564, + "mean_token_accuracy": 0.7379327416419983, + "num_tokens": 380211456.0, + "step": 14961 + }, + { + "epoch": 1.6430924665056006, + "grad_norm": 2.233010768890381, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7178120613098145, + "num_tokens": 380235842.0, + "step": 14962 + }, + { + "epoch": 1.6432022842082143, + "grad_norm": 2.2474050521850586, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7169517278671265, + "num_tokens": 380261839.0, + "step": 14963 + }, + { + "epoch": 1.643312101910828, + "grad_norm": 2.0730984210968018, + "learning_rate": 1e-06, + "loss": 1.0266, + "mean_token_accuracy": 0.686027467250824, + "num_tokens": 380293387.0, + "step": 14964 + }, + { + "epoch": 1.6434219196134416, + "grad_norm": 2.3036885261535645, + "learning_rate": 1e-06, + "loss": 0.9933, + "mean_token_accuracy": 0.7014374136924744, + "num_tokens": 380318937.0, + "step": 14965 + }, + { + "epoch": 1.6435317373160554, + "grad_norm": 2.144463300704956, + "learning_rate": 1e-06, + "loss": 0.952, + "mean_token_accuracy": 0.7083869576454163, + "num_tokens": 380345388.0, + "step": 14966 + }, + { + "epoch": 1.643641555018669, + "grad_norm": 2.2802538871765137, + "learning_rate": 1e-06, + "loss": 0.8096, + "mean_token_accuracy": 0.7411525249481201, + "num_tokens": 380369190.0, + "step": 14967 + }, + { + "epoch": 1.6437513727212827, + "grad_norm": 2.5474629402160645, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7354717254638672, + "num_tokens": 380389432.0, + "step": 14968 + }, + { + "epoch": 1.6438611904238964, + "grad_norm": 2.046752452850342, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7200272679328918, + "num_tokens": 380422048.0, + "step": 14969 + }, + { + "epoch": 1.64397100812651, + "grad_norm": 2.3123362064361572, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7287876605987549, + "num_tokens": 380445018.0, + "step": 14970 + }, + { + "epoch": 1.6440808258291235, + "grad_norm": 2.2165627479553223, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7039743065834045, + "num_tokens": 380474200.0, + "step": 14971 + }, + { + "epoch": 1.6441906435317373, + "grad_norm": 2.2517778873443604, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7254841923713684, + "num_tokens": 380499965.0, + "step": 14972 + }, + { + "epoch": 1.644300461234351, + "grad_norm": 2.4858217239379883, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7415288686752319, + "num_tokens": 380520981.0, + "step": 14973 + }, + { + "epoch": 1.6444102789369648, + "grad_norm": 2.167013645172119, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7206404209136963, + "num_tokens": 380547726.0, + "step": 14974 + }, + { + "epoch": 1.6445200966395783, + "grad_norm": 2.283843994140625, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7415215969085693, + "num_tokens": 380573463.0, + "step": 14975 + }, + { + "epoch": 1.6446299143421919, + "grad_norm": 2.632509231567383, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7227135896682739, + "num_tokens": 380593232.0, + "step": 14976 + }, + { + "epoch": 1.6447397320448056, + "grad_norm": 2.279210090637207, + "learning_rate": 1e-06, + "loss": 0.9268, + "mean_token_accuracy": 0.7054698467254639, + "num_tokens": 380618729.0, + "step": 14977 + }, + { + "epoch": 1.6448495497474194, + "grad_norm": 2.234863758087158, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7145500779151917, + "num_tokens": 380644763.0, + "step": 14978 + }, + { + "epoch": 1.644959367450033, + "grad_norm": 2.007523775100708, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7071411609649658, + "num_tokens": 380677222.0, + "step": 14979 + }, + { + "epoch": 1.6450691851526464, + "grad_norm": 2.2613649368286133, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7357123494148254, + "num_tokens": 380702118.0, + "step": 14980 + }, + { + "epoch": 1.6451790028552602, + "grad_norm": 2.453795909881592, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7172656059265137, + "num_tokens": 380725833.0, + "step": 14981 + }, + { + "epoch": 1.645288820557874, + "grad_norm": 2.239658832550049, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7225697636604309, + "num_tokens": 380752369.0, + "step": 14982 + }, + { + "epoch": 1.6453986382604877, + "grad_norm": 2.27893328666687, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7208787202835083, + "num_tokens": 380777988.0, + "step": 14983 + }, + { + "epoch": 1.6455084559631012, + "grad_norm": 2.1660304069519043, + "learning_rate": 1e-06, + "loss": 1.0073, + "mean_token_accuracy": 0.6930645108222961, + "num_tokens": 380807860.0, + "step": 14984 + }, + { + "epoch": 1.6456182736657148, + "grad_norm": 2.3976521492004395, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.738838791847229, + "num_tokens": 380831345.0, + "step": 14985 + }, + { + "epoch": 1.6457280913683285, + "grad_norm": 1.9434380531311035, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7019718289375305, + "num_tokens": 380864011.0, + "step": 14986 + }, + { + "epoch": 1.6458379090709423, + "grad_norm": 2.2250750064849854, + "learning_rate": 1e-06, + "loss": 1.0199, + "mean_token_accuracy": 0.6891970038414001, + "num_tokens": 380891195.0, + "step": 14987 + }, + { + "epoch": 1.645947726773556, + "grad_norm": 2.239309549331665, + "learning_rate": 1e-06, + "loss": 0.8872, + "mean_token_accuracy": 0.7220586538314819, + "num_tokens": 380916652.0, + "step": 14988 + }, + { + "epoch": 1.6460575444761696, + "grad_norm": 2.560976028442383, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7259643077850342, + "num_tokens": 380939038.0, + "step": 14989 + }, + { + "epoch": 1.6461673621787831, + "grad_norm": 2.4128715991973877, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.6998594999313354, + "num_tokens": 380963330.0, + "step": 14990 + }, + { + "epoch": 1.6462771798813969, + "grad_norm": 2.1899709701538086, + "learning_rate": 1e-06, + "loss": 0.7906, + "mean_token_accuracy": 0.751274824142456, + "num_tokens": 380987463.0, + "step": 14991 + }, + { + "epoch": 1.6463869975840106, + "grad_norm": 2.2184488773345947, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7126981019973755, + "num_tokens": 381016514.0, + "step": 14992 + }, + { + "epoch": 1.6464968152866242, + "grad_norm": 2.429652452468872, + "learning_rate": 1e-06, + "loss": 0.8014, + "mean_token_accuracy": 0.7419477105140686, + "num_tokens": 381037635.0, + "step": 14993 + }, + { + "epoch": 1.6466066329892377, + "grad_norm": 2.3546109199523926, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7219456434249878, + "num_tokens": 381062847.0, + "step": 14994 + }, + { + "epoch": 1.6467164506918515, + "grad_norm": 2.011950731277466, + "learning_rate": 1e-06, + "loss": 0.9689, + "mean_token_accuracy": 0.702201247215271, + "num_tokens": 381092830.0, + "step": 14995 + }, + { + "epoch": 1.6468262683944652, + "grad_norm": 2.4307117462158203, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7189061641693115, + "num_tokens": 381114612.0, + "step": 14996 + }, + { + "epoch": 1.646936086097079, + "grad_norm": 2.231748104095459, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7040594816207886, + "num_tokens": 381142267.0, + "step": 14997 + }, + { + "epoch": 1.6470459037996925, + "grad_norm": 2.17854642868042, + "learning_rate": 1e-06, + "loss": 0.9967, + "mean_token_accuracy": 0.6984118819236755, + "num_tokens": 381169293.0, + "step": 14998 + }, + { + "epoch": 1.647155721502306, + "grad_norm": 2.3051071166992188, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7281349897384644, + "num_tokens": 381195158.0, + "step": 14999 + }, + { + "epoch": 1.6472655392049198, + "grad_norm": 2.4874162673950195, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7038029432296753, + "num_tokens": 381218260.0, + "step": 15000 + }, + { + "epoch": 1.6473753569075336, + "grad_norm": 2.1536684036254883, + "learning_rate": 1e-06, + "loss": 0.7308, + "mean_token_accuracy": 0.7633306980133057, + "num_tokens": 381241286.0, + "step": 15001 + }, + { + "epoch": 1.6474851746101473, + "grad_norm": 2.1309828758239746, + "learning_rate": 1e-06, + "loss": 0.9885, + "mean_token_accuracy": 0.6906818747520447, + "num_tokens": 381268487.0, + "step": 15002 + }, + { + "epoch": 1.6475949923127609, + "grad_norm": 2.2875306606292725, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7111941576004028, + "num_tokens": 381293489.0, + "step": 15003 + }, + { + "epoch": 1.6477048100153744, + "grad_norm": 2.5157485008239746, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.7135686874389648, + "num_tokens": 381316206.0, + "step": 15004 + }, + { + "epoch": 1.6478146277179881, + "grad_norm": 2.216453790664673, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.716951310634613, + "num_tokens": 381342748.0, + "step": 15005 + }, + { + "epoch": 1.647924445420602, + "grad_norm": 2.310206890106201, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7312413454055786, + "num_tokens": 381366836.0, + "step": 15006 + }, + { + "epoch": 1.6480342631232154, + "grad_norm": 2.2773170471191406, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7294889688491821, + "num_tokens": 381390945.0, + "step": 15007 + }, + { + "epoch": 1.648144080825829, + "grad_norm": 2.397801160812378, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7223613262176514, + "num_tokens": 381414898.0, + "step": 15008 + }, + { + "epoch": 1.6482538985284427, + "grad_norm": 2.1301345825195312, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7218663692474365, + "num_tokens": 381443862.0, + "step": 15009 + }, + { + "epoch": 1.6483637162310565, + "grad_norm": 2.10550856590271, + "learning_rate": 1e-06, + "loss": 1.0187, + "mean_token_accuracy": 0.6928300261497498, + "num_tokens": 381474240.0, + "step": 15010 + }, + { + "epoch": 1.6484735339336702, + "grad_norm": 2.5035133361816406, + "learning_rate": 1e-06, + "loss": 0.9685, + "mean_token_accuracy": 0.7067687511444092, + "num_tokens": 381496467.0, + "step": 15011 + }, + { + "epoch": 1.6485833516362838, + "grad_norm": 2.00531005859375, + "learning_rate": 1e-06, + "loss": 1.0046, + "mean_token_accuracy": 0.6922281980514526, + "num_tokens": 381529530.0, + "step": 15012 + }, + { + "epoch": 1.6486931693388973, + "grad_norm": 2.0249571800231934, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.723011314868927, + "num_tokens": 381560073.0, + "step": 15013 + }, + { + "epoch": 1.648802987041511, + "grad_norm": 2.256218910217285, + "learning_rate": 1e-06, + "loss": 0.9515, + "mean_token_accuracy": 0.7098665237426758, + "num_tokens": 381586531.0, + "step": 15014 + }, + { + "epoch": 1.6489128047441248, + "grad_norm": 2.252406358718872, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7357977032661438, + "num_tokens": 381612680.0, + "step": 15015 + }, + { + "epoch": 1.6490226224467384, + "grad_norm": 2.323359966278076, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7285544872283936, + "num_tokens": 381637410.0, + "step": 15016 + }, + { + "epoch": 1.6491324401493521, + "grad_norm": 2.2118499279022217, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7173029184341431, + "num_tokens": 381663793.0, + "step": 15017 + }, + { + "epoch": 1.6492422578519657, + "grad_norm": 2.205460548400879, + "learning_rate": 1e-06, + "loss": 0.7739, + "mean_token_accuracy": 0.754689633846283, + "num_tokens": 381688301.0, + "step": 15018 + }, + { + "epoch": 1.6493520755545794, + "grad_norm": 2.003504514694214, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7156977653503418, + "num_tokens": 381718124.0, + "step": 15019 + }, + { + "epoch": 1.6494618932571932, + "grad_norm": 2.1506505012512207, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7057616114616394, + "num_tokens": 381747274.0, + "step": 15020 + }, + { + "epoch": 1.6495717109598067, + "grad_norm": 2.1016979217529297, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7200695276260376, + "num_tokens": 381777725.0, + "step": 15021 + }, + { + "epoch": 1.6496815286624202, + "grad_norm": 1.8195159435272217, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.7008483409881592, + "num_tokens": 381814875.0, + "step": 15022 + }, + { + "epoch": 1.649791346365034, + "grad_norm": 2.470362663269043, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7168687582015991, + "num_tokens": 381838264.0, + "step": 15023 + }, + { + "epoch": 1.6499011640676478, + "grad_norm": 2.163191080093384, + "learning_rate": 1e-06, + "loss": 0.8233, + "mean_token_accuracy": 0.7336431741714478, + "num_tokens": 381864413.0, + "step": 15024 + }, + { + "epoch": 1.6500109817702615, + "grad_norm": 2.2001354694366455, + "learning_rate": 1e-06, + "loss": 0.9586, + "mean_token_accuracy": 0.7103772163391113, + "num_tokens": 381891034.0, + "step": 15025 + }, + { + "epoch": 1.650120799472875, + "grad_norm": 2.2277746200561523, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.7190357446670532, + "num_tokens": 381918740.0, + "step": 15026 + }, + { + "epoch": 1.6502306171754886, + "grad_norm": 2.228767156600952, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7031652927398682, + "num_tokens": 381944931.0, + "step": 15027 + }, + { + "epoch": 1.6503404348781023, + "grad_norm": 2.0178728103637695, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7073055505752563, + "num_tokens": 381977170.0, + "step": 15028 + }, + { + "epoch": 1.650450252580716, + "grad_norm": 2.3491854667663574, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.70074462890625, + "num_tokens": 382002901.0, + "step": 15029 + }, + { + "epoch": 1.6505600702833296, + "grad_norm": 2.233412742614746, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7516482472419739, + "num_tokens": 382027472.0, + "step": 15030 + }, + { + "epoch": 1.6506698879859434, + "grad_norm": 2.210808038711548, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7134470343589783, + "num_tokens": 382053526.0, + "step": 15031 + }, + { + "epoch": 1.650779705688557, + "grad_norm": 2.2027902603149414, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7174527645111084, + "num_tokens": 382080740.0, + "step": 15032 + }, + { + "epoch": 1.6508895233911707, + "grad_norm": 2.13778018951416, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7215006351470947, + "num_tokens": 382106239.0, + "step": 15033 + }, + { + "epoch": 1.6509993410937844, + "grad_norm": 2.3299593925476074, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7266180515289307, + "num_tokens": 382133914.0, + "step": 15034 + }, + { + "epoch": 1.651109158796398, + "grad_norm": 2.2881836891174316, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7315632104873657, + "num_tokens": 382158591.0, + "step": 15035 + }, + { + "epoch": 1.6512189764990115, + "grad_norm": 2.2947640419006348, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7212334275245667, + "num_tokens": 382183288.0, + "step": 15036 + }, + { + "epoch": 1.6513287942016253, + "grad_norm": 2.301522970199585, + "learning_rate": 1e-06, + "loss": 0.8588, + "mean_token_accuracy": 0.7385817766189575, + "num_tokens": 382206059.0, + "step": 15037 + }, + { + "epoch": 1.651438611904239, + "grad_norm": 2.3532965183258057, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7206231951713562, + "num_tokens": 382230798.0, + "step": 15038 + }, + { + "epoch": 1.6515484296068528, + "grad_norm": 2.441991090774536, + "learning_rate": 1e-06, + "loss": 0.9866, + "mean_token_accuracy": 0.6958977580070496, + "num_tokens": 382255407.0, + "step": 15039 + }, + { + "epoch": 1.6516582473094663, + "grad_norm": 2.257905960083008, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7220735549926758, + "num_tokens": 382281104.0, + "step": 15040 + }, + { + "epoch": 1.6517680650120798, + "grad_norm": 2.130929470062256, + "learning_rate": 1e-06, + "loss": 0.9257, + "mean_token_accuracy": 0.7155880331993103, + "num_tokens": 382309009.0, + "step": 15041 + }, + { + "epoch": 1.6518778827146936, + "grad_norm": 2.100743532180786, + "learning_rate": 1e-06, + "loss": 0.9269, + "mean_token_accuracy": 0.7177315354347229, + "num_tokens": 382337952.0, + "step": 15042 + }, + { + "epoch": 1.6519877004173074, + "grad_norm": 1.9612960815429688, + "learning_rate": 1e-06, + "loss": 0.9927, + "mean_token_accuracy": 0.6922764778137207, + "num_tokens": 382368788.0, + "step": 15043 + }, + { + "epoch": 1.652097518119921, + "grad_norm": 2.065788984298706, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.712110161781311, + "num_tokens": 382399111.0, + "step": 15044 + }, + { + "epoch": 1.6522073358225344, + "grad_norm": 2.436673164367676, + "learning_rate": 1e-06, + "loss": 0.9446, + "mean_token_accuracy": 0.7112576961517334, + "num_tokens": 382421212.0, + "step": 15045 + }, + { + "epoch": 1.6523171535251482, + "grad_norm": 2.5133135318756104, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7171326279640198, + "num_tokens": 382443341.0, + "step": 15046 + }, + { + "epoch": 1.652426971227762, + "grad_norm": 2.3565187454223633, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7174159288406372, + "num_tokens": 382466596.0, + "step": 15047 + }, + { + "epoch": 1.6525367889303757, + "grad_norm": 2.1599597930908203, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7222948670387268, + "num_tokens": 382494075.0, + "step": 15048 + }, + { + "epoch": 1.6526466066329892, + "grad_norm": 2.186546564102173, + "learning_rate": 1e-06, + "loss": 0.9768, + "mean_token_accuracy": 0.7026199102401733, + "num_tokens": 382520737.0, + "step": 15049 + }, + { + "epoch": 1.6527564243356028, + "grad_norm": 2.25728440284729, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7271945476531982, + "num_tokens": 382546229.0, + "step": 15050 + }, + { + "epoch": 1.6528662420382165, + "grad_norm": 2.569143295288086, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7230274677276611, + "num_tokens": 382566144.0, + "step": 15051 + }, + { + "epoch": 1.6529760597408303, + "grad_norm": 2.363750457763672, + "learning_rate": 1e-06, + "loss": 0.8541, + "mean_token_accuracy": 0.7297058701515198, + "num_tokens": 382587320.0, + "step": 15052 + }, + { + "epoch": 1.653085877443444, + "grad_norm": 2.3391928672790527, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7255966067314148, + "num_tokens": 382611672.0, + "step": 15053 + }, + { + "epoch": 1.6531956951460576, + "grad_norm": 2.4206206798553467, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7083587646484375, + "num_tokens": 382633916.0, + "step": 15054 + }, + { + "epoch": 1.653305512848671, + "grad_norm": 2.654036521911621, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7357438206672668, + "num_tokens": 382652625.0, + "step": 15055 + }, + { + "epoch": 1.6534153305512849, + "grad_norm": 2.311884641647339, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.6993507146835327, + "num_tokens": 382677243.0, + "step": 15056 + }, + { + "epoch": 1.6535251482538986, + "grad_norm": 2.138314723968506, + "learning_rate": 1e-06, + "loss": 0.7748, + "mean_token_accuracy": 0.7497297525405884, + "num_tokens": 382701678.0, + "step": 15057 + }, + { + "epoch": 1.6536349659565122, + "grad_norm": 2.3712615966796875, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7262555360794067, + "num_tokens": 382725198.0, + "step": 15058 + }, + { + "epoch": 1.6537447836591257, + "grad_norm": 2.1548774242401123, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7174239158630371, + "num_tokens": 382750817.0, + "step": 15059 + }, + { + "epoch": 1.6538546013617395, + "grad_norm": 2.506148338317871, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7308121919631958, + "num_tokens": 382771057.0, + "step": 15060 + }, + { + "epoch": 1.6539644190643532, + "grad_norm": 2.264665126800537, + "learning_rate": 1e-06, + "loss": 0.8073, + "mean_token_accuracy": 0.7494773864746094, + "num_tokens": 382793016.0, + "step": 15061 + }, + { + "epoch": 1.654074236766967, + "grad_norm": 2.402193069458008, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7071706652641296, + "num_tokens": 382816804.0, + "step": 15062 + }, + { + "epoch": 1.6541840544695805, + "grad_norm": 2.077230215072632, + "learning_rate": 1e-06, + "loss": 0.9456, + "mean_token_accuracy": 0.7026058435440063, + "num_tokens": 382846231.0, + "step": 15063 + }, + { + "epoch": 1.654293872172194, + "grad_norm": 2.119908094406128, + "learning_rate": 1e-06, + "loss": 0.9214, + "mean_token_accuracy": 0.7161632180213928, + "num_tokens": 382874632.0, + "step": 15064 + }, + { + "epoch": 1.6544036898748078, + "grad_norm": 2.3567330837249756, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7265017032623291, + "num_tokens": 382899723.0, + "step": 15065 + }, + { + "epoch": 1.6545135075774215, + "grad_norm": 2.3029024600982666, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7254998683929443, + "num_tokens": 382924680.0, + "step": 15066 + }, + { + "epoch": 1.6546233252800353, + "grad_norm": 2.337585687637329, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7270059585571289, + "num_tokens": 382950241.0, + "step": 15067 + }, + { + "epoch": 1.6547331429826488, + "grad_norm": 2.441359519958496, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7376067638397217, + "num_tokens": 382973564.0, + "step": 15068 + }, + { + "epoch": 1.6548429606852624, + "grad_norm": 2.571138381958008, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7192911505699158, + "num_tokens": 382992760.0, + "step": 15069 + }, + { + "epoch": 1.6549527783878761, + "grad_norm": 2.4172046184539795, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7249906659126282, + "num_tokens": 383016583.0, + "step": 15070 + }, + { + "epoch": 1.65506259609049, + "grad_norm": 1.9895790815353394, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7275627255439758, + "num_tokens": 383046241.0, + "step": 15071 + }, + { + "epoch": 1.6551724137931034, + "grad_norm": 2.178544759750366, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7210218906402588, + "num_tokens": 383072844.0, + "step": 15072 + }, + { + "epoch": 1.655282231495717, + "grad_norm": 2.5370278358459473, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7210410833358765, + "num_tokens": 383093012.0, + "step": 15073 + }, + { + "epoch": 1.6553920491983307, + "grad_norm": 2.800581216812134, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7238723039627075, + "num_tokens": 383110116.0, + "step": 15074 + }, + { + "epoch": 1.6555018669009445, + "grad_norm": 2.028695583343506, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7034140825271606, + "num_tokens": 383140379.0, + "step": 15075 + }, + { + "epoch": 1.6556116846035582, + "grad_norm": 2.0052597522735596, + "learning_rate": 1e-06, + "loss": 0.9037, + "mean_token_accuracy": 0.7250857353210449, + "num_tokens": 383171014.0, + "step": 15076 + }, + { + "epoch": 1.6557215023061718, + "grad_norm": 2.384464979171753, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.728402853012085, + "num_tokens": 383194935.0, + "step": 15077 + }, + { + "epoch": 1.6558313200087853, + "grad_norm": 2.4736263751983643, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7184162139892578, + "num_tokens": 383217169.0, + "step": 15078 + }, + { + "epoch": 1.655941137711399, + "grad_norm": 2.621544599533081, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7190756797790527, + "num_tokens": 383239664.0, + "step": 15079 + }, + { + "epoch": 1.6560509554140128, + "grad_norm": 2.1738414764404297, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7270721197128296, + "num_tokens": 383268965.0, + "step": 15080 + }, + { + "epoch": 1.6561607731166264, + "grad_norm": 2.447486162185669, + "learning_rate": 1e-06, + "loss": 0.9564, + "mean_token_accuracy": 0.7048800587654114, + "num_tokens": 383292033.0, + "step": 15081 + }, + { + "epoch": 1.65627059081924, + "grad_norm": 2.044358491897583, + "learning_rate": 1e-06, + "loss": 0.9608, + "mean_token_accuracy": 0.7037776708602905, + "num_tokens": 383322448.0, + "step": 15082 + }, + { + "epoch": 1.6563804085218536, + "grad_norm": 2.1266844272613525, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7164942026138306, + "num_tokens": 383349218.0, + "step": 15083 + }, + { + "epoch": 1.6564902262244674, + "grad_norm": 2.4786431789398193, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7349317073822021, + "num_tokens": 383369892.0, + "step": 15084 + }, + { + "epoch": 1.6566000439270812, + "grad_norm": 2.2437455654144287, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7317186594009399, + "num_tokens": 383396806.0, + "step": 15085 + }, + { + "epoch": 1.6567098616296947, + "grad_norm": 1.9876232147216797, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7216933965682983, + "num_tokens": 383425774.0, + "step": 15086 + }, + { + "epoch": 1.6568196793323082, + "grad_norm": 1.9975775480270386, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7231550216674805, + "num_tokens": 383456325.0, + "step": 15087 + }, + { + "epoch": 1.656929497034922, + "grad_norm": 2.422657012939453, + "learning_rate": 1e-06, + "loss": 0.7968, + "mean_token_accuracy": 0.7423018217086792, + "num_tokens": 383477552.0, + "step": 15088 + }, + { + "epoch": 1.6570393147375357, + "grad_norm": 2.271803617477417, + "learning_rate": 1e-06, + "loss": 0.8052, + "mean_token_accuracy": 0.7458725571632385, + "num_tokens": 383502555.0, + "step": 15089 + }, + { + "epoch": 1.6571491324401495, + "grad_norm": 2.2162973880767822, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7211966514587402, + "num_tokens": 383529229.0, + "step": 15090 + }, + { + "epoch": 1.657258950142763, + "grad_norm": 1.981894612312317, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7314661741256714, + "num_tokens": 383558667.0, + "step": 15091 + }, + { + "epoch": 1.6573687678453766, + "grad_norm": 2.189265489578247, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.705318808555603, + "num_tokens": 383586426.0, + "step": 15092 + }, + { + "epoch": 1.6574785855479903, + "grad_norm": 2.423654079437256, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7405279874801636, + "num_tokens": 383607818.0, + "step": 15093 + }, + { + "epoch": 1.657588403250604, + "grad_norm": 2.369814872741699, + "learning_rate": 1e-06, + "loss": 0.8208, + "mean_token_accuracy": 0.7412341833114624, + "num_tokens": 383631674.0, + "step": 15094 + }, + { + "epoch": 1.6576982209532176, + "grad_norm": 2.5127110481262207, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7294047474861145, + "num_tokens": 383653422.0, + "step": 15095 + }, + { + "epoch": 1.6578080386558314, + "grad_norm": 2.2268142700195312, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7118769884109497, + "num_tokens": 383682355.0, + "step": 15096 + }, + { + "epoch": 1.657917856358445, + "grad_norm": 2.2834627628326416, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7286702394485474, + "num_tokens": 383707429.0, + "step": 15097 + }, + { + "epoch": 1.6580276740610587, + "grad_norm": 2.0791733264923096, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7139796018600464, + "num_tokens": 383737170.0, + "step": 15098 + }, + { + "epoch": 1.6581374917636724, + "grad_norm": 2.12284517288208, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7267230749130249, + "num_tokens": 383765123.0, + "step": 15099 + }, + { + "epoch": 1.658247309466286, + "grad_norm": 2.156789779663086, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.739521861076355, + "num_tokens": 383790206.0, + "step": 15100 + }, + { + "epoch": 1.6583571271688995, + "grad_norm": 2.7218377590179443, + "learning_rate": 1e-06, + "loss": 0.8045, + "mean_token_accuracy": 0.7412222623825073, + "num_tokens": 383808753.0, + "step": 15101 + }, + { + "epoch": 1.6584669448715132, + "grad_norm": 2.21806263923645, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7310073375701904, + "num_tokens": 383832850.0, + "step": 15102 + }, + { + "epoch": 1.658576762574127, + "grad_norm": 2.2062652111053467, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7183143496513367, + "num_tokens": 383858462.0, + "step": 15103 + }, + { + "epoch": 1.6586865802767408, + "grad_norm": 2.51322865486145, + "learning_rate": 1e-06, + "loss": 0.7866, + "mean_token_accuracy": 0.751529335975647, + "num_tokens": 383877643.0, + "step": 15104 + }, + { + "epoch": 1.6587963979793543, + "grad_norm": 2.297656536102295, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7168149948120117, + "num_tokens": 383903822.0, + "step": 15105 + }, + { + "epoch": 1.6589062156819678, + "grad_norm": 2.3698933124542236, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7016452550888062, + "num_tokens": 383926737.0, + "step": 15106 + }, + { + "epoch": 1.6590160333845816, + "grad_norm": 2.0844762325286865, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.6970346570014954, + "num_tokens": 383956448.0, + "step": 15107 + }, + { + "epoch": 1.6591258510871953, + "grad_norm": 2.1774518489837646, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7248741388320923, + "num_tokens": 383982253.0, + "step": 15108 + }, + { + "epoch": 1.6592356687898089, + "grad_norm": 2.2337656021118164, + "learning_rate": 1e-06, + "loss": 0.7599, + "mean_token_accuracy": 0.7586895227432251, + "num_tokens": 384004634.0, + "step": 15109 + }, + { + "epoch": 1.6593454864924224, + "grad_norm": 2.351945400238037, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7219885587692261, + "num_tokens": 384026434.0, + "step": 15110 + }, + { + "epoch": 1.6594553041950362, + "grad_norm": 2.1185991764068604, + "learning_rate": 1e-06, + "loss": 0.9559, + "mean_token_accuracy": 0.7020922303199768, + "num_tokens": 384053272.0, + "step": 15111 + }, + { + "epoch": 1.65956512189765, + "grad_norm": 1.976990818977356, + "learning_rate": 1e-06, + "loss": 1.0039, + "mean_token_accuracy": 0.6976989507675171, + "num_tokens": 384084606.0, + "step": 15112 + }, + { + "epoch": 1.6596749396002637, + "grad_norm": 2.329415798187256, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.740058958530426, + "num_tokens": 384105386.0, + "step": 15113 + }, + { + "epoch": 1.6597847573028772, + "grad_norm": 2.318081855773926, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7221457958221436, + "num_tokens": 384129943.0, + "step": 15114 + }, + { + "epoch": 1.6598945750054908, + "grad_norm": 2.201373338699341, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7014809250831604, + "num_tokens": 384156167.0, + "step": 15115 + }, + { + "epoch": 1.6600043927081045, + "grad_norm": 2.2384235858917236, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7148410081863403, + "num_tokens": 384183591.0, + "step": 15116 + }, + { + "epoch": 1.6601142104107183, + "grad_norm": 2.1730310916900635, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7234610319137573, + "num_tokens": 384210438.0, + "step": 15117 + }, + { + "epoch": 1.660224028113332, + "grad_norm": 1.9722588062286377, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7112458944320679, + "num_tokens": 384240987.0, + "step": 15118 + }, + { + "epoch": 1.6603338458159456, + "grad_norm": 2.295994520187378, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7135449647903442, + "num_tokens": 384267515.0, + "step": 15119 + }, + { + "epoch": 1.660443663518559, + "grad_norm": 2.560847043991089, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7319133281707764, + "num_tokens": 384288231.0, + "step": 15120 + }, + { + "epoch": 1.6605534812211729, + "grad_norm": 2.094350814819336, + "learning_rate": 1e-06, + "loss": 0.817, + "mean_token_accuracy": 0.7382409572601318, + "num_tokens": 384313681.0, + "step": 15121 + }, + { + "epoch": 1.6606632989237866, + "grad_norm": 2.1360931396484375, + "learning_rate": 1e-06, + "loss": 0.7922, + "mean_token_accuracy": 0.7503328323364258, + "num_tokens": 384338012.0, + "step": 15122 + }, + { + "epoch": 1.6607731166264001, + "grad_norm": 2.027712345123291, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.7145036458969116, + "num_tokens": 384368819.0, + "step": 15123 + }, + { + "epoch": 1.6608829343290137, + "grad_norm": 2.5924735069274902, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7209381461143494, + "num_tokens": 384390996.0, + "step": 15124 + }, + { + "epoch": 1.6609927520316274, + "grad_norm": 2.3313941955566406, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7230575084686279, + "num_tokens": 384416417.0, + "step": 15125 + }, + { + "epoch": 1.6611025697342412, + "grad_norm": 2.0771868228912354, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7143948078155518, + "num_tokens": 384445118.0, + "step": 15126 + }, + { + "epoch": 1.661212387436855, + "grad_norm": 2.3727049827575684, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7143236398696899, + "num_tokens": 384467444.0, + "step": 15127 + }, + { + "epoch": 1.6613222051394685, + "grad_norm": 2.0262558460235596, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.712458610534668, + "num_tokens": 384496628.0, + "step": 15128 + }, + { + "epoch": 1.661432022842082, + "grad_norm": 2.3222262859344482, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7296444177627563, + "num_tokens": 384521406.0, + "step": 15129 + }, + { + "epoch": 1.6615418405446958, + "grad_norm": 2.12434983253479, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7332149744033813, + "num_tokens": 384550204.0, + "step": 15130 + }, + { + "epoch": 1.6616516582473095, + "grad_norm": 2.114903450012207, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.701893150806427, + "num_tokens": 384580696.0, + "step": 15131 + }, + { + "epoch": 1.661761475949923, + "grad_norm": 2.094749927520752, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.7138713598251343, + "num_tokens": 384607805.0, + "step": 15132 + }, + { + "epoch": 1.6618712936525368, + "grad_norm": 2.2037813663482666, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7254433631896973, + "num_tokens": 384634299.0, + "step": 15133 + }, + { + "epoch": 1.6619811113551504, + "grad_norm": 2.048161745071411, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7252655625343323, + "num_tokens": 384663052.0, + "step": 15134 + }, + { + "epoch": 1.6620909290577641, + "grad_norm": 2.3004260063171387, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7018279433250427, + "num_tokens": 384689025.0, + "step": 15135 + }, + { + "epoch": 1.6622007467603779, + "grad_norm": 2.1951849460601807, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7083334922790527, + "num_tokens": 384715988.0, + "step": 15136 + }, + { + "epoch": 1.6623105644629914, + "grad_norm": 2.387464761734009, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7230446338653564, + "num_tokens": 384737525.0, + "step": 15137 + }, + { + "epoch": 1.662420382165605, + "grad_norm": 2.055717945098877, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7249841094017029, + "num_tokens": 384769208.0, + "step": 15138 + }, + { + "epoch": 1.6625301998682187, + "grad_norm": 2.303744316101074, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7169699668884277, + "num_tokens": 384792589.0, + "step": 15139 + }, + { + "epoch": 1.6626400175708325, + "grad_norm": 1.7727186679840088, + "learning_rate": 1e-06, + "loss": 0.9983, + "mean_token_accuracy": 0.6914771795272827, + "num_tokens": 384830228.0, + "step": 15140 + }, + { + "epoch": 1.6627498352734462, + "grad_norm": 2.7266387939453125, + "learning_rate": 1e-06, + "loss": 0.8166, + "mean_token_accuracy": 0.7459149956703186, + "num_tokens": 384848147.0, + "step": 15141 + }, + { + "epoch": 1.6628596529760598, + "grad_norm": 2.1327719688415527, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7354472875595093, + "num_tokens": 384873919.0, + "step": 15142 + }, + { + "epoch": 1.6629694706786733, + "grad_norm": 2.549203872680664, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.711521565914154, + "num_tokens": 384895949.0, + "step": 15143 + }, + { + "epoch": 1.663079288381287, + "grad_norm": 2.2658543586730957, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7369093298912048, + "num_tokens": 384919432.0, + "step": 15144 + }, + { + "epoch": 1.6631891060839008, + "grad_norm": 2.455897092819214, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7393604516983032, + "num_tokens": 384943409.0, + "step": 15145 + }, + { + "epoch": 1.6632989237865143, + "grad_norm": 2.395563840866089, + "learning_rate": 1e-06, + "loss": 0.7996, + "mean_token_accuracy": 0.7440958023071289, + "num_tokens": 384964862.0, + "step": 15146 + }, + { + "epoch": 1.663408741489128, + "grad_norm": 2.16996169090271, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7063103914260864, + "num_tokens": 384991871.0, + "step": 15147 + }, + { + "epoch": 1.6635185591917416, + "grad_norm": 2.327360153198242, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7382440567016602, + "num_tokens": 385013727.0, + "step": 15148 + }, + { + "epoch": 1.6636283768943554, + "grad_norm": 2.190945625305176, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.6941186189651489, + "num_tokens": 385041573.0, + "step": 15149 + }, + { + "epoch": 1.6637381945969691, + "grad_norm": 2.0390100479125977, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7323118448257446, + "num_tokens": 385069297.0, + "step": 15150 + }, + { + "epoch": 1.6638480122995827, + "grad_norm": 2.550008535385132, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7224398851394653, + "num_tokens": 385088957.0, + "step": 15151 + }, + { + "epoch": 1.6639578300021962, + "grad_norm": 2.349245071411133, + "learning_rate": 1e-06, + "loss": 0.9644, + "mean_token_accuracy": 0.701880693435669, + "num_tokens": 385113818.0, + "step": 15152 + }, + { + "epoch": 1.66406764770481, + "grad_norm": 2.655308485031128, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7252671718597412, + "num_tokens": 385133562.0, + "step": 15153 + }, + { + "epoch": 1.6641774654074237, + "grad_norm": 2.2078635692596436, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.7054717540740967, + "num_tokens": 385160674.0, + "step": 15154 + }, + { + "epoch": 1.6642872831100375, + "grad_norm": 2.058925151824951, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.716790497303009, + "num_tokens": 385190446.0, + "step": 15155 + }, + { + "epoch": 1.664397100812651, + "grad_norm": 2.191044807434082, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7155097723007202, + "num_tokens": 385217706.0, + "step": 15156 + }, + { + "epoch": 1.6645069185152646, + "grad_norm": 2.061908721923828, + "learning_rate": 1e-06, + "loss": 0.8297, + "mean_token_accuracy": 0.7440005540847778, + "num_tokens": 385246725.0, + "step": 15157 + }, + { + "epoch": 1.6646167362178783, + "grad_norm": 2.311161756515503, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7077277898788452, + "num_tokens": 385271157.0, + "step": 15158 + }, + { + "epoch": 1.664726553920492, + "grad_norm": 2.3324050903320312, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7089744806289673, + "num_tokens": 385297305.0, + "step": 15159 + }, + { + "epoch": 1.6648363716231056, + "grad_norm": 2.304816246032715, + "learning_rate": 1e-06, + "loss": 0.8473, + "mean_token_accuracy": 0.7328497171401978, + "num_tokens": 385320341.0, + "step": 15160 + }, + { + "epoch": 1.6649461893257194, + "grad_norm": 2.2192375659942627, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7368307113647461, + "num_tokens": 385344796.0, + "step": 15161 + }, + { + "epoch": 1.665056007028333, + "grad_norm": 1.9549918174743652, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7071070075035095, + "num_tokens": 385379367.0, + "step": 15162 + }, + { + "epoch": 1.6651658247309467, + "grad_norm": 1.956403374671936, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7120972871780396, + "num_tokens": 385412994.0, + "step": 15163 + }, + { + "epoch": 1.6652756424335604, + "grad_norm": 2.2471203804016113, + "learning_rate": 1e-06, + "loss": 0.9894, + "mean_token_accuracy": 0.692197322845459, + "num_tokens": 385439025.0, + "step": 15164 + }, + { + "epoch": 1.665385460136174, + "grad_norm": 2.351989507675171, + "learning_rate": 1e-06, + "loss": 0.9264, + "mean_token_accuracy": 0.7185596823692322, + "num_tokens": 385461552.0, + "step": 15165 + }, + { + "epoch": 1.6654952778387875, + "grad_norm": 2.246591567993164, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7290158271789551, + "num_tokens": 385487724.0, + "step": 15166 + }, + { + "epoch": 1.6656050955414012, + "grad_norm": 2.376199722290039, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7278497815132141, + "num_tokens": 385509189.0, + "step": 15167 + }, + { + "epoch": 1.665714913244015, + "grad_norm": 2.088611125946045, + "learning_rate": 1e-06, + "loss": 0.9324, + "mean_token_accuracy": 0.707711935043335, + "num_tokens": 385538921.0, + "step": 15168 + }, + { + "epoch": 1.6658247309466288, + "grad_norm": 2.3461811542510986, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7418391704559326, + "num_tokens": 385561860.0, + "step": 15169 + }, + { + "epoch": 1.6659345486492423, + "grad_norm": 2.0901646614074707, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7071506977081299, + "num_tokens": 385589314.0, + "step": 15170 + }, + { + "epoch": 1.6660443663518558, + "grad_norm": 2.1270978450775146, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7334210276603699, + "num_tokens": 385617641.0, + "step": 15171 + }, + { + "epoch": 1.6661541840544696, + "grad_norm": 2.102309226989746, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.6969477534294128, + "num_tokens": 385648114.0, + "step": 15172 + }, + { + "epoch": 1.6662640017570833, + "grad_norm": 2.127084970474243, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7235420942306519, + "num_tokens": 385674821.0, + "step": 15173 + }, + { + "epoch": 1.6663738194596969, + "grad_norm": 2.0416276454925537, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7168530225753784, + "num_tokens": 385704625.0, + "step": 15174 + }, + { + "epoch": 1.6664836371623104, + "grad_norm": 2.151454448699951, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.727178156375885, + "num_tokens": 385730289.0, + "step": 15175 + }, + { + "epoch": 1.6665934548649242, + "grad_norm": 2.01259708404541, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7340171933174133, + "num_tokens": 385761482.0, + "step": 15176 + }, + { + "epoch": 1.666703272567538, + "grad_norm": 2.3102200031280518, + "learning_rate": 1e-06, + "loss": 0.9785, + "mean_token_accuracy": 0.694903552532196, + "num_tokens": 385788193.0, + "step": 15177 + }, + { + "epoch": 1.6668130902701517, + "grad_norm": 2.043165445327759, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.7291863560676575, + "num_tokens": 385816001.0, + "step": 15178 + }, + { + "epoch": 1.6669229079727652, + "grad_norm": 2.249415874481201, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7227814197540283, + "num_tokens": 385841386.0, + "step": 15179 + }, + { + "epoch": 1.6670327256753787, + "grad_norm": 2.0436320304870605, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.693287193775177, + "num_tokens": 385872538.0, + "step": 15180 + }, + { + "epoch": 1.6671425433779925, + "grad_norm": 2.41414475440979, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.7353463172912598, + "num_tokens": 385894830.0, + "step": 15181 + }, + { + "epoch": 1.6672523610806063, + "grad_norm": 2.366096258163452, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.737942099571228, + "num_tokens": 385918350.0, + "step": 15182 + }, + { + "epoch": 1.66736217878322, + "grad_norm": 2.2673773765563965, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7382577657699585, + "num_tokens": 385941232.0, + "step": 15183 + }, + { + "epoch": 1.6674719964858336, + "grad_norm": 2.4113543033599854, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7347811460494995, + "num_tokens": 385963099.0, + "step": 15184 + }, + { + "epoch": 1.667581814188447, + "grad_norm": 2.6573357582092285, + "learning_rate": 1e-06, + "loss": 0.8751, + "mean_token_accuracy": 0.7212692499160767, + "num_tokens": 385982960.0, + "step": 15185 + }, + { + "epoch": 1.6676916318910608, + "grad_norm": 2.1422886848449707, + "learning_rate": 1e-06, + "loss": 0.9491, + "mean_token_accuracy": 0.7055708765983582, + "num_tokens": 386011572.0, + "step": 15186 + }, + { + "epoch": 1.6678014495936746, + "grad_norm": 2.44889497756958, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7178558111190796, + "num_tokens": 386033781.0, + "step": 15187 + }, + { + "epoch": 1.6679112672962881, + "grad_norm": 2.3780620098114014, + "learning_rate": 1e-06, + "loss": 0.9573, + "mean_token_accuracy": 0.7086641192436218, + "num_tokens": 386057750.0, + "step": 15188 + }, + { + "epoch": 1.6680210849989017, + "grad_norm": 2.2864394187927246, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.7083548903465271, + "num_tokens": 386083995.0, + "step": 15189 + }, + { + "epoch": 1.6681309027015154, + "grad_norm": 2.482243299484253, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7129205465316772, + "num_tokens": 386105904.0, + "step": 15190 + }, + { + "epoch": 1.6682407204041292, + "grad_norm": 2.0240211486816406, + "learning_rate": 1e-06, + "loss": 0.8489, + "mean_token_accuracy": 0.7416813373565674, + "num_tokens": 386134748.0, + "step": 15191 + }, + { + "epoch": 1.668350538106743, + "grad_norm": 1.9964332580566406, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7286455631256104, + "num_tokens": 386165876.0, + "step": 15192 + }, + { + "epoch": 1.6684603558093565, + "grad_norm": 2.1451501846313477, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6997584700584412, + "num_tokens": 386193547.0, + "step": 15193 + }, + { + "epoch": 1.66857017351197, + "grad_norm": 2.461028575897217, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7413776516914368, + "num_tokens": 386214236.0, + "step": 15194 + }, + { + "epoch": 1.6686799912145838, + "grad_norm": 2.5467514991760254, + "learning_rate": 1e-06, + "loss": 0.846, + "mean_token_accuracy": 0.7361389398574829, + "num_tokens": 386233118.0, + "step": 15195 + }, + { + "epoch": 1.6687898089171975, + "grad_norm": 2.2125470638275146, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7274555563926697, + "num_tokens": 386258989.0, + "step": 15196 + }, + { + "epoch": 1.668899626619811, + "grad_norm": 2.2325427532196045, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7286006212234497, + "num_tokens": 386283636.0, + "step": 15197 + }, + { + "epoch": 1.6690094443224248, + "grad_norm": 2.299823760986328, + "learning_rate": 1e-06, + "loss": 0.9186, + "mean_token_accuracy": 0.7130232453346252, + "num_tokens": 386309534.0, + "step": 15198 + }, + { + "epoch": 1.6691192620250384, + "grad_norm": 1.904690146446228, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7393862009048462, + "num_tokens": 386342341.0, + "step": 15199 + }, + { + "epoch": 1.669229079727652, + "grad_norm": 2.229898691177368, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7234611511230469, + "num_tokens": 386368077.0, + "step": 15200 + }, + { + "epoch": 1.6693388974302659, + "grad_norm": 2.196969509124756, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7192273139953613, + "num_tokens": 386395006.0, + "step": 15201 + }, + { + "epoch": 1.6694487151328794, + "grad_norm": 2.2007834911346436, + "learning_rate": 1e-06, + "loss": 0.9057, + "mean_token_accuracy": 0.7138510942459106, + "num_tokens": 386420895.0, + "step": 15202 + }, + { + "epoch": 1.669558532835493, + "grad_norm": 2.176764965057373, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.720261812210083, + "num_tokens": 386447460.0, + "step": 15203 + }, + { + "epoch": 1.6696683505381067, + "grad_norm": 2.4325153827667236, + "learning_rate": 1e-06, + "loss": 0.7869, + "mean_token_accuracy": 0.7475510835647583, + "num_tokens": 386468094.0, + "step": 15204 + }, + { + "epoch": 1.6697781682407205, + "grad_norm": 2.4230148792266846, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.709409236907959, + "num_tokens": 386492490.0, + "step": 15205 + }, + { + "epoch": 1.6698879859433342, + "grad_norm": 2.1426913738250732, + "learning_rate": 1e-06, + "loss": 0.9369, + "mean_token_accuracy": 0.7221592664718628, + "num_tokens": 386520782.0, + "step": 15206 + }, + { + "epoch": 1.6699978036459477, + "grad_norm": 2.0569233894348145, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7081378698348999, + "num_tokens": 386549770.0, + "step": 15207 + }, + { + "epoch": 1.6701076213485613, + "grad_norm": 2.143683910369873, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7157578468322754, + "num_tokens": 386577425.0, + "step": 15208 + }, + { + "epoch": 1.670217439051175, + "grad_norm": 2.407355308532715, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7119677066802979, + "num_tokens": 386599686.0, + "step": 15209 + }, + { + "epoch": 1.6703272567537888, + "grad_norm": 2.5381386280059814, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7118500471115112, + "num_tokens": 386620873.0, + "step": 15210 + }, + { + "epoch": 1.6704370744564023, + "grad_norm": 2.415290355682373, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.715010941028595, + "num_tokens": 386643369.0, + "step": 15211 + }, + { + "epoch": 1.670546892159016, + "grad_norm": 2.3288557529449463, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7397453784942627, + "num_tokens": 386665989.0, + "step": 15212 + }, + { + "epoch": 1.6706567098616296, + "grad_norm": 2.74607515335083, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.7363649606704712, + "num_tokens": 386683501.0, + "step": 15213 + }, + { + "epoch": 1.6707665275642434, + "grad_norm": 2.3647937774658203, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7297316193580627, + "num_tokens": 386704616.0, + "step": 15214 + }, + { + "epoch": 1.6708763452668571, + "grad_norm": 2.5526580810546875, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7163088321685791, + "num_tokens": 386726418.0, + "step": 15215 + }, + { + "epoch": 1.6709861629694707, + "grad_norm": 2.232680320739746, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.710912823677063, + "num_tokens": 386754216.0, + "step": 15216 + }, + { + "epoch": 1.6710959806720842, + "grad_norm": 2.1702303886413574, + "learning_rate": 1e-06, + "loss": 0.9418, + "mean_token_accuracy": 0.7177145481109619, + "num_tokens": 386780038.0, + "step": 15217 + }, + { + "epoch": 1.671205798374698, + "grad_norm": 2.249331474304199, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7202221155166626, + "num_tokens": 386805660.0, + "step": 15218 + }, + { + "epoch": 1.6713156160773117, + "grad_norm": 2.2275028228759766, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7359274625778198, + "num_tokens": 386830459.0, + "step": 15219 + }, + { + "epoch": 1.6714254337799255, + "grad_norm": 2.1334598064422607, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7129303812980652, + "num_tokens": 386859882.0, + "step": 15220 + }, + { + "epoch": 1.671535251482539, + "grad_norm": 2.1016948223114014, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7282788753509521, + "num_tokens": 386887163.0, + "step": 15221 + }, + { + "epoch": 1.6716450691851525, + "grad_norm": 2.228288412094116, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.697201132774353, + "num_tokens": 386914022.0, + "step": 15222 + }, + { + "epoch": 1.6717548868877663, + "grad_norm": 2.49601674079895, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7133708596229553, + "num_tokens": 386934545.0, + "step": 15223 + }, + { + "epoch": 1.67186470459038, + "grad_norm": 2.2547738552093506, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.719750165939331, + "num_tokens": 386960970.0, + "step": 15224 + }, + { + "epoch": 1.6719745222929936, + "grad_norm": 2.2109057903289795, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7162927985191345, + "num_tokens": 386986816.0, + "step": 15225 + }, + { + "epoch": 1.6720843399956071, + "grad_norm": 2.365281820297241, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7119723558425903, + "num_tokens": 387009370.0, + "step": 15226 + }, + { + "epoch": 1.6721941576982209, + "grad_norm": 1.9567434787750244, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7252371311187744, + "num_tokens": 387040136.0, + "step": 15227 + }, + { + "epoch": 1.6723039754008346, + "grad_norm": 2.653280735015869, + "learning_rate": 1e-06, + "loss": 0.7916, + "mean_token_accuracy": 0.7456810474395752, + "num_tokens": 387057835.0, + "step": 15228 + }, + { + "epoch": 1.6724137931034484, + "grad_norm": 2.080974817276001, + "learning_rate": 1e-06, + "loss": 0.9623, + "mean_token_accuracy": 0.7058357000350952, + "num_tokens": 387087520.0, + "step": 15229 + }, + { + "epoch": 1.672523610806062, + "grad_norm": 2.2459750175476074, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7219597101211548, + "num_tokens": 387115150.0, + "step": 15230 + }, + { + "epoch": 1.6726334285086755, + "grad_norm": 2.5028038024902344, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.733995258808136, + "num_tokens": 387135703.0, + "step": 15231 + }, + { + "epoch": 1.6727432462112892, + "grad_norm": 2.1617064476013184, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.7111150622367859, + "num_tokens": 387163193.0, + "step": 15232 + }, + { + "epoch": 1.672853063913903, + "grad_norm": 2.6091926097869873, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7305141687393188, + "num_tokens": 387183530.0, + "step": 15233 + }, + { + "epoch": 1.6729628816165167, + "grad_norm": 2.7465648651123047, + "learning_rate": 1e-06, + "loss": 0.8713, + "mean_token_accuracy": 0.7280852794647217, + "num_tokens": 387201158.0, + "step": 15234 + }, + { + "epoch": 1.6730726993191303, + "grad_norm": 2.6539111137390137, + "learning_rate": 1e-06, + "loss": 0.8681, + "mean_token_accuracy": 0.7294183969497681, + "num_tokens": 387221586.0, + "step": 15235 + }, + { + "epoch": 1.6731825170217438, + "grad_norm": 2.289888620376587, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7367932200431824, + "num_tokens": 387245471.0, + "step": 15236 + }, + { + "epoch": 1.6732923347243576, + "grad_norm": 2.192481279373169, + "learning_rate": 1e-06, + "loss": 0.9293, + "mean_token_accuracy": 0.7129333019256592, + "num_tokens": 387272525.0, + "step": 15237 + }, + { + "epoch": 1.6734021524269713, + "grad_norm": 2.549212694168091, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7367991805076599, + "num_tokens": 387291938.0, + "step": 15238 + }, + { + "epoch": 1.6735119701295849, + "grad_norm": 2.8491156101226807, + "learning_rate": 1e-06, + "loss": 0.7942, + "mean_token_accuracy": 0.7426740527153015, + "num_tokens": 387309346.0, + "step": 15239 + }, + { + "epoch": 1.6736217878321984, + "grad_norm": 2.29126238822937, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7264343500137329, + "num_tokens": 387332855.0, + "step": 15240 + }, + { + "epoch": 1.6737316055348122, + "grad_norm": 2.4119160175323486, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7373777627944946, + "num_tokens": 387354403.0, + "step": 15241 + }, + { + "epoch": 1.673841423237426, + "grad_norm": 2.241642951965332, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.7266063094139099, + "num_tokens": 387378584.0, + "step": 15242 + }, + { + "epoch": 1.6739512409400397, + "grad_norm": 2.226424217224121, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.719984769821167, + "num_tokens": 387404727.0, + "step": 15243 + }, + { + "epoch": 1.6740610586426532, + "grad_norm": 2.225320339202881, + "learning_rate": 1e-06, + "loss": 0.8213, + "mean_token_accuracy": 0.7490538358688354, + "num_tokens": 387429083.0, + "step": 15244 + }, + { + "epoch": 1.6741708763452667, + "grad_norm": 2.2351770401000977, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.6980342268943787, + "num_tokens": 387456118.0, + "step": 15245 + }, + { + "epoch": 1.6742806940478805, + "grad_norm": 2.5027718544006348, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7261731624603271, + "num_tokens": 387477044.0, + "step": 15246 + }, + { + "epoch": 1.6743905117504942, + "grad_norm": 2.0335607528686523, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.718215823173523, + "num_tokens": 387505849.0, + "step": 15247 + }, + { + "epoch": 1.674500329453108, + "grad_norm": 1.959810733795166, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7276430726051331, + "num_tokens": 387539754.0, + "step": 15248 + }, + { + "epoch": 1.6746101471557215, + "grad_norm": 2.329303741455078, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7133329510688782, + "num_tokens": 387563084.0, + "step": 15249 + }, + { + "epoch": 1.674719964858335, + "grad_norm": 2.158738613128662, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7269797325134277, + "num_tokens": 387590607.0, + "step": 15250 + }, + { + "epoch": 1.6748297825609488, + "grad_norm": 1.9645029306411743, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7225633263587952, + "num_tokens": 387622468.0, + "step": 15251 + }, + { + "epoch": 1.6749396002635626, + "grad_norm": 2.183333396911621, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7102966904640198, + "num_tokens": 387648525.0, + "step": 15252 + }, + { + "epoch": 1.6750494179661761, + "grad_norm": 2.2077648639678955, + "learning_rate": 1e-06, + "loss": 0.9839, + "mean_token_accuracy": 0.6948206424713135, + "num_tokens": 387675554.0, + "step": 15253 + }, + { + "epoch": 1.6751592356687897, + "grad_norm": 2.2035694122314453, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7100909948348999, + "num_tokens": 387702306.0, + "step": 15254 + }, + { + "epoch": 1.6752690533714034, + "grad_norm": 2.7292990684509277, + "learning_rate": 1e-06, + "loss": 0.8484, + "mean_token_accuracy": 0.7359319925308228, + "num_tokens": 387721448.0, + "step": 15255 + }, + { + "epoch": 1.6753788710740172, + "grad_norm": 2.318498373031616, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.7275265455245972, + "num_tokens": 387744694.0, + "step": 15256 + }, + { + "epoch": 1.675488688776631, + "grad_norm": 2.3454537391662598, + "learning_rate": 1e-06, + "loss": 0.9646, + "mean_token_accuracy": 0.7118433713912964, + "num_tokens": 387770687.0, + "step": 15257 + }, + { + "epoch": 1.6755985064792445, + "grad_norm": 2.132495641708374, + "learning_rate": 1e-06, + "loss": 0.933, + "mean_token_accuracy": 0.7060517072677612, + "num_tokens": 387798243.0, + "step": 15258 + }, + { + "epoch": 1.675708324181858, + "grad_norm": 2.1400249004364014, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7186269164085388, + "num_tokens": 387825381.0, + "step": 15259 + }, + { + "epoch": 1.6758181418844718, + "grad_norm": 2.0139353275299072, + "learning_rate": 1e-06, + "loss": 0.895, + "mean_token_accuracy": 0.7195678949356079, + "num_tokens": 387854737.0, + "step": 15260 + }, + { + "epoch": 1.6759279595870855, + "grad_norm": 2.312983274459839, + "learning_rate": 1e-06, + "loss": 0.9655, + "mean_token_accuracy": 0.7086490392684937, + "num_tokens": 387880232.0, + "step": 15261 + }, + { + "epoch": 1.676037777289699, + "grad_norm": 2.013113498687744, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7150017619132996, + "num_tokens": 387911557.0, + "step": 15262 + }, + { + "epoch": 1.6761475949923128, + "grad_norm": 2.039043426513672, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7134816646575928, + "num_tokens": 387943916.0, + "step": 15263 + }, + { + "epoch": 1.6762574126949263, + "grad_norm": 2.411652088165283, + "learning_rate": 1e-06, + "loss": 0.7911, + "mean_token_accuracy": 0.7514269351959229, + "num_tokens": 387965396.0, + "step": 15264 + }, + { + "epoch": 1.67636723039754, + "grad_norm": 2.0132973194122314, + "learning_rate": 1e-06, + "loss": 0.9693, + "mean_token_accuracy": 0.7031926512718201, + "num_tokens": 387994814.0, + "step": 15265 + }, + { + "epoch": 1.6764770481001539, + "grad_norm": 2.2221579551696777, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7364751696586609, + "num_tokens": 388019559.0, + "step": 15266 + }, + { + "epoch": 1.6765868658027674, + "grad_norm": 2.117173433303833, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7129932045936584, + "num_tokens": 388049513.0, + "step": 15267 + }, + { + "epoch": 1.676696683505381, + "grad_norm": 2.306781053543091, + "learning_rate": 1e-06, + "loss": 0.8895, + "mean_token_accuracy": 0.7285122275352478, + "num_tokens": 388074281.0, + "step": 15268 + }, + { + "epoch": 1.6768065012079947, + "grad_norm": 2.14009428024292, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7119238376617432, + "num_tokens": 388101744.0, + "step": 15269 + }, + { + "epoch": 1.6769163189106084, + "grad_norm": 2.036789894104004, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7324186563491821, + "num_tokens": 388129620.0, + "step": 15270 + }, + { + "epoch": 1.6770261366132222, + "grad_norm": 2.2272868156433105, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7237606644630432, + "num_tokens": 388155906.0, + "step": 15271 + }, + { + "epoch": 1.6771359543158357, + "grad_norm": 2.008305788040161, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7096664905548096, + "num_tokens": 388187893.0, + "step": 15272 + }, + { + "epoch": 1.6772457720184493, + "grad_norm": 2.181086778640747, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7102069854736328, + "num_tokens": 388216190.0, + "step": 15273 + }, + { + "epoch": 1.677355589721063, + "grad_norm": 2.127164602279663, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.7040436863899231, + "num_tokens": 388244830.0, + "step": 15274 + }, + { + "epoch": 1.6774654074236768, + "grad_norm": 2.0467212200164795, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7078698873519897, + "num_tokens": 388275009.0, + "step": 15275 + }, + { + "epoch": 1.6775752251262903, + "grad_norm": 2.4535419940948486, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7173577547073364, + "num_tokens": 388297328.0, + "step": 15276 + }, + { + "epoch": 1.677685042828904, + "grad_norm": 2.1242666244506836, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.7197625041007996, + "num_tokens": 388326060.0, + "step": 15277 + }, + { + "epoch": 1.6777948605315176, + "grad_norm": 2.2282707691192627, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7474578022956848, + "num_tokens": 388351019.0, + "step": 15278 + }, + { + "epoch": 1.6779046782341314, + "grad_norm": 2.433917284011841, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7192541360855103, + "num_tokens": 388372589.0, + "step": 15279 + }, + { + "epoch": 1.6780144959367451, + "grad_norm": 2.2798328399658203, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.713977575302124, + "num_tokens": 388397830.0, + "step": 15280 + }, + { + "epoch": 1.6781243136393587, + "grad_norm": 2.070664644241333, + "learning_rate": 1e-06, + "loss": 0.9679, + "mean_token_accuracy": 0.7038775682449341, + "num_tokens": 388428500.0, + "step": 15281 + }, + { + "epoch": 1.6782341313419722, + "grad_norm": 2.2803993225097656, + "learning_rate": 1e-06, + "loss": 0.8916, + "mean_token_accuracy": 0.7209041118621826, + "num_tokens": 388452443.0, + "step": 15282 + }, + { + "epoch": 1.678343949044586, + "grad_norm": 2.3362507820129395, + "learning_rate": 1e-06, + "loss": 0.8597, + "mean_token_accuracy": 0.7304328680038452, + "num_tokens": 388476061.0, + "step": 15283 + }, + { + "epoch": 1.6784537667471997, + "grad_norm": 2.3651912212371826, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7162508964538574, + "num_tokens": 388500623.0, + "step": 15284 + }, + { + "epoch": 1.6785635844498135, + "grad_norm": 2.396087884902954, + "learning_rate": 1e-06, + "loss": 0.757, + "mean_token_accuracy": 0.7654120326042175, + "num_tokens": 388521979.0, + "step": 15285 + }, + { + "epoch": 1.678673402152427, + "grad_norm": 2.4090845584869385, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7409145832061768, + "num_tokens": 388543155.0, + "step": 15286 + }, + { + "epoch": 1.6787832198550405, + "grad_norm": 2.197115182876587, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7084636688232422, + "num_tokens": 388572728.0, + "step": 15287 + }, + { + "epoch": 1.6788930375576543, + "grad_norm": 2.4880805015563965, + "learning_rate": 1e-06, + "loss": 0.8478, + "mean_token_accuracy": 0.7316051721572876, + "num_tokens": 388592902.0, + "step": 15288 + }, + { + "epoch": 1.679002855260268, + "grad_norm": 2.4022111892700195, + "learning_rate": 1e-06, + "loss": 0.96, + "mean_token_accuracy": 0.7025801539421082, + "num_tokens": 388617476.0, + "step": 15289 + }, + { + "epoch": 1.6791126729628816, + "grad_norm": 2.410318374633789, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.731895923614502, + "num_tokens": 388639609.0, + "step": 15290 + }, + { + "epoch": 1.6792224906654951, + "grad_norm": 2.5840651988983154, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7136077880859375, + "num_tokens": 388661145.0, + "step": 15291 + }, + { + "epoch": 1.6793323083681089, + "grad_norm": 2.021043300628662, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.744260847568512, + "num_tokens": 388689373.0, + "step": 15292 + }, + { + "epoch": 1.6794421260707226, + "grad_norm": 2.3424651622772217, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7104337215423584, + "num_tokens": 388715359.0, + "step": 15293 + }, + { + "epoch": 1.6795519437733364, + "grad_norm": 2.221639394760132, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7139050960540771, + "num_tokens": 388741677.0, + "step": 15294 + }, + { + "epoch": 1.67966176147595, + "grad_norm": 2.251831293106079, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7147205471992493, + "num_tokens": 388767038.0, + "step": 15295 + }, + { + "epoch": 1.6797715791785635, + "grad_norm": 2.2768938541412354, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7105616331100464, + "num_tokens": 388791720.0, + "step": 15296 + }, + { + "epoch": 1.6798813968811772, + "grad_norm": 2.401625871658325, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7459414005279541, + "num_tokens": 388812301.0, + "step": 15297 + }, + { + "epoch": 1.679991214583791, + "grad_norm": 2.1658098697662354, + "learning_rate": 1e-06, + "loss": 0.9704, + "mean_token_accuracy": 0.695742666721344, + "num_tokens": 388839487.0, + "step": 15298 + }, + { + "epoch": 1.6801010322864047, + "grad_norm": 2.3903391361236572, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7115785479545593, + "num_tokens": 388863265.0, + "step": 15299 + }, + { + "epoch": 1.6802108499890183, + "grad_norm": 2.0528993606567383, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7420291900634766, + "num_tokens": 388892165.0, + "step": 15300 + }, + { + "epoch": 1.6803206676916318, + "grad_norm": 2.31986927986145, + "learning_rate": 1e-06, + "loss": 0.995, + "mean_token_accuracy": 0.6994841694831848, + "num_tokens": 388920084.0, + "step": 15301 + }, + { + "epoch": 1.6804304853942456, + "grad_norm": 1.996566891670227, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.6903451681137085, + "num_tokens": 388955894.0, + "step": 15302 + }, + { + "epoch": 1.6805403030968593, + "grad_norm": 2.167914628982544, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7191234827041626, + "num_tokens": 388984224.0, + "step": 15303 + }, + { + "epoch": 1.6806501207994728, + "grad_norm": 2.2078237533569336, + "learning_rate": 1e-06, + "loss": 0.8347, + "mean_token_accuracy": 0.7328077554702759, + "num_tokens": 389012513.0, + "step": 15304 + }, + { + "epoch": 1.6807599385020864, + "grad_norm": 2.342853307723999, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7140380144119263, + "num_tokens": 389034857.0, + "step": 15305 + }, + { + "epoch": 1.6808697562047001, + "grad_norm": 2.076756000518799, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7003165483474731, + "num_tokens": 389064648.0, + "step": 15306 + }, + { + "epoch": 1.680979573907314, + "grad_norm": 1.9435092210769653, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7114184498786926, + "num_tokens": 389096932.0, + "step": 15307 + }, + { + "epoch": 1.6810893916099277, + "grad_norm": 2.4202311038970947, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7282557487487793, + "num_tokens": 389117731.0, + "step": 15308 + }, + { + "epoch": 1.6811992093125412, + "grad_norm": 2.5726428031921387, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7388108372688293, + "num_tokens": 389137370.0, + "step": 15309 + }, + { + "epoch": 1.6813090270151547, + "grad_norm": 2.132509469985962, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.6973751187324524, + "num_tokens": 389167792.0, + "step": 15310 + }, + { + "epoch": 1.6814188447177685, + "grad_norm": 2.1725480556488037, + "learning_rate": 1e-06, + "loss": 0.9854, + "mean_token_accuracy": 0.701514482498169, + "num_tokens": 389194069.0, + "step": 15311 + }, + { + "epoch": 1.6815286624203822, + "grad_norm": 2.137887954711914, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7070409059524536, + "num_tokens": 389223313.0, + "step": 15312 + }, + { + "epoch": 1.681638480122996, + "grad_norm": 2.5271763801574707, + "learning_rate": 1e-06, + "loss": 0.7661, + "mean_token_accuracy": 0.7517392635345459, + "num_tokens": 389244624.0, + "step": 15313 + }, + { + "epoch": 1.6817482978256095, + "grad_norm": 2.803147077560425, + "learning_rate": 1e-06, + "loss": 0.8372, + "mean_token_accuracy": 0.7254864573478699, + "num_tokens": 389261944.0, + "step": 15314 + }, + { + "epoch": 1.681858115528223, + "grad_norm": 2.591111421585083, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7346636056900024, + "num_tokens": 389281117.0, + "step": 15315 + }, + { + "epoch": 1.6819679332308368, + "grad_norm": 2.535769462585449, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7320411801338196, + "num_tokens": 389303448.0, + "step": 15316 + }, + { + "epoch": 1.6820777509334506, + "grad_norm": 2.3736228942871094, + "learning_rate": 1e-06, + "loss": 0.8288, + "mean_token_accuracy": 0.7358975410461426, + "num_tokens": 389326766.0, + "step": 15317 + }, + { + "epoch": 1.6821875686360641, + "grad_norm": 2.023531675338745, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7131460905075073, + "num_tokens": 389358678.0, + "step": 15318 + }, + { + "epoch": 1.6822973863386776, + "grad_norm": 2.224543571472168, + "learning_rate": 1e-06, + "loss": 1.0204, + "mean_token_accuracy": 0.6826037168502808, + "num_tokens": 389386837.0, + "step": 15319 + }, + { + "epoch": 1.6824072040412914, + "grad_norm": 2.4636342525482178, + "learning_rate": 1e-06, + "loss": 0.9361, + "mean_token_accuracy": 0.707072377204895, + "num_tokens": 389410139.0, + "step": 15320 + }, + { + "epoch": 1.6825170217439052, + "grad_norm": 2.0972890853881836, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7316657304763794, + "num_tokens": 389438260.0, + "step": 15321 + }, + { + "epoch": 1.682626839446519, + "grad_norm": 2.0991439819335938, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7142423391342163, + "num_tokens": 389468308.0, + "step": 15322 + }, + { + "epoch": 1.6827366571491325, + "grad_norm": 2.5356245040893555, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7378470301628113, + "num_tokens": 389488819.0, + "step": 15323 + }, + { + "epoch": 1.682846474851746, + "grad_norm": 2.167428731918335, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7105295062065125, + "num_tokens": 389514659.0, + "step": 15324 + }, + { + "epoch": 1.6829562925543597, + "grad_norm": 2.6121103763580322, + "learning_rate": 1e-06, + "loss": 0.7478, + "mean_token_accuracy": 0.7652953863143921, + "num_tokens": 389533715.0, + "step": 15325 + }, + { + "epoch": 1.6830661102569735, + "grad_norm": 2.3845746517181396, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.731696367263794, + "num_tokens": 389556173.0, + "step": 15326 + }, + { + "epoch": 1.683175927959587, + "grad_norm": 2.1606874465942383, + "learning_rate": 1e-06, + "loss": 0.9635, + "mean_token_accuracy": 0.694851279258728, + "num_tokens": 389586568.0, + "step": 15327 + }, + { + "epoch": 1.6832857456622008, + "grad_norm": 2.6047167778015137, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7206082344055176, + "num_tokens": 389606248.0, + "step": 15328 + }, + { + "epoch": 1.6833955633648143, + "grad_norm": 2.454315662384033, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7264325618743896, + "num_tokens": 389626904.0, + "step": 15329 + }, + { + "epoch": 1.683505381067428, + "grad_norm": 2.436793327331543, + "learning_rate": 1e-06, + "loss": 0.7789, + "mean_token_accuracy": 0.7468979954719543, + "num_tokens": 389647499.0, + "step": 15330 + }, + { + "epoch": 1.6836151987700418, + "grad_norm": 2.3151538372039795, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7324647307395935, + "num_tokens": 389670117.0, + "step": 15331 + }, + { + "epoch": 1.6837250164726554, + "grad_norm": 2.059178113937378, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7265926599502563, + "num_tokens": 389698265.0, + "step": 15332 + }, + { + "epoch": 1.683834834175269, + "grad_norm": 2.275177478790283, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7091268301010132, + "num_tokens": 389724009.0, + "step": 15333 + }, + { + "epoch": 1.6839446518778827, + "grad_norm": 2.245234966278076, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7300092577934265, + "num_tokens": 389746736.0, + "step": 15334 + }, + { + "epoch": 1.6840544695804964, + "grad_norm": 1.7945061922073364, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7198770046234131, + "num_tokens": 389784011.0, + "step": 15335 + }, + { + "epoch": 1.6841642872831102, + "grad_norm": 2.4490346908569336, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7194260954856873, + "num_tokens": 389808111.0, + "step": 15336 + }, + { + "epoch": 1.6842741049857237, + "grad_norm": 2.090122938156128, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7244209051132202, + "num_tokens": 389836653.0, + "step": 15337 + }, + { + "epoch": 1.6843839226883373, + "grad_norm": 2.045166015625, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7314825654029846, + "num_tokens": 389864579.0, + "step": 15338 + }, + { + "epoch": 1.684493740390951, + "grad_norm": 2.1627585887908936, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7097247838973999, + "num_tokens": 389892516.0, + "step": 15339 + }, + { + "epoch": 1.6846035580935648, + "grad_norm": 2.244370698928833, + "learning_rate": 1e-06, + "loss": 0.7537, + "mean_token_accuracy": 0.7598807215690613, + "num_tokens": 389917017.0, + "step": 15340 + }, + { + "epoch": 1.6847133757961783, + "grad_norm": 2.4283430576324463, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7540608644485474, + "num_tokens": 389937697.0, + "step": 15341 + }, + { + "epoch": 1.684823193498792, + "grad_norm": 2.428746461868286, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7157933712005615, + "num_tokens": 389961634.0, + "step": 15342 + }, + { + "epoch": 1.6849330112014056, + "grad_norm": 2.1964967250823975, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7339011430740356, + "num_tokens": 389986878.0, + "step": 15343 + }, + { + "epoch": 1.6850428289040194, + "grad_norm": 2.5988681316375732, + "learning_rate": 1e-06, + "loss": 0.7479, + "mean_token_accuracy": 0.752778172492981, + "num_tokens": 390005225.0, + "step": 15344 + }, + { + "epoch": 1.685152646606633, + "grad_norm": 2.372863292694092, + "learning_rate": 1e-06, + "loss": 0.783, + "mean_token_accuracy": 0.7556867599487305, + "num_tokens": 390026076.0, + "step": 15345 + }, + { + "epoch": 1.6852624643092466, + "grad_norm": 2.169121503829956, + "learning_rate": 1e-06, + "loss": 0.9373, + "mean_token_accuracy": 0.7092728614807129, + "num_tokens": 390054923.0, + "step": 15346 + }, + { + "epoch": 1.6853722820118602, + "grad_norm": 2.1580753326416016, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7440739870071411, + "num_tokens": 390080719.0, + "step": 15347 + }, + { + "epoch": 1.685482099714474, + "grad_norm": 2.1400551795959473, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7262684106826782, + "num_tokens": 390108085.0, + "step": 15348 + }, + { + "epoch": 1.6855919174170877, + "grad_norm": 2.4375762939453125, + "learning_rate": 1e-06, + "loss": 0.9604, + "mean_token_accuracy": 0.7044434547424316, + "num_tokens": 390131185.0, + "step": 15349 + }, + { + "epoch": 1.6857017351197014, + "grad_norm": 2.2149252891540527, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.741796612739563, + "num_tokens": 390157176.0, + "step": 15350 + }, + { + "epoch": 1.685811552822315, + "grad_norm": 2.1287271976470947, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7128043174743652, + "num_tokens": 390185142.0, + "step": 15351 + }, + { + "epoch": 1.6859213705249285, + "grad_norm": 2.2541539669036865, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7154239416122437, + "num_tokens": 390210153.0, + "step": 15352 + }, + { + "epoch": 1.6860311882275423, + "grad_norm": 2.109560489654541, + "learning_rate": 1e-06, + "loss": 0.9452, + "mean_token_accuracy": 0.7077522277832031, + "num_tokens": 390239232.0, + "step": 15353 + }, + { + "epoch": 1.686141005930156, + "grad_norm": 2.010523557662964, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7020437717437744, + "num_tokens": 390269652.0, + "step": 15354 + }, + { + "epoch": 1.6862508236327696, + "grad_norm": 2.198225259780884, + "learning_rate": 1e-06, + "loss": 0.9334, + "mean_token_accuracy": 0.718402624130249, + "num_tokens": 390296902.0, + "step": 15355 + }, + { + "epoch": 1.686360641335383, + "grad_norm": 2.4413211345672607, + "learning_rate": 1e-06, + "loss": 0.8983, + "mean_token_accuracy": 0.7176305055618286, + "num_tokens": 390317857.0, + "step": 15356 + }, + { + "epoch": 1.6864704590379969, + "grad_norm": 2.8849599361419678, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.7357469201087952, + "num_tokens": 390334819.0, + "step": 15357 + }, + { + "epoch": 1.6865802767406106, + "grad_norm": 2.0794014930725098, + "learning_rate": 1e-06, + "loss": 1.038, + "mean_token_accuracy": 0.6829564571380615, + "num_tokens": 390366872.0, + "step": 15358 + }, + { + "epoch": 1.6866900944432244, + "grad_norm": 2.4298524856567383, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7334145903587341, + "num_tokens": 390388716.0, + "step": 15359 + }, + { + "epoch": 1.686799912145838, + "grad_norm": 2.8907485008239746, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.721971333026886, + "num_tokens": 390406712.0, + "step": 15360 + }, + { + "epoch": 1.6869097298484514, + "grad_norm": 2.380340099334717, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7237033843994141, + "num_tokens": 390429954.0, + "step": 15361 + }, + { + "epoch": 1.6870195475510652, + "grad_norm": 2.172821521759033, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7143004536628723, + "num_tokens": 390457058.0, + "step": 15362 + }, + { + "epoch": 1.687129365253679, + "grad_norm": 2.29445743560791, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.7044036388397217, + "num_tokens": 390482849.0, + "step": 15363 + }, + { + "epoch": 1.6872391829562927, + "grad_norm": 2.1396918296813965, + "learning_rate": 1e-06, + "loss": 0.8755, + "mean_token_accuracy": 0.722514271736145, + "num_tokens": 390510580.0, + "step": 15364 + }, + { + "epoch": 1.6873490006589063, + "grad_norm": 2.4366369247436523, + "learning_rate": 1e-06, + "loss": 0.8386, + "mean_token_accuracy": 0.7346522212028503, + "num_tokens": 390535015.0, + "step": 15365 + }, + { + "epoch": 1.6874588183615198, + "grad_norm": 2.0821428298950195, + "learning_rate": 1e-06, + "loss": 1.0254, + "mean_token_accuracy": 0.7036314606666565, + "num_tokens": 390566894.0, + "step": 15366 + }, + { + "epoch": 1.6875686360641335, + "grad_norm": 2.2992334365844727, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7037760019302368, + "num_tokens": 390591858.0, + "step": 15367 + }, + { + "epoch": 1.6876784537667473, + "grad_norm": 2.0132389068603516, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7276685237884521, + "num_tokens": 390620762.0, + "step": 15368 + }, + { + "epoch": 1.6877882714693608, + "grad_norm": 2.3946075439453125, + "learning_rate": 1e-06, + "loss": 0.7315, + "mean_token_accuracy": 0.7579560875892639, + "num_tokens": 390641823.0, + "step": 15369 + }, + { + "epoch": 1.6878980891719744, + "grad_norm": 2.405583143234253, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7195925712585449, + "num_tokens": 390665053.0, + "step": 15370 + }, + { + "epoch": 1.6880079068745881, + "grad_norm": 2.28814435005188, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.7043545842170715, + "num_tokens": 390689177.0, + "step": 15371 + }, + { + "epoch": 1.6881177245772019, + "grad_norm": 2.6172430515289307, + "learning_rate": 1e-06, + "loss": 0.8378, + "mean_token_accuracy": 0.7359078526496887, + "num_tokens": 390708883.0, + "step": 15372 + }, + { + "epoch": 1.6882275422798156, + "grad_norm": 2.014277219772339, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.71358323097229, + "num_tokens": 390739769.0, + "step": 15373 + }, + { + "epoch": 1.6883373599824292, + "grad_norm": 2.055096387863159, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7156189680099487, + "num_tokens": 390769562.0, + "step": 15374 + }, + { + "epoch": 1.6884471776850427, + "grad_norm": 2.1745376586914062, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7029293179512024, + "num_tokens": 390796457.0, + "step": 15375 + }, + { + "epoch": 1.6885569953876565, + "grad_norm": 2.5307729244232178, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7222768068313599, + "num_tokens": 390818219.0, + "step": 15376 + }, + { + "epoch": 1.6886668130902702, + "grad_norm": 2.246396780014038, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7148692011833191, + "num_tokens": 390844810.0, + "step": 15377 + }, + { + "epoch": 1.6887766307928838, + "grad_norm": 2.141521692276001, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7030442953109741, + "num_tokens": 390875689.0, + "step": 15378 + }, + { + "epoch": 1.6888864484954975, + "grad_norm": 2.3876681327819824, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7165367007255554, + "num_tokens": 390900636.0, + "step": 15379 + }, + { + "epoch": 1.688996266198111, + "grad_norm": 2.5610718727111816, + "learning_rate": 1e-06, + "loss": 0.8542, + "mean_token_accuracy": 0.7331439256668091, + "num_tokens": 390921000.0, + "step": 15380 + }, + { + "epoch": 1.6891060839007248, + "grad_norm": 2.159080982208252, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7198547124862671, + "num_tokens": 390948635.0, + "step": 15381 + }, + { + "epoch": 1.6892159016033386, + "grad_norm": 2.139641523361206, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.7114731073379517, + "num_tokens": 390975773.0, + "step": 15382 + }, + { + "epoch": 1.689325719305952, + "grad_norm": 1.8830947875976562, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7027024626731873, + "num_tokens": 391012522.0, + "step": 15383 + }, + { + "epoch": 1.6894355370085656, + "grad_norm": 2.1284842491149902, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.714423656463623, + "num_tokens": 391040116.0, + "step": 15384 + }, + { + "epoch": 1.6895453547111794, + "grad_norm": 2.3364522457122803, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7067228555679321, + "num_tokens": 391067384.0, + "step": 15385 + }, + { + "epoch": 1.6896551724137931, + "grad_norm": 2.375474691390991, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.7018171548843384, + "num_tokens": 391091235.0, + "step": 15386 + }, + { + "epoch": 1.689764990116407, + "grad_norm": 2.1710946559906006, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7100189328193665, + "num_tokens": 391119350.0, + "step": 15387 + }, + { + "epoch": 1.6898748078190204, + "grad_norm": 2.5147175788879395, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.704601526260376, + "num_tokens": 391140950.0, + "step": 15388 + }, + { + "epoch": 1.689984625521634, + "grad_norm": 2.195399045944214, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7283793091773987, + "num_tokens": 391166638.0, + "step": 15389 + }, + { + "epoch": 1.6900944432242477, + "grad_norm": 2.3633272647857666, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7134292125701904, + "num_tokens": 391189751.0, + "step": 15390 + }, + { + "epoch": 1.6902042609268615, + "grad_norm": 2.2338600158691406, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7281903624534607, + "num_tokens": 391215856.0, + "step": 15391 + }, + { + "epoch": 1.690314078629475, + "grad_norm": 1.8787263631820679, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7241145968437195, + "num_tokens": 391248100.0, + "step": 15392 + }, + { + "epoch": 1.6904238963320888, + "grad_norm": 2.0865559577941895, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.716426432132721, + "num_tokens": 391278318.0, + "step": 15393 + }, + { + "epoch": 1.6905337140347023, + "grad_norm": 2.4064888954162598, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7313278317451477, + "num_tokens": 391301192.0, + "step": 15394 + }, + { + "epoch": 1.690643531737316, + "grad_norm": 2.4552793502807617, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7131259441375732, + "num_tokens": 391323478.0, + "step": 15395 + }, + { + "epoch": 1.6907533494399298, + "grad_norm": 2.294299364089966, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7291904091835022, + "num_tokens": 391347432.0, + "step": 15396 + }, + { + "epoch": 1.6908631671425434, + "grad_norm": 2.2077934741973877, + "learning_rate": 1e-06, + "loss": 0.9543, + "mean_token_accuracy": 0.7040791511535645, + "num_tokens": 391373040.0, + "step": 15397 + }, + { + "epoch": 1.690972984845157, + "grad_norm": 2.058270215988159, + "learning_rate": 1e-06, + "loss": 1.0026, + "mean_token_accuracy": 0.6941544413566589, + "num_tokens": 391407383.0, + "step": 15398 + }, + { + "epoch": 1.6910828025477707, + "grad_norm": 2.358292579650879, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.721444845199585, + "num_tokens": 391431806.0, + "step": 15399 + }, + { + "epoch": 1.6911926202503844, + "grad_norm": 2.157848358154297, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7096805572509766, + "num_tokens": 391459071.0, + "step": 15400 + }, + { + "epoch": 1.6913024379529982, + "grad_norm": 2.244091272354126, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.712202787399292, + "num_tokens": 391485327.0, + "step": 15401 + }, + { + "epoch": 1.6914122556556117, + "grad_norm": 2.1975197792053223, + "learning_rate": 1e-06, + "loss": 0.87, + "mean_token_accuracy": 0.7258652448654175, + "num_tokens": 391511536.0, + "step": 15402 + }, + { + "epoch": 1.6915220733582252, + "grad_norm": 2.258054733276367, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7176462411880493, + "num_tokens": 391537775.0, + "step": 15403 + }, + { + "epoch": 1.691631891060839, + "grad_norm": 1.9410409927368164, + "learning_rate": 1e-06, + "loss": 0.9643, + "mean_token_accuracy": 0.7016899585723877, + "num_tokens": 391572257.0, + "step": 15404 + }, + { + "epoch": 1.6917417087634528, + "grad_norm": 2.144676923751831, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7188752293586731, + "num_tokens": 391599834.0, + "step": 15405 + }, + { + "epoch": 1.6918515264660663, + "grad_norm": 2.1964821815490723, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7150914669036865, + "num_tokens": 391627052.0, + "step": 15406 + }, + { + "epoch": 1.6919613441686798, + "grad_norm": 1.8981834650039673, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7000162601470947, + "num_tokens": 391661078.0, + "step": 15407 + }, + { + "epoch": 1.6920711618712936, + "grad_norm": 2.4343512058258057, + "learning_rate": 1e-06, + "loss": 0.8691, + "mean_token_accuracy": 0.7281137108802795, + "num_tokens": 391683159.0, + "step": 15408 + }, + { + "epoch": 1.6921809795739073, + "grad_norm": 2.180806875228882, + "learning_rate": 1e-06, + "loss": 0.976, + "mean_token_accuracy": 0.7094680666923523, + "num_tokens": 391714159.0, + "step": 15409 + }, + { + "epoch": 1.692290797276521, + "grad_norm": 1.9675710201263428, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7243540287017822, + "num_tokens": 391744653.0, + "step": 15410 + }, + { + "epoch": 1.6924006149791346, + "grad_norm": 1.9069011211395264, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7005912065505981, + "num_tokens": 391778660.0, + "step": 15411 + }, + { + "epoch": 1.6925104326817482, + "grad_norm": 2.286313056945801, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7276175022125244, + "num_tokens": 391801718.0, + "step": 15412 + }, + { + "epoch": 1.692620250384362, + "grad_norm": 2.2256312370300293, + "learning_rate": 1e-06, + "loss": 1.0069, + "mean_token_accuracy": 0.6990582346916199, + "num_tokens": 391829164.0, + "step": 15413 + }, + { + "epoch": 1.6927300680869757, + "grad_norm": 1.9207611083984375, + "learning_rate": 1e-06, + "loss": 0.8114, + "mean_token_accuracy": 0.7449314594268799, + "num_tokens": 391859664.0, + "step": 15414 + }, + { + "epoch": 1.6928398857895894, + "grad_norm": 2.41762375831604, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7333950996398926, + "num_tokens": 391881462.0, + "step": 15415 + }, + { + "epoch": 1.692949703492203, + "grad_norm": 2.0694468021392822, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7184077501296997, + "num_tokens": 391908065.0, + "step": 15416 + }, + { + "epoch": 1.6930595211948165, + "grad_norm": 2.370469331741333, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7336009740829468, + "num_tokens": 391929283.0, + "step": 15417 + }, + { + "epoch": 1.6931693388974303, + "grad_norm": 2.2327089309692383, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.727997899055481, + "num_tokens": 391954128.0, + "step": 15418 + }, + { + "epoch": 1.693279156600044, + "grad_norm": 2.1898820400238037, + "learning_rate": 1e-06, + "loss": 0.8509, + "mean_token_accuracy": 0.7371925115585327, + "num_tokens": 391979376.0, + "step": 15419 + }, + { + "epoch": 1.6933889743026576, + "grad_norm": 2.389622449874878, + "learning_rate": 1e-06, + "loss": 0.8324, + "mean_token_accuracy": 0.7335858941078186, + "num_tokens": 392000383.0, + "step": 15420 + }, + { + "epoch": 1.693498792005271, + "grad_norm": 2.8522849082946777, + "learning_rate": 1e-06, + "loss": 0.8019, + "mean_token_accuracy": 0.7426224946975708, + "num_tokens": 392017734.0, + "step": 15421 + }, + { + "epoch": 1.6936086097078848, + "grad_norm": 1.8608802556991577, + "learning_rate": 1e-06, + "loss": 0.9652, + "mean_token_accuracy": 0.6996413469314575, + "num_tokens": 392054168.0, + "step": 15422 + }, + { + "epoch": 1.6937184274104986, + "grad_norm": 2.3692071437835693, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7143768072128296, + "num_tokens": 392077350.0, + "step": 15423 + }, + { + "epoch": 1.6938282451131124, + "grad_norm": 2.4279398918151855, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.698192834854126, + "num_tokens": 392102250.0, + "step": 15424 + }, + { + "epoch": 1.693938062815726, + "grad_norm": 2.191331148147583, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7090968489646912, + "num_tokens": 392130536.0, + "step": 15425 + }, + { + "epoch": 1.6940478805183394, + "grad_norm": 2.1836636066436768, + "learning_rate": 1e-06, + "loss": 0.9506, + "mean_token_accuracy": 0.7048181295394897, + "num_tokens": 392156819.0, + "step": 15426 + }, + { + "epoch": 1.6941576982209532, + "grad_norm": 2.4524872303009033, + "learning_rate": 1e-06, + "loss": 0.8528, + "mean_token_accuracy": 0.7338125705718994, + "num_tokens": 392178862.0, + "step": 15427 + }, + { + "epoch": 1.694267515923567, + "grad_norm": 2.3922438621520996, + "learning_rate": 1e-06, + "loss": 0.7924, + "mean_token_accuracy": 0.7467969655990601, + "num_tokens": 392202070.0, + "step": 15428 + }, + { + "epoch": 1.6943773336261807, + "grad_norm": 2.490225076675415, + "learning_rate": 1e-06, + "loss": 0.85, + "mean_token_accuracy": 0.7299742698669434, + "num_tokens": 392224528.0, + "step": 15429 + }, + { + "epoch": 1.6944871513287942, + "grad_norm": 2.19744873046875, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7179591059684753, + "num_tokens": 392251178.0, + "step": 15430 + }, + { + "epoch": 1.6945969690314078, + "grad_norm": 2.2468271255493164, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7440352439880371, + "num_tokens": 392275774.0, + "step": 15431 + }, + { + "epoch": 1.6947067867340215, + "grad_norm": 2.174978733062744, + "learning_rate": 1e-06, + "loss": 0.9106, + "mean_token_accuracy": 0.7241476774215698, + "num_tokens": 392303930.0, + "step": 15432 + }, + { + "epoch": 1.6948166044366353, + "grad_norm": 2.1116859912872314, + "learning_rate": 1e-06, + "loss": 0.8879, + "mean_token_accuracy": 0.7229809761047363, + "num_tokens": 392330396.0, + "step": 15433 + }, + { + "epoch": 1.6949264221392488, + "grad_norm": 2.2192180156707764, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7208515405654907, + "num_tokens": 392355324.0, + "step": 15434 + }, + { + "epoch": 1.6950362398418624, + "grad_norm": 2.0720036029815674, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.6971031427383423, + "num_tokens": 392385100.0, + "step": 15435 + }, + { + "epoch": 1.6951460575444761, + "grad_norm": 2.3027656078338623, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7093623876571655, + "num_tokens": 392409226.0, + "step": 15436 + }, + { + "epoch": 1.6952558752470899, + "grad_norm": 2.47684645652771, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7119287252426147, + "num_tokens": 392431506.0, + "step": 15437 + }, + { + "epoch": 1.6953656929497036, + "grad_norm": 2.117521286010742, + "learning_rate": 1e-06, + "loss": 0.9451, + "mean_token_accuracy": 0.7058543562889099, + "num_tokens": 392456492.0, + "step": 15438 + }, + { + "epoch": 1.6954755106523172, + "grad_norm": 2.0758278369903564, + "learning_rate": 1e-06, + "loss": 0.9647, + "mean_token_accuracy": 0.7065435647964478, + "num_tokens": 392487757.0, + "step": 15439 + }, + { + "epoch": 1.6955853283549307, + "grad_norm": 2.1359875202178955, + "learning_rate": 1e-06, + "loss": 0.9075, + "mean_token_accuracy": 0.7185380458831787, + "num_tokens": 392513325.0, + "step": 15440 + }, + { + "epoch": 1.6956951460575445, + "grad_norm": 2.182149887084961, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7143199443817139, + "num_tokens": 392539811.0, + "step": 15441 + }, + { + "epoch": 1.6958049637601582, + "grad_norm": 2.731157064437866, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7283366918563843, + "num_tokens": 392558972.0, + "step": 15442 + }, + { + "epoch": 1.6959147814627717, + "grad_norm": 2.2409369945526123, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7249895930290222, + "num_tokens": 392583611.0, + "step": 15443 + }, + { + "epoch": 1.6960245991653855, + "grad_norm": 2.385214328765869, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7183859944343567, + "num_tokens": 392607328.0, + "step": 15444 + }, + { + "epoch": 1.696134416867999, + "grad_norm": 2.554497003555298, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7211692929267883, + "num_tokens": 392628216.0, + "step": 15445 + }, + { + "epoch": 1.6962442345706128, + "grad_norm": 2.241783380508423, + "learning_rate": 1e-06, + "loss": 0.7542, + "mean_token_accuracy": 0.7598102688789368, + "num_tokens": 392652545.0, + "step": 15446 + }, + { + "epoch": 1.6963540522732266, + "grad_norm": 2.095597505569458, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7213988304138184, + "num_tokens": 392678538.0, + "step": 15447 + }, + { + "epoch": 1.69646386997584, + "grad_norm": 2.2651054859161377, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.7389460802078247, + "num_tokens": 392701921.0, + "step": 15448 + }, + { + "epoch": 1.6965736876784536, + "grad_norm": 2.2728281021118164, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.716985821723938, + "num_tokens": 392726729.0, + "step": 15449 + }, + { + "epoch": 1.6966835053810674, + "grad_norm": 2.0267350673675537, + "learning_rate": 1e-06, + "loss": 0.8409, + "mean_token_accuracy": 0.7360663414001465, + "num_tokens": 392756489.0, + "step": 15450 + }, + { + "epoch": 1.6967933230836811, + "grad_norm": 2.1479547023773193, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7300200462341309, + "num_tokens": 392782172.0, + "step": 15451 + }, + { + "epoch": 1.696903140786295, + "grad_norm": 2.463224411010742, + "learning_rate": 1e-06, + "loss": 0.7752, + "mean_token_accuracy": 0.7543166279792786, + "num_tokens": 392802568.0, + "step": 15452 + }, + { + "epoch": 1.6970129584889084, + "grad_norm": 1.9927784204483032, + "learning_rate": 1e-06, + "loss": 0.9391, + "mean_token_accuracy": 0.7052721977233887, + "num_tokens": 392833462.0, + "step": 15453 + }, + { + "epoch": 1.697122776191522, + "grad_norm": 4.112300872802734, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7272788286209106, + "num_tokens": 392857932.0, + "step": 15454 + }, + { + "epoch": 1.6972325938941357, + "grad_norm": 2.2771081924438477, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7259398102760315, + "num_tokens": 392884413.0, + "step": 15455 + }, + { + "epoch": 1.6973424115967495, + "grad_norm": 2.429398775100708, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7213075160980225, + "num_tokens": 392906830.0, + "step": 15456 + }, + { + "epoch": 1.697452229299363, + "grad_norm": 2.611849784851074, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.7374513149261475, + "num_tokens": 392926562.0, + "step": 15457 + }, + { + "epoch": 1.6975620470019768, + "grad_norm": 2.0743191242218018, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7077886462211609, + "num_tokens": 392955280.0, + "step": 15458 + }, + { + "epoch": 1.6976718647045903, + "grad_norm": 2.4177260398864746, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.7035585045814514, + "num_tokens": 392977902.0, + "step": 15459 + }, + { + "epoch": 1.697781682407204, + "grad_norm": 2.412088394165039, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7388094067573547, + "num_tokens": 392999902.0, + "step": 15460 + }, + { + "epoch": 1.6978915001098178, + "grad_norm": 2.4765865802764893, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.7297665476799011, + "num_tokens": 393022540.0, + "step": 15461 + }, + { + "epoch": 1.6980013178124314, + "grad_norm": 2.2468421459198, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7134916186332703, + "num_tokens": 393047839.0, + "step": 15462 + }, + { + "epoch": 1.698111135515045, + "grad_norm": 2.577692985534668, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7235482335090637, + "num_tokens": 393070739.0, + "step": 15463 + }, + { + "epoch": 1.6982209532176586, + "grad_norm": 1.8629801273345947, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7116414308547974, + "num_tokens": 393102754.0, + "step": 15464 + }, + { + "epoch": 1.6983307709202724, + "grad_norm": 2.345586061477661, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7210783958435059, + "num_tokens": 393124983.0, + "step": 15465 + }, + { + "epoch": 1.6984405886228862, + "grad_norm": 2.2520079612731934, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7133415937423706, + "num_tokens": 393150002.0, + "step": 15466 + }, + { + "epoch": 1.6985504063254997, + "grad_norm": 2.2931289672851562, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7096685767173767, + "num_tokens": 393175000.0, + "step": 15467 + }, + { + "epoch": 1.6986602240281132, + "grad_norm": 2.2856051921844482, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7256661653518677, + "num_tokens": 393201766.0, + "step": 15468 + }, + { + "epoch": 1.698770041730727, + "grad_norm": 1.894797921180725, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7194157242774963, + "num_tokens": 393235509.0, + "step": 15469 + }, + { + "epoch": 1.6988798594333407, + "grad_norm": 1.9442099332809448, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7055611610412598, + "num_tokens": 393265649.0, + "step": 15470 + }, + { + "epoch": 1.6989896771359543, + "grad_norm": 2.3489928245544434, + "learning_rate": 1e-06, + "loss": 0.9306, + "mean_token_accuracy": 0.7106956243515015, + "num_tokens": 393289347.0, + "step": 15471 + }, + { + "epoch": 1.6990994948385678, + "grad_norm": 2.0289463996887207, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.70716792345047, + "num_tokens": 393318898.0, + "step": 15472 + }, + { + "epoch": 1.6992093125411816, + "grad_norm": 2.0985727310180664, + "learning_rate": 1e-06, + "loss": 0.9908, + "mean_token_accuracy": 0.7076963186264038, + "num_tokens": 393345927.0, + "step": 15473 + }, + { + "epoch": 1.6993191302437953, + "grad_norm": 2.761704683303833, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.714500904083252, + "num_tokens": 393363850.0, + "step": 15474 + }, + { + "epoch": 1.699428947946409, + "grad_norm": 2.3462181091308594, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7411607503890991, + "num_tokens": 393386151.0, + "step": 15475 + }, + { + "epoch": 1.6995387656490226, + "grad_norm": 2.1609833240509033, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7336176037788391, + "num_tokens": 393411337.0, + "step": 15476 + }, + { + "epoch": 1.6996485833516362, + "grad_norm": 2.3892805576324463, + "learning_rate": 1e-06, + "loss": 0.9544, + "mean_token_accuracy": 0.7058020234107971, + "num_tokens": 393435929.0, + "step": 15477 + }, + { + "epoch": 1.69975840105425, + "grad_norm": 2.381195068359375, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7224583625793457, + "num_tokens": 393457322.0, + "step": 15478 + }, + { + "epoch": 1.6998682187568637, + "grad_norm": 2.490568161010742, + "learning_rate": 1e-06, + "loss": 0.8222, + "mean_token_accuracy": 0.7384319305419922, + "num_tokens": 393477557.0, + "step": 15479 + }, + { + "epoch": 1.6999780364594774, + "grad_norm": 2.42454195022583, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.719221293926239, + "num_tokens": 393500697.0, + "step": 15480 + }, + { + "epoch": 1.700087854162091, + "grad_norm": 2.620598077774048, + "learning_rate": 1e-06, + "loss": 0.7889, + "mean_token_accuracy": 0.7480192184448242, + "num_tokens": 393519662.0, + "step": 15481 + }, + { + "epoch": 1.7001976718647045, + "grad_norm": 2.190228223800659, + "learning_rate": 1e-06, + "loss": 0.9067, + "mean_token_accuracy": 0.7148957252502441, + "num_tokens": 393546380.0, + "step": 15482 + }, + { + "epoch": 1.7003074895673183, + "grad_norm": 2.392117500305176, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7373679876327515, + "num_tokens": 393568683.0, + "step": 15483 + }, + { + "epoch": 1.700417307269932, + "grad_norm": 2.275096893310547, + "learning_rate": 1e-06, + "loss": 0.9371, + "mean_token_accuracy": 0.7085906863212585, + "num_tokens": 393596315.0, + "step": 15484 + }, + { + "epoch": 1.7005271249725455, + "grad_norm": 1.9295427799224854, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7203299403190613, + "num_tokens": 393626787.0, + "step": 15485 + }, + { + "epoch": 1.700636942675159, + "grad_norm": 1.932929515838623, + "learning_rate": 1e-06, + "loss": 0.9787, + "mean_token_accuracy": 0.7044813632965088, + "num_tokens": 393660305.0, + "step": 15486 + }, + { + "epoch": 1.7007467603777728, + "grad_norm": 2.358994722366333, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7231851816177368, + "num_tokens": 393685526.0, + "step": 15487 + }, + { + "epoch": 1.7008565780803866, + "grad_norm": 2.3911237716674805, + "learning_rate": 1e-06, + "loss": 0.8906, + "mean_token_accuracy": 0.7267599105834961, + "num_tokens": 393708260.0, + "step": 15488 + }, + { + "epoch": 1.7009663957830004, + "grad_norm": 2.521074056625366, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7311009168624878, + "num_tokens": 393731700.0, + "step": 15489 + }, + { + "epoch": 1.7010762134856139, + "grad_norm": 2.443683385848999, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7313438653945923, + "num_tokens": 393754821.0, + "step": 15490 + }, + { + "epoch": 1.7011860311882274, + "grad_norm": 2.020035743713379, + "learning_rate": 1e-06, + "loss": 0.9978, + "mean_token_accuracy": 0.6945340633392334, + "num_tokens": 393788437.0, + "step": 15491 + }, + { + "epoch": 1.7012958488908412, + "grad_norm": 2.153546094894409, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7263801097869873, + "num_tokens": 393812711.0, + "step": 15492 + }, + { + "epoch": 1.701405666593455, + "grad_norm": 2.160491704940796, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7344034314155579, + "num_tokens": 393837230.0, + "step": 15493 + }, + { + "epoch": 1.7015154842960687, + "grad_norm": 2.1921064853668213, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7326469421386719, + "num_tokens": 393865150.0, + "step": 15494 + }, + { + "epoch": 1.7016253019986822, + "grad_norm": 2.183480978012085, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7110158205032349, + "num_tokens": 393892664.0, + "step": 15495 + }, + { + "epoch": 1.7017351197012958, + "grad_norm": 2.329730987548828, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7128051519393921, + "num_tokens": 393917395.0, + "step": 15496 + }, + { + "epoch": 1.7018449374039095, + "grad_norm": 2.2016994953155518, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7285146713256836, + "num_tokens": 393942557.0, + "step": 15497 + }, + { + "epoch": 1.7019547551065233, + "grad_norm": 2.185105323791504, + "learning_rate": 1e-06, + "loss": 0.9809, + "mean_token_accuracy": 0.7150757312774658, + "num_tokens": 393967788.0, + "step": 15498 + }, + { + "epoch": 1.7020645728091368, + "grad_norm": 2.2948150634765625, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7495653629302979, + "num_tokens": 393991305.0, + "step": 15499 + }, + { + "epoch": 1.7021743905117503, + "grad_norm": 2.080861806869507, + "learning_rate": 1e-06, + "loss": 0.9881, + "mean_token_accuracy": 0.6987398266792297, + "num_tokens": 394018639.0, + "step": 15500 + }, + { + "epoch": 1.702284208214364, + "grad_norm": 2.2239012718200684, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7160485982894897, + "num_tokens": 394043562.0, + "step": 15501 + }, + { + "epoch": 1.7023940259169779, + "grad_norm": 2.2665951251983643, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7439277768135071, + "num_tokens": 394067113.0, + "step": 15502 + }, + { + "epoch": 1.7025038436195916, + "grad_norm": 2.0097014904022217, + "learning_rate": 1e-06, + "loss": 1.0036, + "mean_token_accuracy": 0.689195990562439, + "num_tokens": 394098487.0, + "step": 15503 + }, + { + "epoch": 1.7026136613222052, + "grad_norm": 2.3417718410491943, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7149573564529419, + "num_tokens": 394123798.0, + "step": 15504 + }, + { + "epoch": 1.7027234790248187, + "grad_norm": 2.6998794078826904, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7290791273117065, + "num_tokens": 394143198.0, + "step": 15505 + }, + { + "epoch": 1.7028332967274324, + "grad_norm": 2.6401071548461914, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7282220125198364, + "num_tokens": 394164260.0, + "step": 15506 + }, + { + "epoch": 1.7029431144300462, + "grad_norm": 2.0261526107788086, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7140847444534302, + "num_tokens": 394193724.0, + "step": 15507 + }, + { + "epoch": 1.7030529321326597, + "grad_norm": 2.3015313148498535, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7162648439407349, + "num_tokens": 394220298.0, + "step": 15508 + }, + { + "epoch": 1.7031627498352735, + "grad_norm": 2.4120609760284424, + "learning_rate": 1e-06, + "loss": 0.816, + "mean_token_accuracy": 0.7414262294769287, + "num_tokens": 394242911.0, + "step": 15509 + }, + { + "epoch": 1.703272567537887, + "grad_norm": 2.2673351764678955, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7401586174964905, + "num_tokens": 394265968.0, + "step": 15510 + }, + { + "epoch": 1.7033823852405008, + "grad_norm": 1.8799078464508057, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7156635522842407, + "num_tokens": 394301424.0, + "step": 15511 + }, + { + "epoch": 1.7034922029431145, + "grad_norm": 2.505134105682373, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7332364320755005, + "num_tokens": 394324347.0, + "step": 15512 + }, + { + "epoch": 1.703602020645728, + "grad_norm": 2.666411876678467, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7375850677490234, + "num_tokens": 394342043.0, + "step": 15513 + }, + { + "epoch": 1.7037118383483416, + "grad_norm": 2.3532779216766357, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7217094898223877, + "num_tokens": 394366925.0, + "step": 15514 + }, + { + "epoch": 1.7038216560509554, + "grad_norm": 2.24245023727417, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.734700083732605, + "num_tokens": 394390628.0, + "step": 15515 + }, + { + "epoch": 1.7039314737535691, + "grad_norm": 2.2008779048919678, + "learning_rate": 1e-06, + "loss": 0.9461, + "mean_token_accuracy": 0.7061469554901123, + "num_tokens": 394416128.0, + "step": 15516 + }, + { + "epoch": 1.7040412914561829, + "grad_norm": 2.332676887512207, + "learning_rate": 1e-06, + "loss": 0.8789, + "mean_token_accuracy": 0.7217351794242859, + "num_tokens": 394437819.0, + "step": 15517 + }, + { + "epoch": 1.7041511091587964, + "grad_norm": 2.3099677562713623, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.7161406874656677, + "num_tokens": 394462611.0, + "step": 15518 + }, + { + "epoch": 1.70426092686141, + "grad_norm": 2.6084208488464355, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7327585220336914, + "num_tokens": 394481391.0, + "step": 15519 + }, + { + "epoch": 1.7043707445640237, + "grad_norm": 2.274714708328247, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7134435176849365, + "num_tokens": 394507639.0, + "step": 15520 + }, + { + "epoch": 1.7044805622666375, + "grad_norm": 2.3023757934570312, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.718073844909668, + "num_tokens": 394531428.0, + "step": 15521 + }, + { + "epoch": 1.704590379969251, + "grad_norm": 2.189600706100464, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7243995666503906, + "num_tokens": 394556457.0, + "step": 15522 + }, + { + "epoch": 1.7047001976718648, + "grad_norm": 2.1409831047058105, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7110694050788879, + "num_tokens": 394585807.0, + "step": 15523 + }, + { + "epoch": 1.7048100153744783, + "grad_norm": 2.389508008956909, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7102590799331665, + "num_tokens": 394610497.0, + "step": 15524 + }, + { + "epoch": 1.704919833077092, + "grad_norm": 2.003279447555542, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7185173630714417, + "num_tokens": 394642926.0, + "step": 15525 + }, + { + "epoch": 1.7050296507797058, + "grad_norm": 2.501040458679199, + "learning_rate": 1e-06, + "loss": 0.8864, + "mean_token_accuracy": 0.7242883443832397, + "num_tokens": 394663673.0, + "step": 15526 + }, + { + "epoch": 1.7051394684823193, + "grad_norm": 2.293147325515747, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.71510249376297, + "num_tokens": 394688614.0, + "step": 15527 + }, + { + "epoch": 1.7052492861849329, + "grad_norm": 2.295422077178955, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7297070622444153, + "num_tokens": 394710926.0, + "step": 15528 + }, + { + "epoch": 1.7053591038875466, + "grad_norm": 2.395994186401367, + "learning_rate": 1e-06, + "loss": 0.9617, + "mean_token_accuracy": 0.712411642074585, + "num_tokens": 394732314.0, + "step": 15529 + }, + { + "epoch": 1.7054689215901604, + "grad_norm": 2.2102110385894775, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7181215286254883, + "num_tokens": 394758705.0, + "step": 15530 + }, + { + "epoch": 1.7055787392927741, + "grad_norm": 2.364849328994751, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.7273024916648865, + "num_tokens": 394781876.0, + "step": 15531 + }, + { + "epoch": 1.7056885569953877, + "grad_norm": 1.960832118988037, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7123358249664307, + "num_tokens": 394815946.0, + "step": 15532 + }, + { + "epoch": 1.7057983746980012, + "grad_norm": 2.273670196533203, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.719096302986145, + "num_tokens": 394840971.0, + "step": 15533 + }, + { + "epoch": 1.705908192400615, + "grad_norm": 2.2004034519195557, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7138094902038574, + "num_tokens": 394867482.0, + "step": 15534 + }, + { + "epoch": 1.7060180101032287, + "grad_norm": 2.3755011558532715, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7337822914123535, + "num_tokens": 394892115.0, + "step": 15535 + }, + { + "epoch": 1.7061278278058423, + "grad_norm": 2.074486017227173, + "learning_rate": 1e-06, + "loss": 1.0146, + "mean_token_accuracy": 0.6911740303039551, + "num_tokens": 394924087.0, + "step": 15536 + }, + { + "epoch": 1.7062376455084558, + "grad_norm": 2.524296998977661, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7316357493400574, + "num_tokens": 394943964.0, + "step": 15537 + }, + { + "epoch": 1.7063474632110696, + "grad_norm": 2.3890061378479004, + "learning_rate": 1e-06, + "loss": 0.9368, + "mean_token_accuracy": 0.7216581106185913, + "num_tokens": 394967468.0, + "step": 15538 + }, + { + "epoch": 1.7064572809136833, + "grad_norm": 2.2817933559417725, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7197079062461853, + "num_tokens": 394991396.0, + "step": 15539 + }, + { + "epoch": 1.706567098616297, + "grad_norm": 2.613602638244629, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7179224491119385, + "num_tokens": 395011500.0, + "step": 15540 + }, + { + "epoch": 1.7066769163189106, + "grad_norm": 2.036769151687622, + "learning_rate": 1e-06, + "loss": 0.8336, + "mean_token_accuracy": 0.7347408533096313, + "num_tokens": 395043111.0, + "step": 15541 + }, + { + "epoch": 1.7067867340215241, + "grad_norm": 2.0900280475616455, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7217031717300415, + "num_tokens": 395073043.0, + "step": 15542 + }, + { + "epoch": 1.706896551724138, + "grad_norm": 2.424217700958252, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7257140874862671, + "num_tokens": 395094466.0, + "step": 15543 + }, + { + "epoch": 1.7070063694267517, + "grad_norm": 2.330259084701538, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7060260772705078, + "num_tokens": 395119923.0, + "step": 15544 + }, + { + "epoch": 1.7071161871293654, + "grad_norm": 2.4813268184661865, + "learning_rate": 1e-06, + "loss": 0.8946, + "mean_token_accuracy": 0.7195411920547485, + "num_tokens": 395142426.0, + "step": 15545 + }, + { + "epoch": 1.707226004831979, + "grad_norm": 2.4245951175689697, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7232546210289001, + "num_tokens": 395163607.0, + "step": 15546 + }, + { + "epoch": 1.7073358225345925, + "grad_norm": 2.032097101211548, + "learning_rate": 1e-06, + "loss": 0.8123, + "mean_token_accuracy": 0.7415783405303955, + "num_tokens": 395189459.0, + "step": 15547 + }, + { + "epoch": 1.7074456402372062, + "grad_norm": 2.302859306335449, + "learning_rate": 1e-06, + "loss": 0.8398, + "mean_token_accuracy": 0.7295992374420166, + "num_tokens": 395213707.0, + "step": 15548 + }, + { + "epoch": 1.70755545793982, + "grad_norm": 2.1699798107147217, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7014144659042358, + "num_tokens": 395241507.0, + "step": 15549 + }, + { + "epoch": 1.7076652756424335, + "grad_norm": 2.4386141300201416, + "learning_rate": 1e-06, + "loss": 0.8584, + "mean_token_accuracy": 0.735386312007904, + "num_tokens": 395260654.0, + "step": 15550 + }, + { + "epoch": 1.707775093345047, + "grad_norm": 2.523338556289673, + "learning_rate": 1e-06, + "loss": 0.8471, + "mean_token_accuracy": 0.7352989912033081, + "num_tokens": 395281388.0, + "step": 15551 + }, + { + "epoch": 1.7078849110476608, + "grad_norm": 2.1281838417053223, + "learning_rate": 1e-06, + "loss": 0.8849, + "mean_token_accuracy": 0.7179021239280701, + "num_tokens": 395307109.0, + "step": 15552 + }, + { + "epoch": 1.7079947287502746, + "grad_norm": 2.163066864013672, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7117196321487427, + "num_tokens": 395333856.0, + "step": 15553 + }, + { + "epoch": 1.7081045464528883, + "grad_norm": 2.4548192024230957, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7354267835617065, + "num_tokens": 395355963.0, + "step": 15554 + }, + { + "epoch": 1.7082143641555019, + "grad_norm": 2.4361765384674072, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7357636094093323, + "num_tokens": 395379926.0, + "step": 15555 + }, + { + "epoch": 1.7083241818581154, + "grad_norm": 2.4115912914276123, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7274326086044312, + "num_tokens": 395401356.0, + "step": 15556 + }, + { + "epoch": 1.7084339995607292, + "grad_norm": 2.577718734741211, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7357054948806763, + "num_tokens": 395420182.0, + "step": 15557 + }, + { + "epoch": 1.708543817263343, + "grad_norm": 2.2506604194641113, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.7037414312362671, + "num_tokens": 395445106.0, + "step": 15558 + }, + { + "epoch": 1.7086536349659565, + "grad_norm": 2.2786686420440674, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7181844711303711, + "num_tokens": 395471665.0, + "step": 15559 + }, + { + "epoch": 1.7087634526685702, + "grad_norm": 2.3394455909729004, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7332673072814941, + "num_tokens": 395495441.0, + "step": 15560 + }, + { + "epoch": 1.7088732703711838, + "grad_norm": 2.2939813137054443, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.722029447555542, + "num_tokens": 395519873.0, + "step": 15561 + }, + { + "epoch": 1.7089830880737975, + "grad_norm": 2.3815579414367676, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7187922596931458, + "num_tokens": 395543102.0, + "step": 15562 + }, + { + "epoch": 1.7090929057764113, + "grad_norm": 2.1850666999816895, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.7150328755378723, + "num_tokens": 395568491.0, + "step": 15563 + }, + { + "epoch": 1.7092027234790248, + "grad_norm": 1.9573462009429932, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.711630642414093, + "num_tokens": 395598629.0, + "step": 15564 + }, + { + "epoch": 1.7093125411816383, + "grad_norm": 2.1774022579193115, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7410550117492676, + "num_tokens": 395623705.0, + "step": 15565 + }, + { + "epoch": 1.709422358884252, + "grad_norm": 2.3398313522338867, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7308014631271362, + "num_tokens": 395647942.0, + "step": 15566 + }, + { + "epoch": 1.7095321765868658, + "grad_norm": 1.949181079864502, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7128652930259705, + "num_tokens": 395680309.0, + "step": 15567 + }, + { + "epoch": 1.7096419942894796, + "grad_norm": 2.353698968887329, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6966691017150879, + "num_tokens": 395706056.0, + "step": 15568 + }, + { + "epoch": 1.7097518119920931, + "grad_norm": 2.1765999794006348, + "learning_rate": 1e-06, + "loss": 1.0057, + "mean_token_accuracy": 0.6870777606964111, + "num_tokens": 395732486.0, + "step": 15569 + }, + { + "epoch": 1.7098616296947067, + "grad_norm": 2.571300745010376, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7287750840187073, + "num_tokens": 395754661.0, + "step": 15570 + }, + { + "epoch": 1.7099714473973204, + "grad_norm": 2.581327438354492, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7280468344688416, + "num_tokens": 395774652.0, + "step": 15571 + }, + { + "epoch": 1.7100812650999342, + "grad_norm": 2.135723352432251, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7022918462753296, + "num_tokens": 395802616.0, + "step": 15572 + }, + { + "epoch": 1.7101910828025477, + "grad_norm": 2.448050022125244, + "learning_rate": 1e-06, + "loss": 0.82, + "mean_token_accuracy": 0.74111008644104, + "num_tokens": 395823568.0, + "step": 15573 + }, + { + "epoch": 1.7103009005051615, + "grad_norm": 2.2876267433166504, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7152828574180603, + "num_tokens": 395849025.0, + "step": 15574 + }, + { + "epoch": 1.710410718207775, + "grad_norm": 2.7303576469421387, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7229270935058594, + "num_tokens": 395868179.0, + "step": 15575 + }, + { + "epoch": 1.7105205359103888, + "grad_norm": 2.507891893386841, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7389495968818665, + "num_tokens": 395890308.0, + "step": 15576 + }, + { + "epoch": 1.7106303536130025, + "grad_norm": 2.160020112991333, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.726546585559845, + "num_tokens": 395917664.0, + "step": 15577 + }, + { + "epoch": 1.710740171315616, + "grad_norm": 2.088864803314209, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7242750525474548, + "num_tokens": 395946391.0, + "step": 15578 + }, + { + "epoch": 1.7108499890182296, + "grad_norm": 2.079364061355591, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7372482419013977, + "num_tokens": 395973113.0, + "step": 15579 + }, + { + "epoch": 1.7109598067208434, + "grad_norm": 2.2838265895843506, + "learning_rate": 1e-06, + "loss": 0.8988, + "mean_token_accuracy": 0.7261946201324463, + "num_tokens": 395999004.0, + "step": 15580 + }, + { + "epoch": 1.7110696244234571, + "grad_norm": 2.481299638748169, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7294082641601562, + "num_tokens": 396021657.0, + "step": 15581 + }, + { + "epoch": 1.7111794421260709, + "grad_norm": 2.432786703109741, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7319422960281372, + "num_tokens": 396043613.0, + "step": 15582 + }, + { + "epoch": 1.7112892598286844, + "grad_norm": 2.189635753631592, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.729058027267456, + "num_tokens": 396069645.0, + "step": 15583 + }, + { + "epoch": 1.711399077531298, + "grad_norm": 2.1792898178100586, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7193033695220947, + "num_tokens": 396096747.0, + "step": 15584 + }, + { + "epoch": 1.7115088952339117, + "grad_norm": 2.3390238285064697, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7102752327919006, + "num_tokens": 396123517.0, + "step": 15585 + }, + { + "epoch": 1.7116187129365255, + "grad_norm": 1.9972202777862549, + "learning_rate": 1e-06, + "loss": 0.9529, + "mean_token_accuracy": 0.7167102098464966, + "num_tokens": 396153088.0, + "step": 15586 + }, + { + "epoch": 1.711728530639139, + "grad_norm": 2.1166718006134033, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7365881204605103, + "num_tokens": 396179366.0, + "step": 15587 + }, + { + "epoch": 1.7118383483417525, + "grad_norm": 2.0736498832702637, + "learning_rate": 1e-06, + "loss": 0.7761, + "mean_token_accuracy": 0.752001166343689, + "num_tokens": 396205458.0, + "step": 15588 + }, + { + "epoch": 1.7119481660443663, + "grad_norm": 2.3657426834106445, + "learning_rate": 1e-06, + "loss": 1.0179, + "mean_token_accuracy": 0.6876023411750793, + "num_tokens": 396231395.0, + "step": 15589 + }, + { + "epoch": 1.71205798374698, + "grad_norm": 2.075885772705078, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7197897434234619, + "num_tokens": 396262687.0, + "step": 15590 + }, + { + "epoch": 1.7121678014495938, + "grad_norm": 2.2715280055999756, + "learning_rate": 1e-06, + "loss": 0.7701, + "mean_token_accuracy": 0.7508988380432129, + "num_tokens": 396286915.0, + "step": 15591 + }, + { + "epoch": 1.7122776191522073, + "grad_norm": 2.1117167472839355, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7175494432449341, + "num_tokens": 396313580.0, + "step": 15592 + }, + { + "epoch": 1.7123874368548209, + "grad_norm": 2.459040880203247, + "learning_rate": 1e-06, + "loss": 0.9767, + "mean_token_accuracy": 0.7135570049285889, + "num_tokens": 396336597.0, + "step": 15593 + }, + { + "epoch": 1.7124972545574346, + "grad_norm": 2.1763134002685547, + "learning_rate": 1e-06, + "loss": 0.8326, + "mean_token_accuracy": 0.7406679391860962, + "num_tokens": 396361194.0, + "step": 15594 + }, + { + "epoch": 1.7126070722600484, + "grad_norm": 2.3922531604766846, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7350287437438965, + "num_tokens": 396382152.0, + "step": 15595 + }, + { + "epoch": 1.7127168899626621, + "grad_norm": 2.048572301864624, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7133466005325317, + "num_tokens": 396410463.0, + "step": 15596 + }, + { + "epoch": 1.7128267076652757, + "grad_norm": 2.063075304031372, + "learning_rate": 1e-06, + "loss": 0.9189, + "mean_token_accuracy": 0.7263522148132324, + "num_tokens": 396440148.0, + "step": 15597 + }, + { + "epoch": 1.7129365253678892, + "grad_norm": 2.1007282733917236, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.697525143623352, + "num_tokens": 396471050.0, + "step": 15598 + }, + { + "epoch": 1.713046343070503, + "grad_norm": 2.397393226623535, + "learning_rate": 1e-06, + "loss": 0.9719, + "mean_token_accuracy": 0.697044312953949, + "num_tokens": 396494462.0, + "step": 15599 + }, + { + "epoch": 1.7131561607731167, + "grad_norm": 2.150228261947632, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7382737994194031, + "num_tokens": 396522447.0, + "step": 15600 + }, + { + "epoch": 1.7132659784757303, + "grad_norm": 2.4765210151672363, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.705371618270874, + "num_tokens": 396545822.0, + "step": 15601 + }, + { + "epoch": 1.7133757961783438, + "grad_norm": 2.261181592941284, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7183812856674194, + "num_tokens": 396572066.0, + "step": 15602 + }, + { + "epoch": 1.7134856138809575, + "grad_norm": 2.208742380142212, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7168217897415161, + "num_tokens": 396598195.0, + "step": 15603 + }, + { + "epoch": 1.7135954315835713, + "grad_norm": 2.1264543533325195, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7265084981918335, + "num_tokens": 396629694.0, + "step": 15604 + }, + { + "epoch": 1.713705249286185, + "grad_norm": 1.8091270923614502, + "learning_rate": 1e-06, + "loss": 0.9765, + "mean_token_accuracy": 0.6992914080619812, + "num_tokens": 396668934.0, + "step": 15605 + }, + { + "epoch": 1.7138150669887986, + "grad_norm": 2.3350846767425537, + "learning_rate": 1e-06, + "loss": 0.9338, + "mean_token_accuracy": 0.709181547164917, + "num_tokens": 396694153.0, + "step": 15606 + }, + { + "epoch": 1.7139248846914121, + "grad_norm": 2.165642738342285, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7249183654785156, + "num_tokens": 396720495.0, + "step": 15607 + }, + { + "epoch": 1.7140347023940259, + "grad_norm": 2.0953786373138428, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7279638648033142, + "num_tokens": 396746628.0, + "step": 15608 + }, + { + "epoch": 1.7141445200966396, + "grad_norm": 2.343341588973999, + "learning_rate": 1e-06, + "loss": 1.0479, + "mean_token_accuracy": 0.6813881397247314, + "num_tokens": 396771758.0, + "step": 15609 + }, + { + "epoch": 1.7142543377992534, + "grad_norm": 2.2672626972198486, + "learning_rate": 1e-06, + "loss": 0.9624, + "mean_token_accuracy": 0.710940420627594, + "num_tokens": 396798160.0, + "step": 15610 + }, + { + "epoch": 1.714364155501867, + "grad_norm": 2.3912060260772705, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7235744595527649, + "num_tokens": 396820885.0, + "step": 15611 + }, + { + "epoch": 1.7144739732044805, + "grad_norm": 2.52921199798584, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.6918681859970093, + "num_tokens": 396842332.0, + "step": 15612 + }, + { + "epoch": 1.7145837909070942, + "grad_norm": 2.217810869216919, + "learning_rate": 1e-06, + "loss": 0.9001, + "mean_token_accuracy": 0.7118104100227356, + "num_tokens": 396869053.0, + "step": 15613 + }, + { + "epoch": 1.714693608609708, + "grad_norm": 2.3341832160949707, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.7133368253707886, + "num_tokens": 396895397.0, + "step": 15614 + }, + { + "epoch": 1.7148034263123215, + "grad_norm": 2.2500298023223877, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7165932655334473, + "num_tokens": 396923102.0, + "step": 15615 + }, + { + "epoch": 1.714913244014935, + "grad_norm": 2.369295120239258, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.739732027053833, + "num_tokens": 396944934.0, + "step": 15616 + }, + { + "epoch": 1.7150230617175488, + "grad_norm": 2.1065726280212402, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.717963695526123, + "num_tokens": 396974502.0, + "step": 15617 + }, + { + "epoch": 1.7151328794201626, + "grad_norm": 2.062250852584839, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.7008564472198486, + "num_tokens": 397003729.0, + "step": 15618 + }, + { + "epoch": 1.7152426971227763, + "grad_norm": 2.3333632946014404, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7360852360725403, + "num_tokens": 397027436.0, + "step": 15619 + }, + { + "epoch": 1.7153525148253899, + "grad_norm": 2.323021173477173, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7368316650390625, + "num_tokens": 397051877.0, + "step": 15620 + }, + { + "epoch": 1.7154623325280034, + "grad_norm": 2.506631374359131, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7240135669708252, + "num_tokens": 397072886.0, + "step": 15621 + }, + { + "epoch": 1.7155721502306172, + "grad_norm": 1.9520803689956665, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7037477493286133, + "num_tokens": 397106592.0, + "step": 15622 + }, + { + "epoch": 1.715681967933231, + "grad_norm": 2.1503992080688477, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7292804718017578, + "num_tokens": 397133030.0, + "step": 15623 + }, + { + "epoch": 1.7157917856358444, + "grad_norm": 2.1102054119110107, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7199099063873291, + "num_tokens": 397161186.0, + "step": 15624 + }, + { + "epoch": 1.7159016033384582, + "grad_norm": 1.9798287153244019, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7322166562080383, + "num_tokens": 397190024.0, + "step": 15625 + }, + { + "epoch": 1.7160114210410717, + "grad_norm": 2.2686469554901123, + "learning_rate": 1e-06, + "loss": 0.9109, + "mean_token_accuracy": 0.7115617394447327, + "num_tokens": 397214521.0, + "step": 15626 + }, + { + "epoch": 1.7161212387436855, + "grad_norm": 2.685333490371704, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7253519296646118, + "num_tokens": 397234237.0, + "step": 15627 + }, + { + "epoch": 1.7162310564462993, + "grad_norm": 2.564980983734131, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7200961709022522, + "num_tokens": 397256856.0, + "step": 15628 + }, + { + "epoch": 1.7163408741489128, + "grad_norm": 2.518434524536133, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.730215311050415, + "num_tokens": 397279252.0, + "step": 15629 + }, + { + "epoch": 1.7164506918515263, + "grad_norm": 2.2864983081817627, + "learning_rate": 1e-06, + "loss": 0.9275, + "mean_token_accuracy": 0.715042233467102, + "num_tokens": 397306186.0, + "step": 15630 + }, + { + "epoch": 1.71656050955414, + "grad_norm": 2.6638875007629395, + "learning_rate": 1e-06, + "loss": 0.8638, + "mean_token_accuracy": 0.7335199117660522, + "num_tokens": 397326736.0, + "step": 15631 + }, + { + "epoch": 1.7166703272567538, + "grad_norm": 1.9895068407058716, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7171967029571533, + "num_tokens": 397355265.0, + "step": 15632 + }, + { + "epoch": 1.7167801449593676, + "grad_norm": 2.1003644466400146, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.704555869102478, + "num_tokens": 397384236.0, + "step": 15633 + }, + { + "epoch": 1.7168899626619811, + "grad_norm": 2.3655545711517334, + "learning_rate": 1e-06, + "loss": 0.8251, + "mean_token_accuracy": 0.7320042848587036, + "num_tokens": 397406605.0, + "step": 15634 + }, + { + "epoch": 1.7169997803645947, + "grad_norm": 2.2475950717926025, + "learning_rate": 1e-06, + "loss": 0.9168, + "mean_token_accuracy": 0.7120516300201416, + "num_tokens": 397431555.0, + "step": 15635 + }, + { + "epoch": 1.7171095980672084, + "grad_norm": 2.0991320610046387, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.6985477209091187, + "num_tokens": 397459960.0, + "step": 15636 + }, + { + "epoch": 1.7172194157698222, + "grad_norm": 2.172060012817383, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7204325795173645, + "num_tokens": 397487061.0, + "step": 15637 + }, + { + "epoch": 1.7173292334724357, + "grad_norm": 2.6270036697387695, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.734549343585968, + "num_tokens": 397505734.0, + "step": 15638 + }, + { + "epoch": 1.7174390511750495, + "grad_norm": 2.188305616378784, + "learning_rate": 1e-06, + "loss": 0.9741, + "mean_token_accuracy": 0.7001727819442749, + "num_tokens": 397532402.0, + "step": 15639 + }, + { + "epoch": 1.717548868877663, + "grad_norm": 2.0196890830993652, + "learning_rate": 1e-06, + "loss": 0.9929, + "mean_token_accuracy": 0.6906934380531311, + "num_tokens": 397562292.0, + "step": 15640 + }, + { + "epoch": 1.7176586865802768, + "grad_norm": 2.36933970451355, + "learning_rate": 1e-06, + "loss": 0.8796, + "mean_token_accuracy": 0.7227864861488342, + "num_tokens": 397586700.0, + "step": 15641 + }, + { + "epoch": 1.7177685042828905, + "grad_norm": 2.1353673934936523, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7201536893844604, + "num_tokens": 397615096.0, + "step": 15642 + }, + { + "epoch": 1.717878321985504, + "grad_norm": 2.1672141551971436, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7267720699310303, + "num_tokens": 397642636.0, + "step": 15643 + }, + { + "epoch": 1.7179881396881176, + "grad_norm": 2.471208095550537, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7224279642105103, + "num_tokens": 397664347.0, + "step": 15644 + }, + { + "epoch": 1.7180979573907313, + "grad_norm": 2.044090509414673, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.6991348266601562, + "num_tokens": 397695037.0, + "step": 15645 + }, + { + "epoch": 1.718207775093345, + "grad_norm": 1.9106682538986206, + "learning_rate": 1e-06, + "loss": 0.8952, + "mean_token_accuracy": 0.7182570695877075, + "num_tokens": 397728649.0, + "step": 15646 + }, + { + "epoch": 1.7183175927959589, + "grad_norm": 2.665111541748047, + "learning_rate": 1e-06, + "loss": 0.8249, + "mean_token_accuracy": 0.734139084815979, + "num_tokens": 397748269.0, + "step": 15647 + }, + { + "epoch": 1.7184274104985724, + "grad_norm": 2.333681583404541, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7153197526931763, + "num_tokens": 397771482.0, + "step": 15648 + }, + { + "epoch": 1.718537228201186, + "grad_norm": 2.5034360885620117, + "learning_rate": 1e-06, + "loss": 0.7804, + "mean_token_accuracy": 0.7517606019973755, + "num_tokens": 397791343.0, + "step": 15649 + }, + { + "epoch": 1.7186470459037997, + "grad_norm": 2.3588836193084717, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7533838748931885, + "num_tokens": 397813932.0, + "step": 15650 + }, + { + "epoch": 1.7187568636064134, + "grad_norm": 2.3611481189727783, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7104923129081726, + "num_tokens": 397838866.0, + "step": 15651 + }, + { + "epoch": 1.718866681309027, + "grad_norm": 2.3613762855529785, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7176389694213867, + "num_tokens": 397861238.0, + "step": 15652 + }, + { + "epoch": 1.7189764990116405, + "grad_norm": 2.231415033340454, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7070763111114502, + "num_tokens": 397886192.0, + "step": 15653 + }, + { + "epoch": 1.7190863167142543, + "grad_norm": 2.3889050483703613, + "learning_rate": 1e-06, + "loss": 0.8619, + "mean_token_accuracy": 0.7215886116027832, + "num_tokens": 397908032.0, + "step": 15654 + }, + { + "epoch": 1.719196134416868, + "grad_norm": 2.1239147186279297, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.7176250219345093, + "num_tokens": 397934493.0, + "step": 15655 + }, + { + "epoch": 1.7193059521194818, + "grad_norm": 2.1572091579437256, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7017531394958496, + "num_tokens": 397962459.0, + "step": 15656 + }, + { + "epoch": 1.7194157698220953, + "grad_norm": 2.308875799179077, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7207186222076416, + "num_tokens": 397986733.0, + "step": 15657 + }, + { + "epoch": 1.7195255875247089, + "grad_norm": 2.239858388900757, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7201817035675049, + "num_tokens": 398013078.0, + "step": 15658 + }, + { + "epoch": 1.7196354052273226, + "grad_norm": 2.152994155883789, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7114574909210205, + "num_tokens": 398042169.0, + "step": 15659 + }, + { + "epoch": 1.7197452229299364, + "grad_norm": 2.0984201431274414, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7066460847854614, + "num_tokens": 398070657.0, + "step": 15660 + }, + { + "epoch": 1.7198550406325501, + "grad_norm": 2.3839690685272217, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7215480208396912, + "num_tokens": 398093759.0, + "step": 15661 + }, + { + "epoch": 1.7199648583351637, + "grad_norm": 2.6944832801818848, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7329322099685669, + "num_tokens": 398112766.0, + "step": 15662 + }, + { + "epoch": 1.7200746760377772, + "grad_norm": 2.393897533416748, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.7272260785102844, + "num_tokens": 398135558.0, + "step": 15663 + }, + { + "epoch": 1.720184493740391, + "grad_norm": 2.086864709854126, + "learning_rate": 1e-06, + "loss": 1.0793, + "mean_token_accuracy": 0.6900314092636108, + "num_tokens": 398166780.0, + "step": 15664 + }, + { + "epoch": 1.7202943114430047, + "grad_norm": 2.4156126976013184, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7152931690216064, + "num_tokens": 398190104.0, + "step": 15665 + }, + { + "epoch": 1.7204041291456182, + "grad_norm": 2.2925684452056885, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7044520378112793, + "num_tokens": 398214787.0, + "step": 15666 + }, + { + "epoch": 1.7205139468482318, + "grad_norm": 1.8934725522994995, + "learning_rate": 1e-06, + "loss": 0.9912, + "mean_token_accuracy": 0.6924310326576233, + "num_tokens": 398248450.0, + "step": 15667 + }, + { + "epoch": 1.7206237645508455, + "grad_norm": 2.5972509384155273, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7269224524497986, + "num_tokens": 398269555.0, + "step": 15668 + }, + { + "epoch": 1.7207335822534593, + "grad_norm": 1.9619626998901367, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7122675180435181, + "num_tokens": 398303233.0, + "step": 15669 + }, + { + "epoch": 1.720843399956073, + "grad_norm": 2.287522792816162, + "learning_rate": 1e-06, + "loss": 0.8745, + "mean_token_accuracy": 0.7269296646118164, + "num_tokens": 398327568.0, + "step": 15670 + }, + { + "epoch": 1.7209532176586866, + "grad_norm": 2.3085439205169678, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.713303804397583, + "num_tokens": 398352620.0, + "step": 15671 + }, + { + "epoch": 1.7210630353613001, + "grad_norm": 1.9998409748077393, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7065260410308838, + "num_tokens": 398384438.0, + "step": 15672 + }, + { + "epoch": 1.7211728530639139, + "grad_norm": 2.266247034072876, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.713326096534729, + "num_tokens": 398410688.0, + "step": 15673 + }, + { + "epoch": 1.7212826707665276, + "grad_norm": 2.2354278564453125, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7088937759399414, + "num_tokens": 398437068.0, + "step": 15674 + }, + { + "epoch": 1.7213924884691414, + "grad_norm": 2.1573424339294434, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.6996607184410095, + "num_tokens": 398466531.0, + "step": 15675 + }, + { + "epoch": 1.721502306171755, + "grad_norm": 2.1806585788726807, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7304628491401672, + "num_tokens": 398494505.0, + "step": 15676 + }, + { + "epoch": 1.7216121238743685, + "grad_norm": 2.252987861633301, + "learning_rate": 1e-06, + "loss": 0.8029, + "mean_token_accuracy": 0.7406656742095947, + "num_tokens": 398519518.0, + "step": 15677 + }, + { + "epoch": 1.7217219415769822, + "grad_norm": 2.4318132400512695, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7230567932128906, + "num_tokens": 398540211.0, + "step": 15678 + }, + { + "epoch": 1.721831759279596, + "grad_norm": 2.4869072437286377, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.735449492931366, + "num_tokens": 398560312.0, + "step": 15679 + }, + { + "epoch": 1.7219415769822095, + "grad_norm": 2.211379051208496, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7230151295661926, + "num_tokens": 398586197.0, + "step": 15680 + }, + { + "epoch": 1.722051394684823, + "grad_norm": 2.0462844371795654, + "learning_rate": 1e-06, + "loss": 1.0001, + "mean_token_accuracy": 0.7003433108329773, + "num_tokens": 398617601.0, + "step": 15681 + }, + { + "epoch": 1.7221612123874368, + "grad_norm": 2.481468915939331, + "learning_rate": 1e-06, + "loss": 0.7777, + "mean_token_accuracy": 0.751718282699585, + "num_tokens": 398637384.0, + "step": 15682 + }, + { + "epoch": 1.7222710300900506, + "grad_norm": 2.0668604373931885, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.7136825323104858, + "num_tokens": 398666056.0, + "step": 15683 + }, + { + "epoch": 1.7223808477926643, + "grad_norm": 2.1740620136260986, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.714271068572998, + "num_tokens": 398692911.0, + "step": 15684 + }, + { + "epoch": 1.7224906654952779, + "grad_norm": 2.386808395385742, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7139619588851929, + "num_tokens": 398715803.0, + "step": 15685 + }, + { + "epoch": 1.7226004831978914, + "grad_norm": 2.4490010738372803, + "learning_rate": 1e-06, + "loss": 1.0028, + "mean_token_accuracy": 0.6949982047080994, + "num_tokens": 398738866.0, + "step": 15686 + }, + { + "epoch": 1.7227103009005051, + "grad_norm": 2.58274245262146, + "learning_rate": 1e-06, + "loss": 0.8928, + "mean_token_accuracy": 0.7321412563323975, + "num_tokens": 398759448.0, + "step": 15687 + }, + { + "epoch": 1.722820118603119, + "grad_norm": 2.1672885417938232, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7144280672073364, + "num_tokens": 398785222.0, + "step": 15688 + }, + { + "epoch": 1.7229299363057324, + "grad_norm": 2.1076266765594482, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7099573016166687, + "num_tokens": 398814133.0, + "step": 15689 + }, + { + "epoch": 1.7230397540083462, + "grad_norm": 2.2134711742401123, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7183692455291748, + "num_tokens": 398841149.0, + "step": 15690 + }, + { + "epoch": 1.7231495717109597, + "grad_norm": 1.9490833282470703, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7189233303070068, + "num_tokens": 398871030.0, + "step": 15691 + }, + { + "epoch": 1.7232593894135735, + "grad_norm": 2.4321396350860596, + "learning_rate": 1e-06, + "loss": 0.8319, + "mean_token_accuracy": 0.7380921840667725, + "num_tokens": 398894844.0, + "step": 15692 + }, + { + "epoch": 1.7233692071161872, + "grad_norm": 2.333756446838379, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7330851554870605, + "num_tokens": 398919016.0, + "step": 15693 + }, + { + "epoch": 1.7234790248188008, + "grad_norm": 2.333770751953125, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7071192264556885, + "num_tokens": 398944158.0, + "step": 15694 + }, + { + "epoch": 1.7235888425214143, + "grad_norm": 1.9266393184661865, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6940722465515137, + "num_tokens": 398977535.0, + "step": 15695 + }, + { + "epoch": 1.723698660224028, + "grad_norm": 2.184903621673584, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7209292650222778, + "num_tokens": 399003508.0, + "step": 15696 + }, + { + "epoch": 1.7238084779266418, + "grad_norm": 2.2847304344177246, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7110092639923096, + "num_tokens": 399028674.0, + "step": 15697 + }, + { + "epoch": 1.7239182956292556, + "grad_norm": 2.30429744720459, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7235428094863892, + "num_tokens": 399053841.0, + "step": 15698 + }, + { + "epoch": 1.7240281133318691, + "grad_norm": 2.1141514778137207, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7307446002960205, + "num_tokens": 399080615.0, + "step": 15699 + }, + { + "epoch": 1.7241379310344827, + "grad_norm": 2.1735689640045166, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7177398800849915, + "num_tokens": 399107336.0, + "step": 15700 + }, + { + "epoch": 1.7242477487370964, + "grad_norm": 2.071974039077759, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7114150524139404, + "num_tokens": 399136391.0, + "step": 15701 + }, + { + "epoch": 1.7243575664397102, + "grad_norm": 2.0252652168273926, + "learning_rate": 1e-06, + "loss": 0.9625, + "mean_token_accuracy": 0.7022782564163208, + "num_tokens": 399166802.0, + "step": 15702 + }, + { + "epoch": 1.7244673841423237, + "grad_norm": 2.599104166030884, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.7427870035171509, + "num_tokens": 399185806.0, + "step": 15703 + }, + { + "epoch": 1.7245772018449375, + "grad_norm": 2.507594585418701, + "learning_rate": 1e-06, + "loss": 0.8164, + "mean_token_accuracy": 0.7391513586044312, + "num_tokens": 399206537.0, + "step": 15704 + }, + { + "epoch": 1.724687019547551, + "grad_norm": 2.5285329818725586, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.73862624168396, + "num_tokens": 399226660.0, + "step": 15705 + }, + { + "epoch": 1.7247968372501647, + "grad_norm": 2.322293519973755, + "learning_rate": 1e-06, + "loss": 1.0088, + "mean_token_accuracy": 0.7018512487411499, + "num_tokens": 399252564.0, + "step": 15706 + }, + { + "epoch": 1.7249066549527785, + "grad_norm": 2.711428642272949, + "learning_rate": 1e-06, + "loss": 0.8742, + "mean_token_accuracy": 0.7225580215454102, + "num_tokens": 399272072.0, + "step": 15707 + }, + { + "epoch": 1.725016472655392, + "grad_norm": 2.2698898315429688, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.7449760437011719, + "num_tokens": 399296725.0, + "step": 15708 + }, + { + "epoch": 1.7251262903580056, + "grad_norm": 2.5958170890808105, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.735385537147522, + "num_tokens": 399319678.0, + "step": 15709 + }, + { + "epoch": 1.7252361080606193, + "grad_norm": 2.401291847229004, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7110238671302795, + "num_tokens": 399341879.0, + "step": 15710 + }, + { + "epoch": 1.725345925763233, + "grad_norm": 2.2242062091827393, + "learning_rate": 1e-06, + "loss": 0.89, + "mean_token_accuracy": 0.7247412204742432, + "num_tokens": 399367428.0, + "step": 15711 + }, + { + "epoch": 1.7254557434658468, + "grad_norm": 2.5751795768737793, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7345574498176575, + "num_tokens": 399386848.0, + "step": 15712 + }, + { + "epoch": 1.7255655611684604, + "grad_norm": 2.659482717514038, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7248483896255493, + "num_tokens": 399406525.0, + "step": 15713 + }, + { + "epoch": 1.725675378871074, + "grad_norm": 2.180163621902466, + "learning_rate": 1e-06, + "loss": 0.9317, + "mean_token_accuracy": 0.7102378606796265, + "num_tokens": 399432969.0, + "step": 15714 + }, + { + "epoch": 1.7257851965736877, + "grad_norm": 2.4804599285125732, + "learning_rate": 1e-06, + "loss": 0.8656, + "mean_token_accuracy": 0.7256495356559753, + "num_tokens": 399453982.0, + "step": 15715 + }, + { + "epoch": 1.7258950142763014, + "grad_norm": 2.311523199081421, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7148998379707336, + "num_tokens": 399478242.0, + "step": 15716 + }, + { + "epoch": 1.726004831978915, + "grad_norm": 2.593981981277466, + "learning_rate": 1e-06, + "loss": 0.7783, + "mean_token_accuracy": 0.7540363669395447, + "num_tokens": 399497585.0, + "step": 15717 + }, + { + "epoch": 1.7261146496815285, + "grad_norm": 2.3412461280822754, + "learning_rate": 1e-06, + "loss": 0.7874, + "mean_token_accuracy": 0.7485926151275635, + "num_tokens": 399519304.0, + "step": 15718 + }, + { + "epoch": 1.7262244673841423, + "grad_norm": 2.24526047706604, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7161464691162109, + "num_tokens": 399542813.0, + "step": 15719 + }, + { + "epoch": 1.726334285086756, + "grad_norm": 2.2041118144989014, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7100750803947449, + "num_tokens": 399569461.0, + "step": 15720 + }, + { + "epoch": 1.7264441027893698, + "grad_norm": 2.0775580406188965, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.716217041015625, + "num_tokens": 399598720.0, + "step": 15721 + }, + { + "epoch": 1.7265539204919833, + "grad_norm": 2.301518440246582, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7139727473258972, + "num_tokens": 399623108.0, + "step": 15722 + }, + { + "epoch": 1.7266637381945968, + "grad_norm": 2.21393084526062, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7123636603355408, + "num_tokens": 399651931.0, + "step": 15723 + }, + { + "epoch": 1.7267735558972106, + "grad_norm": 2.368077039718628, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.739649772644043, + "num_tokens": 399674236.0, + "step": 15724 + }, + { + "epoch": 1.7268833735998244, + "grad_norm": 2.390768527984619, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7128227949142456, + "num_tokens": 399698512.0, + "step": 15725 + }, + { + "epoch": 1.7269931913024381, + "grad_norm": 2.0880186557769775, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7301788330078125, + "num_tokens": 399725675.0, + "step": 15726 + }, + { + "epoch": 1.7271030090050516, + "grad_norm": 2.391535758972168, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7362374067306519, + "num_tokens": 399748255.0, + "step": 15727 + }, + { + "epoch": 1.7272128267076652, + "grad_norm": 2.13797664642334, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.720181941986084, + "num_tokens": 399776734.0, + "step": 15728 + }, + { + "epoch": 1.727322644410279, + "grad_norm": 2.2943577766418457, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.721064567565918, + "num_tokens": 399800887.0, + "step": 15729 + }, + { + "epoch": 1.7274324621128927, + "grad_norm": 2.407858371734619, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7192139029502869, + "num_tokens": 399824689.0, + "step": 15730 + }, + { + "epoch": 1.7275422798155062, + "grad_norm": 2.5876917839050293, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7148228883743286, + "num_tokens": 399845094.0, + "step": 15731 + }, + { + "epoch": 1.7276520975181198, + "grad_norm": 2.2222843170166016, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7010120153427124, + "num_tokens": 399873144.0, + "step": 15732 + }, + { + "epoch": 1.7277619152207335, + "grad_norm": 2.0602316856384277, + "learning_rate": 1e-06, + "loss": 1.0089, + "mean_token_accuracy": 0.6950583457946777, + "num_tokens": 399902839.0, + "step": 15733 + }, + { + "epoch": 1.7278717329233473, + "grad_norm": 1.9698625802993774, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7318777441978455, + "num_tokens": 399933540.0, + "step": 15734 + }, + { + "epoch": 1.727981550625961, + "grad_norm": 2.29571533203125, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7225954532623291, + "num_tokens": 399957401.0, + "step": 15735 + }, + { + "epoch": 1.7280913683285746, + "grad_norm": 2.1115663051605225, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7208331823348999, + "num_tokens": 399985429.0, + "step": 15736 + }, + { + "epoch": 1.728201186031188, + "grad_norm": 2.1116209030151367, + "learning_rate": 1e-06, + "loss": 1.0358, + "mean_token_accuracy": 0.6870115995407104, + "num_tokens": 400016433.0, + "step": 15737 + }, + { + "epoch": 1.7283110037338019, + "grad_norm": 2.1271283626556396, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7003599405288696, + "num_tokens": 400046881.0, + "step": 15738 + }, + { + "epoch": 1.7284208214364156, + "grad_norm": 2.1971840858459473, + "learning_rate": 1e-06, + "loss": 0.9792, + "mean_token_accuracy": 0.6949199438095093, + "num_tokens": 400073563.0, + "step": 15739 + }, + { + "epoch": 1.7285306391390292, + "grad_norm": 2.272064685821533, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7094481587409973, + "num_tokens": 400098688.0, + "step": 15740 + }, + { + "epoch": 1.728640456841643, + "grad_norm": 2.1286203861236572, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7162449359893799, + "num_tokens": 400126956.0, + "step": 15741 + }, + { + "epoch": 1.7287502745442564, + "grad_norm": 2.3753273487091064, + "learning_rate": 1e-06, + "loss": 1.0025, + "mean_token_accuracy": 0.6955106258392334, + "num_tokens": 400150612.0, + "step": 15742 + }, + { + "epoch": 1.7288600922468702, + "grad_norm": 1.9133319854736328, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.6987687349319458, + "num_tokens": 400183675.0, + "step": 15743 + }, + { + "epoch": 1.728969909949484, + "grad_norm": 1.981424331665039, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6951481103897095, + "num_tokens": 400217420.0, + "step": 15744 + }, + { + "epoch": 1.7290797276520975, + "grad_norm": 2.090574264526367, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.715329110622406, + "num_tokens": 400247747.0, + "step": 15745 + }, + { + "epoch": 1.729189545354711, + "grad_norm": 2.2126669883728027, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7206981778144836, + "num_tokens": 400273551.0, + "step": 15746 + }, + { + "epoch": 1.7292993630573248, + "grad_norm": 2.2137763500213623, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7324286699295044, + "num_tokens": 400298193.0, + "step": 15747 + }, + { + "epoch": 1.7294091807599385, + "grad_norm": 2.0689823627471924, + "learning_rate": 1e-06, + "loss": 0.9605, + "mean_token_accuracy": 0.7028483152389526, + "num_tokens": 400325919.0, + "step": 15748 + }, + { + "epoch": 1.7295189984625523, + "grad_norm": 1.9405560493469238, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7053584456443787, + "num_tokens": 400357601.0, + "step": 15749 + }, + { + "epoch": 1.7296288161651658, + "grad_norm": 2.1446797847747803, + "learning_rate": 1e-06, + "loss": 0.8932, + "mean_token_accuracy": 0.7183932065963745, + "num_tokens": 400385127.0, + "step": 15750 + }, + { + "epoch": 1.7297386338677794, + "grad_norm": 2.111833333969116, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7109173536300659, + "num_tokens": 400414106.0, + "step": 15751 + }, + { + "epoch": 1.7298484515703931, + "grad_norm": 2.008784055709839, + "learning_rate": 1e-06, + "loss": 1.0008, + "mean_token_accuracy": 0.6922304034233093, + "num_tokens": 400446957.0, + "step": 15752 + }, + { + "epoch": 1.7299582692730069, + "grad_norm": 2.0749123096466064, + "learning_rate": 1e-06, + "loss": 0.9799, + "mean_token_accuracy": 0.7001854181289673, + "num_tokens": 400476887.0, + "step": 15753 + }, + { + "epoch": 1.7300680869756204, + "grad_norm": 2.0930004119873047, + "learning_rate": 1e-06, + "loss": 0.8436, + "mean_token_accuracy": 0.7327530384063721, + "num_tokens": 400505472.0, + "step": 15754 + }, + { + "epoch": 1.7301779046782342, + "grad_norm": 1.891626000404358, + "learning_rate": 1e-06, + "loss": 1.0152, + "mean_token_accuracy": 0.6916815042495728, + "num_tokens": 400538719.0, + "step": 15755 + }, + { + "epoch": 1.7302877223808477, + "grad_norm": 2.1712310314178467, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7291920781135559, + "num_tokens": 400565064.0, + "step": 15756 + }, + { + "epoch": 1.7303975400834615, + "grad_norm": 2.2614243030548096, + "learning_rate": 1e-06, + "loss": 0.9117, + "mean_token_accuracy": 0.7245593070983887, + "num_tokens": 400589889.0, + "step": 15757 + }, + { + "epoch": 1.7305073577860752, + "grad_norm": 2.237823247909546, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7198036909103394, + "num_tokens": 400616828.0, + "step": 15758 + }, + { + "epoch": 1.7306171754886888, + "grad_norm": 2.022510290145874, + "learning_rate": 1e-06, + "loss": 0.9907, + "mean_token_accuracy": 0.6887701749801636, + "num_tokens": 400649388.0, + "step": 15759 + }, + { + "epoch": 1.7307269931913023, + "grad_norm": 2.3586108684539795, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.726693868637085, + "num_tokens": 400672393.0, + "step": 15760 + }, + { + "epoch": 1.730836810893916, + "grad_norm": 2.183765172958374, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7086437940597534, + "num_tokens": 400701980.0, + "step": 15761 + }, + { + "epoch": 1.7309466285965298, + "grad_norm": 2.2129573822021484, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7236438989639282, + "num_tokens": 400730022.0, + "step": 15762 + }, + { + "epoch": 1.7310564462991436, + "grad_norm": 2.837524652481079, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7399327754974365, + "num_tokens": 400746575.0, + "step": 15763 + }, + { + "epoch": 1.731166264001757, + "grad_norm": 2.100342273712158, + "learning_rate": 1e-06, + "loss": 0.9272, + "mean_token_accuracy": 0.7107486724853516, + "num_tokens": 400776243.0, + "step": 15764 + }, + { + "epoch": 1.7312760817043706, + "grad_norm": 2.0565290451049805, + "learning_rate": 1e-06, + "loss": 0.9651, + "mean_token_accuracy": 0.7004034519195557, + "num_tokens": 400806745.0, + "step": 15765 + }, + { + "epoch": 1.7313858994069844, + "grad_norm": 2.384413957595825, + "learning_rate": 1e-06, + "loss": 0.8368, + "mean_token_accuracy": 0.7412956953048706, + "num_tokens": 400830560.0, + "step": 15766 + }, + { + "epoch": 1.7314957171095982, + "grad_norm": 2.090749740600586, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7377899289131165, + "num_tokens": 400856687.0, + "step": 15767 + }, + { + "epoch": 1.7316055348122117, + "grad_norm": 2.0826416015625, + "learning_rate": 1e-06, + "loss": 0.9212, + "mean_token_accuracy": 0.7156966328620911, + "num_tokens": 400890342.0, + "step": 15768 + }, + { + "epoch": 1.7317153525148252, + "grad_norm": 1.9541304111480713, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7032208442687988, + "num_tokens": 400924496.0, + "step": 15769 + }, + { + "epoch": 1.731825170217439, + "grad_norm": 2.4584193229675293, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7081484794616699, + "num_tokens": 400947528.0, + "step": 15770 + }, + { + "epoch": 1.7319349879200527, + "grad_norm": 2.298114061355591, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.734032928943634, + "num_tokens": 400971421.0, + "step": 15771 + }, + { + "epoch": 1.7320448056226665, + "grad_norm": 2.2862467765808105, + "learning_rate": 1e-06, + "loss": 0.7801, + "mean_token_accuracy": 0.7531259655952454, + "num_tokens": 400995888.0, + "step": 15772 + }, + { + "epoch": 1.73215462332528, + "grad_norm": 2.049030303955078, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7120650410652161, + "num_tokens": 401025086.0, + "step": 15773 + }, + { + "epoch": 1.7322644410278936, + "grad_norm": 2.020824670791626, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.7217569351196289, + "num_tokens": 401057124.0, + "step": 15774 + }, + { + "epoch": 1.7323742587305073, + "grad_norm": 2.33449125289917, + "learning_rate": 1e-06, + "loss": 0.9462, + "mean_token_accuracy": 0.7043070197105408, + "num_tokens": 401081591.0, + "step": 15775 + }, + { + "epoch": 1.732484076433121, + "grad_norm": 2.2262401580810547, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7434070706367493, + "num_tokens": 401106895.0, + "step": 15776 + }, + { + "epoch": 1.7325938941357348, + "grad_norm": 2.015111207962036, + "learning_rate": 1e-06, + "loss": 0.8982, + "mean_token_accuracy": 0.731459379196167, + "num_tokens": 401136018.0, + "step": 15777 + }, + { + "epoch": 1.7327037118383484, + "grad_norm": 2.387855052947998, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7045233249664307, + "num_tokens": 401159903.0, + "step": 15778 + }, + { + "epoch": 1.732813529540962, + "grad_norm": 2.590273857116699, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7244663238525391, + "num_tokens": 401180274.0, + "step": 15779 + }, + { + "epoch": 1.7329233472435757, + "grad_norm": 2.4286277294158936, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7129486203193665, + "num_tokens": 401203107.0, + "step": 15780 + }, + { + "epoch": 1.7330331649461894, + "grad_norm": 2.143221616744995, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.719427227973938, + "num_tokens": 401229752.0, + "step": 15781 + }, + { + "epoch": 1.733142982648803, + "grad_norm": 2.2092983722686768, + "learning_rate": 1e-06, + "loss": 0.8775, + "mean_token_accuracy": 0.7239938974380493, + "num_tokens": 401255235.0, + "step": 15782 + }, + { + "epoch": 1.7332528003514165, + "grad_norm": 2.3711931705474854, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7389529943466187, + "num_tokens": 401278150.0, + "step": 15783 + }, + { + "epoch": 1.7333626180540302, + "grad_norm": 2.6039721965789795, + "learning_rate": 1e-06, + "loss": 0.8014, + "mean_token_accuracy": 0.7441979646682739, + "num_tokens": 401296917.0, + "step": 15784 + }, + { + "epoch": 1.733472435756644, + "grad_norm": 2.149650812149048, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7318813800811768, + "num_tokens": 401322665.0, + "step": 15785 + }, + { + "epoch": 1.7335822534592578, + "grad_norm": 2.2409496307373047, + "learning_rate": 1e-06, + "loss": 0.7466, + "mean_token_accuracy": 0.7587476968765259, + "num_tokens": 401345765.0, + "step": 15786 + }, + { + "epoch": 1.7336920711618713, + "grad_norm": 2.3126139640808105, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7112654447555542, + "num_tokens": 401369592.0, + "step": 15787 + }, + { + "epoch": 1.7338018888644848, + "grad_norm": 2.056915521621704, + "learning_rate": 1e-06, + "loss": 1.0123, + "mean_token_accuracy": 0.689521074295044, + "num_tokens": 401400269.0, + "step": 15788 + }, + { + "epoch": 1.7339117065670986, + "grad_norm": 2.4218547344207764, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7311748266220093, + "num_tokens": 401422096.0, + "step": 15789 + }, + { + "epoch": 1.7340215242697123, + "grad_norm": 2.0021088123321533, + "learning_rate": 1e-06, + "loss": 0.9724, + "mean_token_accuracy": 0.6963415145874023, + "num_tokens": 401454921.0, + "step": 15790 + }, + { + "epoch": 1.734131341972326, + "grad_norm": 2.373710870742798, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7179262638092041, + "num_tokens": 401477235.0, + "step": 15791 + }, + { + "epoch": 1.7342411596749396, + "grad_norm": 2.1734821796417236, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.709658682346344, + "num_tokens": 401503497.0, + "step": 15792 + }, + { + "epoch": 1.7343509773775532, + "grad_norm": 2.1567955017089844, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7242740988731384, + "num_tokens": 401530992.0, + "step": 15793 + }, + { + "epoch": 1.734460795080167, + "grad_norm": 2.4907004833221436, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.719785749912262, + "num_tokens": 401553778.0, + "step": 15794 + }, + { + "epoch": 1.7345706127827807, + "grad_norm": 1.9040324687957764, + "learning_rate": 1e-06, + "loss": 0.9877, + "mean_token_accuracy": 0.7005707025527954, + "num_tokens": 401588803.0, + "step": 15795 + }, + { + "epoch": 1.7346804304853942, + "grad_norm": 2.2970104217529297, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7207927107810974, + "num_tokens": 401614712.0, + "step": 15796 + }, + { + "epoch": 1.7347902481880078, + "grad_norm": 2.545341730117798, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7172096967697144, + "num_tokens": 401635505.0, + "step": 15797 + }, + { + "epoch": 1.7349000658906215, + "grad_norm": 2.149794101715088, + "learning_rate": 1e-06, + "loss": 0.8697, + "mean_token_accuracy": 0.7320747375488281, + "num_tokens": 401662473.0, + "step": 15798 + }, + { + "epoch": 1.7350098835932353, + "grad_norm": 2.7393925189971924, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7130728363990784, + "num_tokens": 401681293.0, + "step": 15799 + }, + { + "epoch": 1.735119701295849, + "grad_norm": 2.453796863555908, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.7461181879043579, + "num_tokens": 401701593.0, + "step": 15800 + }, + { + "epoch": 1.7352295189984626, + "grad_norm": 2.0039772987365723, + "learning_rate": 1e-06, + "loss": 0.8612, + "mean_token_accuracy": 0.7311865091323853, + "num_tokens": 401730838.0, + "step": 15801 + }, + { + "epoch": 1.735339336701076, + "grad_norm": 2.068481922149658, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7252804040908813, + "num_tokens": 401759923.0, + "step": 15802 + }, + { + "epoch": 1.7354491544036899, + "grad_norm": 2.2690999507904053, + "learning_rate": 1e-06, + "loss": 0.8414, + "mean_token_accuracy": 0.7353852391242981, + "num_tokens": 401786567.0, + "step": 15803 + }, + { + "epoch": 1.7355589721063036, + "grad_norm": 2.152078628540039, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7254672050476074, + "num_tokens": 401813698.0, + "step": 15804 + }, + { + "epoch": 1.7356687898089171, + "grad_norm": 2.1728527545928955, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7046253681182861, + "num_tokens": 401841262.0, + "step": 15805 + }, + { + "epoch": 1.735778607511531, + "grad_norm": 2.3813765048980713, + "learning_rate": 1e-06, + "loss": 0.9217, + "mean_token_accuracy": 0.7161029577255249, + "num_tokens": 401864706.0, + "step": 15806 + }, + { + "epoch": 1.7358884252141444, + "grad_norm": 2.837352991104126, + "learning_rate": 1e-06, + "loss": 0.8005, + "mean_token_accuracy": 0.7407350540161133, + "num_tokens": 401881090.0, + "step": 15807 + }, + { + "epoch": 1.7359982429167582, + "grad_norm": 2.4366841316223145, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7307430505752563, + "num_tokens": 401903451.0, + "step": 15808 + }, + { + "epoch": 1.736108060619372, + "grad_norm": 2.2453036308288574, + "learning_rate": 1e-06, + "loss": 0.9172, + "mean_token_accuracy": 0.7127286791801453, + "num_tokens": 401928598.0, + "step": 15809 + }, + { + "epoch": 1.7362178783219855, + "grad_norm": 2.0557165145874023, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7204490900039673, + "num_tokens": 401956478.0, + "step": 15810 + }, + { + "epoch": 1.736327696024599, + "grad_norm": 2.0216612815856934, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7225235104560852, + "num_tokens": 401987988.0, + "step": 15811 + }, + { + "epoch": 1.7364375137272128, + "grad_norm": 2.098012924194336, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7190830111503601, + "num_tokens": 402016665.0, + "step": 15812 + }, + { + "epoch": 1.7365473314298265, + "grad_norm": 2.1555168628692627, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7115204334259033, + "num_tokens": 402045720.0, + "step": 15813 + }, + { + "epoch": 1.7366571491324403, + "grad_norm": 2.0559849739074707, + "learning_rate": 1e-06, + "loss": 0.9666, + "mean_token_accuracy": 0.697600245475769, + "num_tokens": 402075641.0, + "step": 15814 + }, + { + "epoch": 1.7367669668350538, + "grad_norm": 2.060892343521118, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7262630462646484, + "num_tokens": 402104230.0, + "step": 15815 + }, + { + "epoch": 1.7368767845376674, + "grad_norm": 2.3843600749969482, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7182184457778931, + "num_tokens": 402128272.0, + "step": 15816 + }, + { + "epoch": 1.7369866022402811, + "grad_norm": 2.2978813648223877, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7230228185653687, + "num_tokens": 402152984.0, + "step": 15817 + }, + { + "epoch": 1.7370964199428949, + "grad_norm": 2.34401273727417, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7070229053497314, + "num_tokens": 402175909.0, + "step": 15818 + }, + { + "epoch": 1.7372062376455084, + "grad_norm": 2.5740532875061035, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7302910089492798, + "num_tokens": 402195595.0, + "step": 15819 + }, + { + "epoch": 1.7373160553481222, + "grad_norm": 2.056195020675659, + "learning_rate": 1e-06, + "loss": 1.0052, + "mean_token_accuracy": 0.6951611042022705, + "num_tokens": 402228564.0, + "step": 15820 + }, + { + "epoch": 1.7374258730507357, + "grad_norm": 2.5015761852264404, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7156039476394653, + "num_tokens": 402249671.0, + "step": 15821 + }, + { + "epoch": 1.7375356907533495, + "grad_norm": 2.238079309463501, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.722184419631958, + "num_tokens": 402279030.0, + "step": 15822 + }, + { + "epoch": 1.7376455084559632, + "grad_norm": 2.241563320159912, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7434249520301819, + "num_tokens": 402304306.0, + "step": 15823 + }, + { + "epoch": 1.7377553261585768, + "grad_norm": 2.3464150428771973, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.708074688911438, + "num_tokens": 402328806.0, + "step": 15824 + }, + { + "epoch": 1.7378651438611903, + "grad_norm": 2.428382396697998, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7037142515182495, + "num_tokens": 402354256.0, + "step": 15825 + }, + { + "epoch": 1.737974961563804, + "grad_norm": 2.105011224746704, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.7054028511047363, + "num_tokens": 402384334.0, + "step": 15826 + }, + { + "epoch": 1.7380847792664178, + "grad_norm": 2.1629374027252197, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7214879393577576, + "num_tokens": 402410421.0, + "step": 15827 + }, + { + "epoch": 1.7381945969690316, + "grad_norm": 2.4283387660980225, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7233134508132935, + "num_tokens": 402434056.0, + "step": 15828 + }, + { + "epoch": 1.738304414671645, + "grad_norm": 2.0566158294677734, + "learning_rate": 1e-06, + "loss": 0.8577, + "mean_token_accuracy": 0.729202151298523, + "num_tokens": 402461254.0, + "step": 15829 + }, + { + "epoch": 1.7384142323742586, + "grad_norm": 1.992843508720398, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7203835248947144, + "num_tokens": 402491272.0, + "step": 15830 + }, + { + "epoch": 1.7385240500768724, + "grad_norm": 2.2574996948242188, + "learning_rate": 1e-06, + "loss": 0.8756, + "mean_token_accuracy": 0.7245789766311646, + "num_tokens": 402517647.0, + "step": 15831 + }, + { + "epoch": 1.7386338677794861, + "grad_norm": 2.283687114715576, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7136294841766357, + "num_tokens": 402543359.0, + "step": 15832 + }, + { + "epoch": 1.7387436854820997, + "grad_norm": 2.4072744846343994, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7192562818527222, + "num_tokens": 402566410.0, + "step": 15833 + }, + { + "epoch": 1.7388535031847132, + "grad_norm": 2.516890525817871, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7370816469192505, + "num_tokens": 402587727.0, + "step": 15834 + }, + { + "epoch": 1.738963320887327, + "grad_norm": 2.3135907649993896, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7097858190536499, + "num_tokens": 402612879.0, + "step": 15835 + }, + { + "epoch": 1.7390731385899407, + "grad_norm": 2.147961139678955, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7111741304397583, + "num_tokens": 402641619.0, + "step": 15836 + }, + { + "epoch": 1.7391829562925545, + "grad_norm": 2.1742777824401855, + "learning_rate": 1e-06, + "loss": 0.9225, + "mean_token_accuracy": 0.7097210884094238, + "num_tokens": 402669653.0, + "step": 15837 + }, + { + "epoch": 1.739292773995168, + "grad_norm": 2.3645966053009033, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7252398729324341, + "num_tokens": 402694778.0, + "step": 15838 + }, + { + "epoch": 1.7394025916977816, + "grad_norm": 2.448058843612671, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7220368385314941, + "num_tokens": 402717462.0, + "step": 15839 + }, + { + "epoch": 1.7395124094003953, + "grad_norm": 2.1297357082366943, + "learning_rate": 1e-06, + "loss": 0.9984, + "mean_token_accuracy": 0.6931625008583069, + "num_tokens": 402745283.0, + "step": 15840 + }, + { + "epoch": 1.739622227103009, + "grad_norm": 2.4758076667785645, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7316513061523438, + "num_tokens": 402767876.0, + "step": 15841 + }, + { + "epoch": 1.7397320448056228, + "grad_norm": 2.1842095851898193, + "learning_rate": 1e-06, + "loss": 0.8774, + "mean_token_accuracy": 0.7256094217300415, + "num_tokens": 402793639.0, + "step": 15842 + }, + { + "epoch": 1.7398418625082364, + "grad_norm": 2.130018949508667, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7143579125404358, + "num_tokens": 402822227.0, + "step": 15843 + }, + { + "epoch": 1.73995168021085, + "grad_norm": 2.3167476654052734, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7160053253173828, + "num_tokens": 402849376.0, + "step": 15844 + }, + { + "epoch": 1.7400614979134637, + "grad_norm": 1.9267315864562988, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7094706892967224, + "num_tokens": 402880853.0, + "step": 15845 + }, + { + "epoch": 1.7401713156160774, + "grad_norm": 2.3947317600250244, + "learning_rate": 1e-06, + "loss": 0.8443, + "mean_token_accuracy": 0.7424190044403076, + "num_tokens": 402903044.0, + "step": 15846 + }, + { + "epoch": 1.740281133318691, + "grad_norm": 2.2785303592681885, + "learning_rate": 1e-06, + "loss": 1.0016, + "mean_token_accuracy": 0.691125750541687, + "num_tokens": 402927791.0, + "step": 15847 + }, + { + "epoch": 1.7403909510213045, + "grad_norm": 2.299542188644409, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7190774083137512, + "num_tokens": 402953177.0, + "step": 15848 + }, + { + "epoch": 1.7405007687239182, + "grad_norm": 1.996114730834961, + "learning_rate": 1e-06, + "loss": 0.7483, + "mean_token_accuracy": 0.7614875435829163, + "num_tokens": 402980861.0, + "step": 15849 + }, + { + "epoch": 1.740610586426532, + "grad_norm": 2.202862501144409, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7250714302062988, + "num_tokens": 403005319.0, + "step": 15850 + }, + { + "epoch": 1.7407204041291457, + "grad_norm": 2.074704170227051, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7188000679016113, + "num_tokens": 403035700.0, + "step": 15851 + }, + { + "epoch": 1.7408302218317593, + "grad_norm": 2.1147007942199707, + "learning_rate": 1e-06, + "loss": 0.8989, + "mean_token_accuracy": 0.7120442986488342, + "num_tokens": 403061823.0, + "step": 15852 + }, + { + "epoch": 1.7409400395343728, + "grad_norm": 2.6759872436523438, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7193253636360168, + "num_tokens": 403082390.0, + "step": 15853 + }, + { + "epoch": 1.7410498572369866, + "grad_norm": 2.0452280044555664, + "learning_rate": 1e-06, + "loss": 0.9947, + "mean_token_accuracy": 0.699549674987793, + "num_tokens": 403110880.0, + "step": 15854 + }, + { + "epoch": 1.7411596749396003, + "grad_norm": 2.342877149581909, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7456080913543701, + "num_tokens": 403133130.0, + "step": 15855 + }, + { + "epoch": 1.741269492642214, + "grad_norm": 2.4313809871673584, + "learning_rate": 1e-06, + "loss": 0.8503, + "mean_token_accuracy": 0.738277792930603, + "num_tokens": 403154872.0, + "step": 15856 + }, + { + "epoch": 1.7413793103448276, + "grad_norm": 2.1837964057922363, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7157804369926453, + "num_tokens": 403181230.0, + "step": 15857 + }, + { + "epoch": 1.7414891280474412, + "grad_norm": 2.1279525756835938, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7033929824829102, + "num_tokens": 403210508.0, + "step": 15858 + }, + { + "epoch": 1.741598945750055, + "grad_norm": 2.470761299133301, + "learning_rate": 1e-06, + "loss": 0.8712, + "mean_token_accuracy": 0.7264386415481567, + "num_tokens": 403232337.0, + "step": 15859 + }, + { + "epoch": 1.7417087634526687, + "grad_norm": 2.141834020614624, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7348002195358276, + "num_tokens": 403258566.0, + "step": 15860 + }, + { + "epoch": 1.7418185811552822, + "grad_norm": 2.3515660762786865, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7214823365211487, + "num_tokens": 403281534.0, + "step": 15861 + }, + { + "epoch": 1.7419283988578957, + "grad_norm": 2.9673149585723877, + "learning_rate": 1e-06, + "loss": 0.8266, + "mean_token_accuracy": 0.736505925655365, + "num_tokens": 403297796.0, + "step": 15862 + }, + { + "epoch": 1.7420382165605095, + "grad_norm": 2.1236493587493896, + "learning_rate": 1e-06, + "loss": 0.8833, + "mean_token_accuracy": 0.72135329246521, + "num_tokens": 403324851.0, + "step": 15863 + }, + { + "epoch": 1.7421480342631233, + "grad_norm": 1.9897948503494263, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7159146666526794, + "num_tokens": 403356801.0, + "step": 15864 + }, + { + "epoch": 1.742257851965737, + "grad_norm": 2.324054718017578, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7173435688018799, + "num_tokens": 403380099.0, + "step": 15865 + }, + { + "epoch": 1.7423676696683505, + "grad_norm": 2.237013578414917, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7177990674972534, + "num_tokens": 403405436.0, + "step": 15866 + }, + { + "epoch": 1.742477487370964, + "grad_norm": 1.938973307609558, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.7047369480133057, + "num_tokens": 403439536.0, + "step": 15867 + }, + { + "epoch": 1.7425873050735778, + "grad_norm": 2.221012592315674, + "learning_rate": 1e-06, + "loss": 0.8706, + "mean_token_accuracy": 0.7272021770477295, + "num_tokens": 403463694.0, + "step": 15868 + }, + { + "epoch": 1.7426971227761916, + "grad_norm": 2.1289591789245605, + "learning_rate": 1e-06, + "loss": 1.0385, + "mean_token_accuracy": 0.6758264899253845, + "num_tokens": 403494372.0, + "step": 15869 + }, + { + "epoch": 1.7428069404788051, + "grad_norm": 2.5255911350250244, + "learning_rate": 1e-06, + "loss": 0.8499, + "mean_token_accuracy": 0.7318832278251648, + "num_tokens": 403516114.0, + "step": 15870 + }, + { + "epoch": 1.742916758181419, + "grad_norm": 1.9935795068740845, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7331201434135437, + "num_tokens": 403545625.0, + "step": 15871 + }, + { + "epoch": 1.7430265758840324, + "grad_norm": 2.242295742034912, + "learning_rate": 1e-06, + "loss": 0.8868, + "mean_token_accuracy": 0.7237230539321899, + "num_tokens": 403570700.0, + "step": 15872 + }, + { + "epoch": 1.7431363935866462, + "grad_norm": 2.5230038166046143, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7102217078208923, + "num_tokens": 403592832.0, + "step": 15873 + }, + { + "epoch": 1.74324621128926, + "grad_norm": 2.4158453941345215, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7340525984764099, + "num_tokens": 403615044.0, + "step": 15874 + }, + { + "epoch": 1.7433560289918735, + "grad_norm": 2.183657169342041, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7445807456970215, + "num_tokens": 403639276.0, + "step": 15875 + }, + { + "epoch": 1.743465846694487, + "grad_norm": 2.3748059272766113, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.729407787322998, + "num_tokens": 403660585.0, + "step": 15876 + }, + { + "epoch": 1.7435756643971008, + "grad_norm": 2.182957410812378, + "learning_rate": 1e-06, + "loss": 0.9585, + "mean_token_accuracy": 0.7058887481689453, + "num_tokens": 403689132.0, + "step": 15877 + }, + { + "epoch": 1.7436854820997145, + "grad_norm": 2.388819456100464, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7207989692687988, + "num_tokens": 403712359.0, + "step": 15878 + }, + { + "epoch": 1.7437952998023283, + "grad_norm": 2.435228109359741, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7169560194015503, + "num_tokens": 403736223.0, + "step": 15879 + }, + { + "epoch": 1.7439051175049418, + "grad_norm": 2.217928409576416, + "learning_rate": 1e-06, + "loss": 0.9303, + "mean_token_accuracy": 0.7135999798774719, + "num_tokens": 403762717.0, + "step": 15880 + }, + { + "epoch": 1.7440149352075554, + "grad_norm": 2.081421136856079, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7126897573471069, + "num_tokens": 403792800.0, + "step": 15881 + }, + { + "epoch": 1.744124752910169, + "grad_norm": 2.359954595565796, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7075755596160889, + "num_tokens": 403816989.0, + "step": 15882 + }, + { + "epoch": 1.7442345706127829, + "grad_norm": 2.376735210418701, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7081363201141357, + "num_tokens": 403840539.0, + "step": 15883 + }, + { + "epoch": 1.7443443883153964, + "grad_norm": 2.12137508392334, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7094862461090088, + "num_tokens": 403868080.0, + "step": 15884 + }, + { + "epoch": 1.7444542060180102, + "grad_norm": 2.6473255157470703, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.7439843416213989, + "num_tokens": 403886251.0, + "step": 15885 + }, + { + "epoch": 1.7445640237206237, + "grad_norm": 2.7179219722747803, + "learning_rate": 1e-06, + "loss": 0.8959, + "mean_token_accuracy": 0.7325579524040222, + "num_tokens": 403905064.0, + "step": 15886 + }, + { + "epoch": 1.7446738414232374, + "grad_norm": 2.3313863277435303, + "learning_rate": 1e-06, + "loss": 0.9504, + "mean_token_accuracy": 0.7067785263061523, + "num_tokens": 403929887.0, + "step": 15887 + }, + { + "epoch": 1.7447836591258512, + "grad_norm": 2.1682820320129395, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.734041690826416, + "num_tokens": 403957825.0, + "step": 15888 + }, + { + "epoch": 1.7448934768284647, + "grad_norm": 2.6058850288391113, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7309403419494629, + "num_tokens": 403977891.0, + "step": 15889 + }, + { + "epoch": 1.7450032945310783, + "grad_norm": 2.2601349353790283, + "learning_rate": 1e-06, + "loss": 0.8091, + "mean_token_accuracy": 0.7455014586448669, + "num_tokens": 404002119.0, + "step": 15890 + }, + { + "epoch": 1.745113112233692, + "grad_norm": 2.3186817169189453, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7149924039840698, + "num_tokens": 404028358.0, + "step": 15891 + }, + { + "epoch": 1.7452229299363058, + "grad_norm": 2.3888089656829834, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7098515033721924, + "num_tokens": 404052296.0, + "step": 15892 + }, + { + "epoch": 1.7453327476389195, + "grad_norm": 2.114384174346924, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.697475790977478, + "num_tokens": 404082761.0, + "step": 15893 + }, + { + "epoch": 1.745442565341533, + "grad_norm": 2.3078489303588867, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.709241509437561, + "num_tokens": 404108362.0, + "step": 15894 + }, + { + "epoch": 1.7455523830441466, + "grad_norm": 2.174436330795288, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7233431339263916, + "num_tokens": 404134849.0, + "step": 15895 + }, + { + "epoch": 1.7456622007467604, + "grad_norm": 2.157963275909424, + "learning_rate": 1e-06, + "loss": 0.943, + "mean_token_accuracy": 0.7062591910362244, + "num_tokens": 404163352.0, + "step": 15896 + }, + { + "epoch": 1.7457720184493741, + "grad_norm": 2.1174280643463135, + "learning_rate": 1e-06, + "loss": 0.7849, + "mean_token_accuracy": 0.7456592321395874, + "num_tokens": 404188887.0, + "step": 15897 + }, + { + "epoch": 1.7458818361519877, + "grad_norm": 2.2640035152435303, + "learning_rate": 1e-06, + "loss": 0.9534, + "mean_token_accuracy": 0.7014105319976807, + "num_tokens": 404212452.0, + "step": 15898 + }, + { + "epoch": 1.7459916538546012, + "grad_norm": 2.1290299892425537, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7134884595870972, + "num_tokens": 404242500.0, + "step": 15899 + }, + { + "epoch": 1.746101471557215, + "grad_norm": 2.1011016368865967, + "learning_rate": 1e-06, + "loss": 0.9579, + "mean_token_accuracy": 0.7023409605026245, + "num_tokens": 404275519.0, + "step": 15900 + }, + { + "epoch": 1.7462112892598287, + "grad_norm": 2.304064989089966, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7333992719650269, + "num_tokens": 404298262.0, + "step": 15901 + }, + { + "epoch": 1.7463211069624425, + "grad_norm": 2.1792619228363037, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7105284929275513, + "num_tokens": 404325009.0, + "step": 15902 + }, + { + "epoch": 1.746430924665056, + "grad_norm": 2.0650722980499268, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7090330123901367, + "num_tokens": 404354589.0, + "step": 15903 + }, + { + "epoch": 1.7465407423676695, + "grad_norm": 2.166440725326538, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.722968578338623, + "num_tokens": 404381937.0, + "step": 15904 + }, + { + "epoch": 1.7466505600702833, + "grad_norm": 2.675022840499878, + "learning_rate": 1e-06, + "loss": 0.7533, + "mean_token_accuracy": 0.7589594125747681, + "num_tokens": 404399147.0, + "step": 15905 + }, + { + "epoch": 1.746760377772897, + "grad_norm": 2.26052188873291, + "learning_rate": 1e-06, + "loss": 0.9359, + "mean_token_accuracy": 0.7083883881568909, + "num_tokens": 404426286.0, + "step": 15906 + }, + { + "epoch": 1.7468701954755108, + "grad_norm": 2.390882968902588, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7130317687988281, + "num_tokens": 404451175.0, + "step": 15907 + }, + { + "epoch": 1.7469800131781243, + "grad_norm": 2.3066678047180176, + "learning_rate": 1e-06, + "loss": 0.858, + "mean_token_accuracy": 0.7271256446838379, + "num_tokens": 404474151.0, + "step": 15908 + }, + { + "epoch": 1.7470898308807379, + "grad_norm": 2.690131425857544, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7287514209747314, + "num_tokens": 404492130.0, + "step": 15909 + }, + { + "epoch": 1.7471996485833516, + "grad_norm": 2.6122031211853027, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7367070913314819, + "num_tokens": 404510910.0, + "step": 15910 + }, + { + "epoch": 1.7473094662859654, + "grad_norm": 1.978630781173706, + "learning_rate": 1e-06, + "loss": 0.9444, + "mean_token_accuracy": 0.7099321484565735, + "num_tokens": 404543412.0, + "step": 15911 + }, + { + "epoch": 1.747419283988579, + "grad_norm": 2.145028829574585, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7283798456192017, + "num_tokens": 404569061.0, + "step": 15912 + }, + { + "epoch": 1.7475291016911925, + "grad_norm": 2.220731735229492, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7002291083335876, + "num_tokens": 404595158.0, + "step": 15913 + }, + { + "epoch": 1.7476389193938062, + "grad_norm": 2.252742052078247, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7000012397766113, + "num_tokens": 404620854.0, + "step": 15914 + }, + { + "epoch": 1.74774873709642, + "grad_norm": 2.002140760421753, + "learning_rate": 1e-06, + "loss": 0.9776, + "mean_token_accuracy": 0.7006336450576782, + "num_tokens": 404654175.0, + "step": 15915 + }, + { + "epoch": 1.7478585547990337, + "grad_norm": 2.045604705810547, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7047373056411743, + "num_tokens": 404683084.0, + "step": 15916 + }, + { + "epoch": 1.7479683725016473, + "grad_norm": 2.2010107040405273, + "learning_rate": 1e-06, + "loss": 0.8782, + "mean_token_accuracy": 0.7253339290618896, + "num_tokens": 404707686.0, + "step": 15917 + }, + { + "epoch": 1.7480781902042608, + "grad_norm": 2.2811906337738037, + "learning_rate": 1e-06, + "loss": 0.8912, + "mean_token_accuracy": 0.7229668498039246, + "num_tokens": 404733709.0, + "step": 15918 + }, + { + "epoch": 1.7481880079068746, + "grad_norm": 2.029297351837158, + "learning_rate": 1e-06, + "loss": 0.8568, + "mean_token_accuracy": 0.727433443069458, + "num_tokens": 404763814.0, + "step": 15919 + }, + { + "epoch": 1.7482978256094883, + "grad_norm": 2.0303189754486084, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7176536321640015, + "num_tokens": 404791869.0, + "step": 15920 + }, + { + "epoch": 1.7484076433121019, + "grad_norm": 2.3950185775756836, + "learning_rate": 1e-06, + "loss": 0.8405, + "mean_token_accuracy": 0.7343669533729553, + "num_tokens": 404815040.0, + "step": 15921 + }, + { + "epoch": 1.7485174610147156, + "grad_norm": 2.105966806411743, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7236084342002869, + "num_tokens": 404841215.0, + "step": 15922 + }, + { + "epoch": 1.7486272787173291, + "grad_norm": 2.559627056121826, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7132043838500977, + "num_tokens": 404862307.0, + "step": 15923 + }, + { + "epoch": 1.748737096419943, + "grad_norm": 2.57368540763855, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7044118642807007, + "num_tokens": 404882509.0, + "step": 15924 + }, + { + "epoch": 1.7488469141225567, + "grad_norm": 2.2256197929382324, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.6970520615577698, + "num_tokens": 404910757.0, + "step": 15925 + }, + { + "epoch": 1.7489567318251702, + "grad_norm": 2.0539190769195557, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7164186835289001, + "num_tokens": 404939823.0, + "step": 15926 + }, + { + "epoch": 1.7490665495277837, + "grad_norm": 2.534283399581909, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.723811149597168, + "num_tokens": 404961943.0, + "step": 15927 + }, + { + "epoch": 1.7491763672303975, + "grad_norm": 2.3027164936065674, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7107070684432983, + "num_tokens": 404986024.0, + "step": 15928 + }, + { + "epoch": 1.7492861849330112, + "grad_norm": 2.282698392868042, + "learning_rate": 1e-06, + "loss": 0.8606, + "mean_token_accuracy": 0.7228822708129883, + "num_tokens": 405011427.0, + "step": 15929 + }, + { + "epoch": 1.749396002635625, + "grad_norm": 2.3340868949890137, + "learning_rate": 1e-06, + "loss": 0.9296, + "mean_token_accuracy": 0.7116533517837524, + "num_tokens": 405036251.0, + "step": 15930 + }, + { + "epoch": 1.7495058203382385, + "grad_norm": 2.283062696456909, + "learning_rate": 1e-06, + "loss": 0.9294, + "mean_token_accuracy": 0.7173985242843628, + "num_tokens": 405061553.0, + "step": 15931 + }, + { + "epoch": 1.749615638040852, + "grad_norm": 2.1886658668518066, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7129252552986145, + "num_tokens": 405087973.0, + "step": 15932 + }, + { + "epoch": 1.7497254557434658, + "grad_norm": 2.1965863704681396, + "learning_rate": 1e-06, + "loss": 1.023, + "mean_token_accuracy": 0.7009899616241455, + "num_tokens": 405116220.0, + "step": 15933 + }, + { + "epoch": 1.7498352734460796, + "grad_norm": 2.03322696685791, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.7037116289138794, + "num_tokens": 405147782.0, + "step": 15934 + }, + { + "epoch": 1.7499450911486931, + "grad_norm": 2.106682538986206, + "learning_rate": 1e-06, + "loss": 0.7836, + "mean_token_accuracy": 0.753829836845398, + "num_tokens": 405174234.0, + "step": 15935 + }, + { + "epoch": 1.7500549088513069, + "grad_norm": 2.1071889400482178, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7149970531463623, + "num_tokens": 405202781.0, + "step": 15936 + }, + { + "epoch": 1.7501647265539204, + "grad_norm": 2.4924392700195312, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.7285752296447754, + "num_tokens": 405224313.0, + "step": 15937 + }, + { + "epoch": 1.7502745442565342, + "grad_norm": 2.0252132415771484, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7023193836212158, + "num_tokens": 405254507.0, + "step": 15938 + }, + { + "epoch": 1.750384361959148, + "grad_norm": 2.017925262451172, + "learning_rate": 1e-06, + "loss": 0.7927, + "mean_token_accuracy": 0.7477803230285645, + "num_tokens": 405281877.0, + "step": 15939 + }, + { + "epoch": 1.7504941796617615, + "grad_norm": 2.268373966217041, + "learning_rate": 1e-06, + "loss": 0.8595, + "mean_token_accuracy": 0.7287216186523438, + "num_tokens": 405306882.0, + "step": 15940 + }, + { + "epoch": 1.750603997364375, + "grad_norm": 2.189842700958252, + "learning_rate": 1e-06, + "loss": 0.8791, + "mean_token_accuracy": 0.7251260280609131, + "num_tokens": 405333078.0, + "step": 15941 + }, + { + "epoch": 1.7507138150669888, + "grad_norm": 2.5040440559387207, + "learning_rate": 1e-06, + "loss": 0.7125, + "mean_token_accuracy": 0.758808970451355, + "num_tokens": 405352279.0, + "step": 15942 + }, + { + "epoch": 1.7508236327696025, + "grad_norm": 2.366239070892334, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7036886215209961, + "num_tokens": 405375273.0, + "step": 15943 + }, + { + "epoch": 1.7509334504722163, + "grad_norm": 2.124727249145508, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7113882303237915, + "num_tokens": 405402693.0, + "step": 15944 + }, + { + "epoch": 1.7510432681748298, + "grad_norm": 2.004962921142578, + "learning_rate": 1e-06, + "loss": 0.959, + "mean_token_accuracy": 0.707150936126709, + "num_tokens": 405432629.0, + "step": 15945 + }, + { + "epoch": 1.7511530858774433, + "grad_norm": 2.4353582859039307, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7313218116760254, + "num_tokens": 405454379.0, + "step": 15946 + }, + { + "epoch": 1.751262903580057, + "grad_norm": 2.5367679595947266, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7441065311431885, + "num_tokens": 405474232.0, + "step": 15947 + }, + { + "epoch": 1.7513727212826709, + "grad_norm": 2.0428309440612793, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.6990793347358704, + "num_tokens": 405502855.0, + "step": 15948 + }, + { + "epoch": 1.7514825389852844, + "grad_norm": 2.247792959213257, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7140867710113525, + "num_tokens": 405528666.0, + "step": 15949 + }, + { + "epoch": 1.7515923566878981, + "grad_norm": 2.3059298992156982, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7038599848747253, + "num_tokens": 405555719.0, + "step": 15950 + }, + { + "epoch": 1.7517021743905117, + "grad_norm": 2.289550542831421, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7097380757331848, + "num_tokens": 405581675.0, + "step": 15951 + }, + { + "epoch": 1.7518119920931254, + "grad_norm": 2.2329599857330322, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7336426973342896, + "num_tokens": 405606377.0, + "step": 15952 + }, + { + "epoch": 1.7519218097957392, + "grad_norm": 2.2635862827301025, + "learning_rate": 1e-06, + "loss": 0.9731, + "mean_token_accuracy": 0.6982761025428772, + "num_tokens": 405633285.0, + "step": 15953 + }, + { + "epoch": 1.7520316274983527, + "grad_norm": 2.3911168575286865, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7248771786689758, + "num_tokens": 405655787.0, + "step": 15954 + }, + { + "epoch": 1.7521414452009663, + "grad_norm": 2.405691385269165, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7187308669090271, + "num_tokens": 405678116.0, + "step": 15955 + }, + { + "epoch": 1.75225126290358, + "grad_norm": 2.3348448276519775, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7233486175537109, + "num_tokens": 405702153.0, + "step": 15956 + }, + { + "epoch": 1.7523610806061938, + "grad_norm": 2.120588779449463, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.728364884853363, + "num_tokens": 405728812.0, + "step": 15957 + }, + { + "epoch": 1.7524708983088075, + "grad_norm": 2.161402463912964, + "learning_rate": 1e-06, + "loss": 0.8685, + "mean_token_accuracy": 0.7297903895378113, + "num_tokens": 405753777.0, + "step": 15958 + }, + { + "epoch": 1.752580716011421, + "grad_norm": 2.0305116176605225, + "learning_rate": 1e-06, + "loss": 1.0239, + "mean_token_accuracy": 0.6876118183135986, + "num_tokens": 405786140.0, + "step": 15959 + }, + { + "epoch": 1.7526905337140346, + "grad_norm": 2.2405612468719482, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7246713638305664, + "num_tokens": 405812334.0, + "step": 15960 + }, + { + "epoch": 1.7528003514166484, + "grad_norm": 2.422985792160034, + "learning_rate": 1e-06, + "loss": 0.8611, + "mean_token_accuracy": 0.7254570126533508, + "num_tokens": 405833376.0, + "step": 15961 + }, + { + "epoch": 1.7529101691192621, + "grad_norm": 2.3889710903167725, + "learning_rate": 1e-06, + "loss": 0.9599, + "mean_token_accuracy": 0.6992642283439636, + "num_tokens": 405857199.0, + "step": 15962 + }, + { + "epoch": 1.7530199868218757, + "grad_norm": 1.9217808246612549, + "learning_rate": 1e-06, + "loss": 1.0178, + "mean_token_accuracy": 0.6861603260040283, + "num_tokens": 405889469.0, + "step": 15963 + }, + { + "epoch": 1.7531298045244892, + "grad_norm": 1.9580832719802856, + "learning_rate": 1e-06, + "loss": 0.9712, + "mean_token_accuracy": 0.6998382210731506, + "num_tokens": 405923069.0, + "step": 15964 + }, + { + "epoch": 1.753239622227103, + "grad_norm": 2.5244555473327637, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7167186737060547, + "num_tokens": 405944073.0, + "step": 15965 + }, + { + "epoch": 1.7533494399297167, + "grad_norm": 2.2447726726531982, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7183334827423096, + "num_tokens": 405968803.0, + "step": 15966 + }, + { + "epoch": 1.7534592576323305, + "grad_norm": 2.033257246017456, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7214202284812927, + "num_tokens": 405999339.0, + "step": 15967 + }, + { + "epoch": 1.753569075334944, + "grad_norm": 2.2691032886505127, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7209527492523193, + "num_tokens": 406025527.0, + "step": 15968 + }, + { + "epoch": 1.7536788930375575, + "grad_norm": 2.0997447967529297, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7047394514083862, + "num_tokens": 406054527.0, + "step": 15969 + }, + { + "epoch": 1.7537887107401713, + "grad_norm": 2.25561261177063, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7227940559387207, + "num_tokens": 406079366.0, + "step": 15970 + }, + { + "epoch": 1.753898528442785, + "grad_norm": 2.345244884490967, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7261906862258911, + "num_tokens": 406103535.0, + "step": 15971 + }, + { + "epoch": 1.7540083461453988, + "grad_norm": 2.425220012664795, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7203710675239563, + "num_tokens": 406126099.0, + "step": 15972 + }, + { + "epoch": 1.7541181638480123, + "grad_norm": 2.3957934379577637, + "learning_rate": 1e-06, + "loss": 0.9367, + "mean_token_accuracy": 0.7139103412628174, + "num_tokens": 406150620.0, + "step": 15973 + }, + { + "epoch": 1.7542279815506259, + "grad_norm": 2.502530574798584, + "learning_rate": 1e-06, + "loss": 0.9107, + "mean_token_accuracy": 0.7140548825263977, + "num_tokens": 406173445.0, + "step": 15974 + }, + { + "epoch": 1.7543377992532396, + "grad_norm": 2.031853199005127, + "learning_rate": 1e-06, + "loss": 0.8321, + "mean_token_accuracy": 0.7385969161987305, + "num_tokens": 406201987.0, + "step": 15975 + }, + { + "epoch": 1.7544476169558534, + "grad_norm": 2.391831159591675, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7065161466598511, + "num_tokens": 406226202.0, + "step": 15976 + }, + { + "epoch": 1.754557434658467, + "grad_norm": 2.240412712097168, + "learning_rate": 1e-06, + "loss": 0.8733, + "mean_token_accuracy": 0.7230265736579895, + "num_tokens": 406250676.0, + "step": 15977 + }, + { + "epoch": 1.7546672523610805, + "grad_norm": 1.9940992593765259, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7014951705932617, + "num_tokens": 406282076.0, + "step": 15978 + }, + { + "epoch": 1.7547770700636942, + "grad_norm": 2.1564157009124756, + "learning_rate": 1e-06, + "loss": 0.9246, + "mean_token_accuracy": 0.7135686874389648, + "num_tokens": 406308942.0, + "step": 15979 + }, + { + "epoch": 1.754886887766308, + "grad_norm": 2.30325984954834, + "learning_rate": 1e-06, + "loss": 0.8657, + "mean_token_accuracy": 0.7267147898674011, + "num_tokens": 406331351.0, + "step": 15980 + }, + { + "epoch": 1.7549967054689217, + "grad_norm": 2.546020269393921, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7402544021606445, + "num_tokens": 406352250.0, + "step": 15981 + }, + { + "epoch": 1.7551065231715353, + "grad_norm": 2.342648506164551, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.726746678352356, + "num_tokens": 406376200.0, + "step": 15982 + }, + { + "epoch": 1.7552163408741488, + "grad_norm": 2.3475699424743652, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7248284220695496, + "num_tokens": 406400086.0, + "step": 15983 + }, + { + "epoch": 1.7553261585767626, + "grad_norm": 2.014415979385376, + "learning_rate": 1e-06, + "loss": 0.8418, + "mean_token_accuracy": 0.7336212396621704, + "num_tokens": 406429150.0, + "step": 15984 + }, + { + "epoch": 1.7554359762793763, + "grad_norm": 2.093167543411255, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6967353820800781, + "num_tokens": 406461254.0, + "step": 15985 + }, + { + "epoch": 1.7555457939819898, + "grad_norm": 2.7074859142303467, + "learning_rate": 1e-06, + "loss": 0.8318, + "mean_token_accuracy": 0.7379551529884338, + "num_tokens": 406480925.0, + "step": 15986 + }, + { + "epoch": 1.7556556116846036, + "grad_norm": 1.880875587463379, + "learning_rate": 1e-06, + "loss": 0.9636, + "mean_token_accuracy": 0.7033759355545044, + "num_tokens": 406512959.0, + "step": 15987 + }, + { + "epoch": 1.7557654293872171, + "grad_norm": 1.9805952310562134, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7063966989517212, + "num_tokens": 406545292.0, + "step": 15988 + }, + { + "epoch": 1.755875247089831, + "grad_norm": 2.199552297592163, + "learning_rate": 1e-06, + "loss": 0.8911, + "mean_token_accuracy": 0.7232226133346558, + "num_tokens": 406570970.0, + "step": 15989 + }, + { + "epoch": 1.7559850647924446, + "grad_norm": 2.2449870109558105, + "learning_rate": 1e-06, + "loss": 1.0128, + "mean_token_accuracy": 0.6896803379058838, + "num_tokens": 406599032.0, + "step": 15990 + }, + { + "epoch": 1.7560948824950582, + "grad_norm": 2.6162872314453125, + "learning_rate": 1e-06, + "loss": 0.8978, + "mean_token_accuracy": 0.725608229637146, + "num_tokens": 406619190.0, + "step": 15991 + }, + { + "epoch": 1.7562047001976717, + "grad_norm": 2.192279577255249, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7108082175254822, + "num_tokens": 406646019.0, + "step": 15992 + }, + { + "epoch": 1.7563145179002855, + "grad_norm": 2.3367221355438232, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7279312610626221, + "num_tokens": 406668832.0, + "step": 15993 + }, + { + "epoch": 1.7564243356028992, + "grad_norm": 2.31577205657959, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7330927848815918, + "num_tokens": 406693789.0, + "step": 15994 + }, + { + "epoch": 1.756534153305513, + "grad_norm": 2.3316287994384766, + "learning_rate": 1e-06, + "loss": 0.8017, + "mean_token_accuracy": 0.7413773536682129, + "num_tokens": 406717752.0, + "step": 15995 + }, + { + "epoch": 1.7566439710081265, + "grad_norm": 2.1768405437469482, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.710552453994751, + "num_tokens": 406742203.0, + "step": 15996 + }, + { + "epoch": 1.75675378871074, + "grad_norm": 2.4738826751708984, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7284320592880249, + "num_tokens": 406763539.0, + "step": 15997 + }, + { + "epoch": 1.7568636064133538, + "grad_norm": 2.346797227859497, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7367513179779053, + "num_tokens": 406786303.0, + "step": 15998 + }, + { + "epoch": 1.7569734241159676, + "grad_norm": 2.586275339126587, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7457460165023804, + "num_tokens": 406804776.0, + "step": 15999 + }, + { + "epoch": 1.757083241818581, + "grad_norm": 2.3874926567077637, + "learning_rate": 1e-06, + "loss": 0.854, + "mean_token_accuracy": 0.7378599047660828, + "num_tokens": 406827593.0, + "step": 16000 + }, + { + "epoch": 1.7571930595211949, + "grad_norm": 2.031132221221924, + "learning_rate": 1e-06, + "loss": 1.0093, + "mean_token_accuracy": 0.688177764415741, + "num_tokens": 406858527.0, + "step": 16001 + }, + { + "epoch": 1.7573028772238084, + "grad_norm": 2.5037527084350586, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7230794429779053, + "num_tokens": 406880502.0, + "step": 16002 + }, + { + "epoch": 1.7574126949264222, + "grad_norm": 2.229250431060791, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7226228713989258, + "num_tokens": 406907077.0, + "step": 16003 + }, + { + "epoch": 1.757522512629036, + "grad_norm": 2.1246447563171387, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7090092897415161, + "num_tokens": 406935196.0, + "step": 16004 + }, + { + "epoch": 1.7576323303316495, + "grad_norm": 2.1115477085113525, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7206429243087769, + "num_tokens": 406963464.0, + "step": 16005 + }, + { + "epoch": 1.757742148034263, + "grad_norm": 2.209333896636963, + "learning_rate": 1e-06, + "loss": 0.926, + "mean_token_accuracy": 0.7232868671417236, + "num_tokens": 406989764.0, + "step": 16006 + }, + { + "epoch": 1.7578519657368767, + "grad_norm": 2.553074359893799, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7236192226409912, + "num_tokens": 407012667.0, + "step": 16007 + }, + { + "epoch": 1.7579617834394905, + "grad_norm": 2.323917865753174, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.712797224521637, + "num_tokens": 407036234.0, + "step": 16008 + }, + { + "epoch": 1.7580716011421043, + "grad_norm": 2.33050537109375, + "learning_rate": 1e-06, + "loss": 0.9496, + "mean_token_accuracy": 0.70240318775177, + "num_tokens": 407061193.0, + "step": 16009 + }, + { + "epoch": 1.7581814188447178, + "grad_norm": 2.524750232696533, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7075839638710022, + "num_tokens": 407082372.0, + "step": 16010 + }, + { + "epoch": 1.7582912365473313, + "grad_norm": 2.4611752033233643, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7320814728736877, + "num_tokens": 407104326.0, + "step": 16011 + }, + { + "epoch": 1.758401054249945, + "grad_norm": 2.1066861152648926, + "learning_rate": 1e-06, + "loss": 1.0094, + "mean_token_accuracy": 0.6957792043685913, + "num_tokens": 407134479.0, + "step": 16012 + }, + { + "epoch": 1.7585108719525588, + "grad_norm": 2.4094598293304443, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7289329171180725, + "num_tokens": 407156609.0, + "step": 16013 + }, + { + "epoch": 1.7586206896551724, + "grad_norm": 2.0603744983673096, + "learning_rate": 1e-06, + "loss": 0.8178, + "mean_token_accuracy": 0.7463353872299194, + "num_tokens": 407184729.0, + "step": 16014 + }, + { + "epoch": 1.758730507357786, + "grad_norm": 2.072918653488159, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7090930342674255, + "num_tokens": 407213759.0, + "step": 16015 + }, + { + "epoch": 1.7588403250603997, + "grad_norm": 2.1450905799865723, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7064254283905029, + "num_tokens": 407241031.0, + "step": 16016 + }, + { + "epoch": 1.7589501427630134, + "grad_norm": 2.038637399673462, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7140525579452515, + "num_tokens": 407271979.0, + "step": 16017 + }, + { + "epoch": 1.7590599604656272, + "grad_norm": 2.139328956604004, + "learning_rate": 1e-06, + "loss": 0.9703, + "mean_token_accuracy": 0.7002295255661011, + "num_tokens": 407300307.0, + "step": 16018 + }, + { + "epoch": 1.7591697781682407, + "grad_norm": 2.6079416275024414, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7410207390785217, + "num_tokens": 407320220.0, + "step": 16019 + }, + { + "epoch": 1.7592795958708543, + "grad_norm": 2.505281448364258, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7274233102798462, + "num_tokens": 407341801.0, + "step": 16020 + }, + { + "epoch": 1.759389413573468, + "grad_norm": 2.571617364883423, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7364467978477478, + "num_tokens": 407360059.0, + "step": 16021 + }, + { + "epoch": 1.7594992312760818, + "grad_norm": 2.224504232406616, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7122568488121033, + "num_tokens": 407383831.0, + "step": 16022 + }, + { + "epoch": 1.7596090489786955, + "grad_norm": 2.062694787979126, + "learning_rate": 1e-06, + "loss": 0.9709, + "mean_token_accuracy": 0.7133063077926636, + "num_tokens": 407413982.0, + "step": 16023 + }, + { + "epoch": 1.759718866681309, + "grad_norm": 2.1349124908447266, + "learning_rate": 1e-06, + "loss": 0.8093, + "mean_token_accuracy": 0.7424339056015015, + "num_tokens": 407440059.0, + "step": 16024 + }, + { + "epoch": 1.7598286843839226, + "grad_norm": 2.2866299152374268, + "learning_rate": 1e-06, + "loss": 0.9717, + "mean_token_accuracy": 0.702700138092041, + "num_tokens": 407467871.0, + "step": 16025 + }, + { + "epoch": 1.7599385020865363, + "grad_norm": 2.163986921310425, + "learning_rate": 1e-06, + "loss": 0.9406, + "mean_token_accuracy": 0.711022138595581, + "num_tokens": 407494132.0, + "step": 16026 + }, + { + "epoch": 1.76004831978915, + "grad_norm": 2.6893680095672607, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7191666960716248, + "num_tokens": 407512496.0, + "step": 16027 + }, + { + "epoch": 1.7601581374917636, + "grad_norm": 1.8698111772537231, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.6969911456108093, + "num_tokens": 407547310.0, + "step": 16028 + }, + { + "epoch": 1.7602679551943772, + "grad_norm": 2.0991902351379395, + "learning_rate": 1e-06, + "loss": 0.9244, + "mean_token_accuracy": 0.7140908241271973, + "num_tokens": 407577702.0, + "step": 16029 + }, + { + "epoch": 1.760377772896991, + "grad_norm": 2.217430830001831, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7264704704284668, + "num_tokens": 407603559.0, + "step": 16030 + }, + { + "epoch": 1.7604875905996047, + "grad_norm": 2.0186595916748047, + "learning_rate": 1e-06, + "loss": 0.8625, + "mean_token_accuracy": 0.7272088527679443, + "num_tokens": 407632979.0, + "step": 16031 + }, + { + "epoch": 1.7605974083022184, + "grad_norm": 2.262495279312134, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7310270071029663, + "num_tokens": 407657630.0, + "step": 16032 + }, + { + "epoch": 1.760707226004832, + "grad_norm": 2.234185218811035, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7238332629203796, + "num_tokens": 407683297.0, + "step": 16033 + }, + { + "epoch": 1.7608170437074455, + "grad_norm": 2.14613676071167, + "learning_rate": 1e-06, + "loss": 0.9062, + "mean_token_accuracy": 0.714676558971405, + "num_tokens": 407712041.0, + "step": 16034 + }, + { + "epoch": 1.7609268614100593, + "grad_norm": 2.101416826248169, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7384907007217407, + "num_tokens": 407740071.0, + "step": 16035 + }, + { + "epoch": 1.761036679112673, + "grad_norm": 2.266244888305664, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7209205627441406, + "num_tokens": 407766037.0, + "step": 16036 + }, + { + "epoch": 1.7611464968152868, + "grad_norm": 2.1186890602111816, + "learning_rate": 1e-06, + "loss": 0.9845, + "mean_token_accuracy": 0.7017219066619873, + "num_tokens": 407796156.0, + "step": 16037 + }, + { + "epoch": 1.7612563145179003, + "grad_norm": 2.173893451690674, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7184362411499023, + "num_tokens": 407823970.0, + "step": 16038 + }, + { + "epoch": 1.7613661322205139, + "grad_norm": 2.126943826675415, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.708091139793396, + "num_tokens": 407851027.0, + "step": 16039 + }, + { + "epoch": 1.7614759499231276, + "grad_norm": 2.3579888343811035, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.6985791921615601, + "num_tokens": 407874221.0, + "step": 16040 + }, + { + "epoch": 1.7615857676257414, + "grad_norm": 2.226043701171875, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7219273447990417, + "num_tokens": 407901122.0, + "step": 16041 + }, + { + "epoch": 1.761695585328355, + "grad_norm": 2.230563163757324, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7332990765571594, + "num_tokens": 407924812.0, + "step": 16042 + }, + { + "epoch": 1.7618054030309684, + "grad_norm": 2.3033175468444824, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7155610918998718, + "num_tokens": 407951128.0, + "step": 16043 + }, + { + "epoch": 1.7619152207335822, + "grad_norm": 2.125173807144165, + "learning_rate": 1e-06, + "loss": 0.8709, + "mean_token_accuracy": 0.7325098514556885, + "num_tokens": 407979369.0, + "step": 16044 + }, + { + "epoch": 1.762025038436196, + "grad_norm": 2.2080016136169434, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.704957902431488, + "num_tokens": 408005825.0, + "step": 16045 + }, + { + "epoch": 1.7621348561388097, + "grad_norm": 2.4839556217193604, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7387540936470032, + "num_tokens": 408027312.0, + "step": 16046 + }, + { + "epoch": 1.7622446738414232, + "grad_norm": 2.2560503482818604, + "learning_rate": 1e-06, + "loss": 0.9058, + "mean_token_accuracy": 0.7183219194412231, + "num_tokens": 408052106.0, + "step": 16047 + }, + { + "epoch": 1.7623544915440368, + "grad_norm": 2.3563246726989746, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.702373743057251, + "num_tokens": 408073422.0, + "step": 16048 + }, + { + "epoch": 1.7624643092466505, + "grad_norm": 2.515536308288574, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7280336618423462, + "num_tokens": 408094986.0, + "step": 16049 + }, + { + "epoch": 1.7625741269492643, + "grad_norm": 2.7760815620422363, + "learning_rate": 1e-06, + "loss": 0.8195, + "mean_token_accuracy": 0.7340688705444336, + "num_tokens": 408112752.0, + "step": 16050 + }, + { + "epoch": 1.7626839446518778, + "grad_norm": 2.3509061336517334, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7227144241333008, + "num_tokens": 408134912.0, + "step": 16051 + }, + { + "epoch": 1.7627937623544916, + "grad_norm": 2.2664544582366943, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7239948511123657, + "num_tokens": 408158465.0, + "step": 16052 + }, + { + "epoch": 1.7629035800571051, + "grad_norm": 1.909363031387329, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7468205094337463, + "num_tokens": 408187453.0, + "step": 16053 + }, + { + "epoch": 1.7630133977597189, + "grad_norm": 2.495011329650879, + "learning_rate": 1e-06, + "loss": 0.9619, + "mean_token_accuracy": 0.7019228339195251, + "num_tokens": 408209639.0, + "step": 16054 + }, + { + "epoch": 1.7631232154623326, + "grad_norm": 2.3086555004119873, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7197964787483215, + "num_tokens": 408232803.0, + "step": 16055 + }, + { + "epoch": 1.7632330331649462, + "grad_norm": 2.1720476150512695, + "learning_rate": 1e-06, + "loss": 0.9678, + "mean_token_accuracy": 0.7058255672454834, + "num_tokens": 408260282.0, + "step": 16056 + }, + { + "epoch": 1.7633428508675597, + "grad_norm": 2.196164131164551, + "learning_rate": 1e-06, + "loss": 0.8624, + "mean_token_accuracy": 0.7298436760902405, + "num_tokens": 408286032.0, + "step": 16057 + }, + { + "epoch": 1.7634526685701735, + "grad_norm": 2.045736789703369, + "learning_rate": 1e-06, + "loss": 0.8261, + "mean_token_accuracy": 0.7402364015579224, + "num_tokens": 408315280.0, + "step": 16058 + }, + { + "epoch": 1.7635624862727872, + "grad_norm": 2.1070737838745117, + "learning_rate": 1e-06, + "loss": 0.9077, + "mean_token_accuracy": 0.7160622477531433, + "num_tokens": 408342147.0, + "step": 16059 + }, + { + "epoch": 1.763672303975401, + "grad_norm": 2.2490131855010986, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7297812104225159, + "num_tokens": 408367668.0, + "step": 16060 + }, + { + "epoch": 1.7637821216780145, + "grad_norm": 2.3569490909576416, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7243620753288269, + "num_tokens": 408391329.0, + "step": 16061 + }, + { + "epoch": 1.763891939380628, + "grad_norm": 2.326303243637085, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.7086411118507385, + "num_tokens": 408416547.0, + "step": 16062 + }, + { + "epoch": 1.7640017570832418, + "grad_norm": 2.2931294441223145, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.729071319103241, + "num_tokens": 408440894.0, + "step": 16063 + }, + { + "epoch": 1.7641115747858556, + "grad_norm": 2.4459385871887207, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7154898643493652, + "num_tokens": 408461779.0, + "step": 16064 + }, + { + "epoch": 1.764221392488469, + "grad_norm": 2.202502489089966, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7313690185546875, + "num_tokens": 408487799.0, + "step": 16065 + }, + { + "epoch": 1.7643312101910829, + "grad_norm": 2.357151508331299, + "learning_rate": 1e-06, + "loss": 0.8842, + "mean_token_accuracy": 0.7182334661483765, + "num_tokens": 408510576.0, + "step": 16066 + }, + { + "epoch": 1.7644410278936964, + "grad_norm": 2.44883131980896, + "learning_rate": 1e-06, + "loss": 0.8464, + "mean_token_accuracy": 0.7329838275909424, + "num_tokens": 408533072.0, + "step": 16067 + }, + { + "epoch": 1.7645508455963101, + "grad_norm": 2.3840696811676025, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.721012532711029, + "num_tokens": 408556598.0, + "step": 16068 + }, + { + "epoch": 1.764660663298924, + "grad_norm": 1.9574639797210693, + "learning_rate": 1e-06, + "loss": 0.9556, + "mean_token_accuracy": 0.6977865695953369, + "num_tokens": 408587658.0, + "step": 16069 + }, + { + "epoch": 1.7647704810015374, + "grad_norm": 2.540980100631714, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7367778420448303, + "num_tokens": 408607882.0, + "step": 16070 + }, + { + "epoch": 1.764880298704151, + "grad_norm": 2.256845235824585, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7217381000518799, + "num_tokens": 408632063.0, + "step": 16071 + }, + { + "epoch": 1.7649901164067647, + "grad_norm": 2.287410020828247, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.7229520082473755, + "num_tokens": 408660188.0, + "step": 16072 + }, + { + "epoch": 1.7650999341093785, + "grad_norm": 2.789130449295044, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.7167488932609558, + "num_tokens": 408676791.0, + "step": 16073 + }, + { + "epoch": 1.7652097518119922, + "grad_norm": 2.3958916664123535, + "learning_rate": 1e-06, + "loss": 0.9321, + "mean_token_accuracy": 0.7223573327064514, + "num_tokens": 408700381.0, + "step": 16074 + }, + { + "epoch": 1.7653195695146058, + "grad_norm": 2.1575703620910645, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7045797109603882, + "num_tokens": 408728314.0, + "step": 16075 + }, + { + "epoch": 1.7654293872172193, + "grad_norm": 2.6316416263580322, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.7263512015342712, + "num_tokens": 408748316.0, + "step": 16076 + }, + { + "epoch": 1.765539204919833, + "grad_norm": 2.156372547149658, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7339937686920166, + "num_tokens": 408775277.0, + "step": 16077 + }, + { + "epoch": 1.7656490226224468, + "grad_norm": 2.076314687728882, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7388450503349304, + "num_tokens": 408804714.0, + "step": 16078 + }, + { + "epoch": 1.7657588403250604, + "grad_norm": 2.2722692489624023, + "learning_rate": 1e-06, + "loss": 0.8197, + "mean_token_accuracy": 0.7411265969276428, + "num_tokens": 408827663.0, + "step": 16079 + }, + { + "epoch": 1.765868658027674, + "grad_norm": 2.3893353939056396, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7191469669342041, + "num_tokens": 408850787.0, + "step": 16080 + }, + { + "epoch": 1.7659784757302877, + "grad_norm": 2.307199001312256, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7070798277854919, + "num_tokens": 408876031.0, + "step": 16081 + }, + { + "epoch": 1.7660882934329014, + "grad_norm": 2.7707157135009766, + "learning_rate": 1e-06, + "loss": 0.9245, + "mean_token_accuracy": 0.7130337953567505, + "num_tokens": 408893593.0, + "step": 16082 + }, + { + "epoch": 1.7661981111355152, + "grad_norm": 2.1085686683654785, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.709652304649353, + "num_tokens": 408920930.0, + "step": 16083 + }, + { + "epoch": 1.7663079288381287, + "grad_norm": 2.3599767684936523, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7338599562644958, + "num_tokens": 408942980.0, + "step": 16084 + }, + { + "epoch": 1.7664177465407422, + "grad_norm": 2.2337241172790527, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7318904399871826, + "num_tokens": 408967241.0, + "step": 16085 + }, + { + "epoch": 1.766527564243356, + "grad_norm": 2.043720245361328, + "learning_rate": 1e-06, + "loss": 0.9591, + "mean_token_accuracy": 0.6998212337493896, + "num_tokens": 408998686.0, + "step": 16086 + }, + { + "epoch": 1.7666373819459698, + "grad_norm": 2.271209239959717, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.703845739364624, + "num_tokens": 409024402.0, + "step": 16087 + }, + { + "epoch": 1.7667471996485835, + "grad_norm": 2.0151686668395996, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6892502903938293, + "num_tokens": 409055773.0, + "step": 16088 + }, + { + "epoch": 1.766857017351197, + "grad_norm": 2.170569896697998, + "learning_rate": 1e-06, + "loss": 0.8867, + "mean_token_accuracy": 0.716389000415802, + "num_tokens": 409081018.0, + "step": 16089 + }, + { + "epoch": 1.7669668350538106, + "grad_norm": 2.0796008110046387, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7276902794837952, + "num_tokens": 409109304.0, + "step": 16090 + }, + { + "epoch": 1.7670766527564243, + "grad_norm": 2.2949342727661133, + "learning_rate": 1e-06, + "loss": 0.9695, + "mean_token_accuracy": 0.7094597816467285, + "num_tokens": 409133502.0, + "step": 16091 + }, + { + "epoch": 1.767186470459038, + "grad_norm": 2.1250126361846924, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7131047248840332, + "num_tokens": 409160516.0, + "step": 16092 + }, + { + "epoch": 1.7672962881616516, + "grad_norm": 2.1090056896209717, + "learning_rate": 1e-06, + "loss": 0.9129, + "mean_token_accuracy": 0.7121748328208923, + "num_tokens": 409188890.0, + "step": 16093 + }, + { + "epoch": 1.7674061058642652, + "grad_norm": 2.165198802947998, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7274606227874756, + "num_tokens": 409214879.0, + "step": 16094 + }, + { + "epoch": 1.767515923566879, + "grad_norm": 2.3133111000061035, + "learning_rate": 1e-06, + "loss": 0.9079, + "mean_token_accuracy": 0.7187806367874146, + "num_tokens": 409239063.0, + "step": 16095 + }, + { + "epoch": 1.7676257412694927, + "grad_norm": 2.2602035999298096, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7143129110336304, + "num_tokens": 409263258.0, + "step": 16096 + }, + { + "epoch": 1.7677355589721064, + "grad_norm": 2.0547144412994385, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7165398597717285, + "num_tokens": 409291875.0, + "step": 16097 + }, + { + "epoch": 1.76784537667472, + "grad_norm": 2.075737476348877, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7294619083404541, + "num_tokens": 409319171.0, + "step": 16098 + }, + { + "epoch": 1.7679551943773335, + "grad_norm": 2.2521936893463135, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7099575996398926, + "num_tokens": 409343842.0, + "step": 16099 + }, + { + "epoch": 1.7680650120799473, + "grad_norm": 2.252537727355957, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7333238124847412, + "num_tokens": 409369860.0, + "step": 16100 + }, + { + "epoch": 1.768174829782561, + "grad_norm": 2.478888750076294, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7204742431640625, + "num_tokens": 409390860.0, + "step": 16101 + }, + { + "epoch": 1.7682846474851748, + "grad_norm": 2.066964864730835, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7236471176147461, + "num_tokens": 409419295.0, + "step": 16102 + }, + { + "epoch": 1.7683944651877883, + "grad_norm": 2.086583375930786, + "learning_rate": 1e-06, + "loss": 0.9386, + "mean_token_accuracy": 0.7061625123023987, + "num_tokens": 409449040.0, + "step": 16103 + }, + { + "epoch": 1.7685042828904018, + "grad_norm": 2.3229236602783203, + "learning_rate": 1e-06, + "loss": 0.8986, + "mean_token_accuracy": 0.7253677845001221, + "num_tokens": 409472647.0, + "step": 16104 + }, + { + "epoch": 1.7686141005930156, + "grad_norm": 2.340024709701538, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7275736331939697, + "num_tokens": 409496524.0, + "step": 16105 + }, + { + "epoch": 1.7687239182956294, + "grad_norm": 2.187913656234741, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7294610142707825, + "num_tokens": 409523054.0, + "step": 16106 + }, + { + "epoch": 1.768833735998243, + "grad_norm": 2.3645317554473877, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.6949106454849243, + "num_tokens": 409547867.0, + "step": 16107 + }, + { + "epoch": 1.7689435537008564, + "grad_norm": 2.240912675857544, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7275681495666504, + "num_tokens": 409573346.0, + "step": 16108 + }, + { + "epoch": 1.7690533714034702, + "grad_norm": 2.678373098373413, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7307918071746826, + "num_tokens": 409592336.0, + "step": 16109 + }, + { + "epoch": 1.769163189106084, + "grad_norm": 2.1734135150909424, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.6997612714767456, + "num_tokens": 409619086.0, + "step": 16110 + }, + { + "epoch": 1.7692730068086977, + "grad_norm": 2.193359375, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7154645919799805, + "num_tokens": 409645403.0, + "step": 16111 + }, + { + "epoch": 1.7693828245113112, + "grad_norm": 2.0679731369018555, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7242239713668823, + "num_tokens": 409672133.0, + "step": 16112 + }, + { + "epoch": 1.7694926422139248, + "grad_norm": 2.4644384384155273, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7076955437660217, + "num_tokens": 409693575.0, + "step": 16113 + }, + { + "epoch": 1.7696024599165385, + "grad_norm": 2.477372884750366, + "learning_rate": 1e-06, + "loss": 0.9468, + "mean_token_accuracy": 0.7063454389572144, + "num_tokens": 409715524.0, + "step": 16114 + }, + { + "epoch": 1.7697122776191523, + "grad_norm": 2.2821311950683594, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.7078143954277039, + "num_tokens": 409741633.0, + "step": 16115 + }, + { + "epoch": 1.7698220953217658, + "grad_norm": 2.546978235244751, + "learning_rate": 1e-06, + "loss": 0.8687, + "mean_token_accuracy": 0.7306523323059082, + "num_tokens": 409762818.0, + "step": 16116 + }, + { + "epoch": 1.7699319130243796, + "grad_norm": 2.4264976978302, + "learning_rate": 1e-06, + "loss": 0.8403, + "mean_token_accuracy": 0.7310212850570679, + "num_tokens": 409784762.0, + "step": 16117 + }, + { + "epoch": 1.7700417307269931, + "grad_norm": 2.2119266986846924, + "learning_rate": 1e-06, + "loss": 0.9327, + "mean_token_accuracy": 0.7137796878814697, + "num_tokens": 409809941.0, + "step": 16118 + }, + { + "epoch": 1.7701515484296069, + "grad_norm": 2.1663272380828857, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7167543172836304, + "num_tokens": 409837023.0, + "step": 16119 + }, + { + "epoch": 1.7702613661322206, + "grad_norm": 2.1524429321289062, + "learning_rate": 1e-06, + "loss": 0.9569, + "mean_token_accuracy": 0.7119560837745667, + "num_tokens": 409864903.0, + "step": 16120 + }, + { + "epoch": 1.7703711838348342, + "grad_norm": 2.381741762161255, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.7237393856048584, + "num_tokens": 409887540.0, + "step": 16121 + }, + { + "epoch": 1.7704810015374477, + "grad_norm": 2.2704453468322754, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7406224012374878, + "num_tokens": 409912596.0, + "step": 16122 + }, + { + "epoch": 1.7705908192400615, + "grad_norm": 2.5929343700408936, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7425769567489624, + "num_tokens": 409932502.0, + "step": 16123 + }, + { + "epoch": 1.7707006369426752, + "grad_norm": 2.5139877796173096, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7319595217704773, + "num_tokens": 409953172.0, + "step": 16124 + }, + { + "epoch": 1.770810454645289, + "grad_norm": 2.271967649459839, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7196197509765625, + "num_tokens": 409981845.0, + "step": 16125 + }, + { + "epoch": 1.7709202723479025, + "grad_norm": 1.88603937625885, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7173040509223938, + "num_tokens": 410016328.0, + "step": 16126 + }, + { + "epoch": 1.771030090050516, + "grad_norm": 2.059955358505249, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7160025835037231, + "num_tokens": 410045469.0, + "step": 16127 + }, + { + "epoch": 1.7711399077531298, + "grad_norm": 2.0175633430480957, + "learning_rate": 1e-06, + "loss": 0.9843, + "mean_token_accuracy": 0.6950492858886719, + "num_tokens": 410075673.0, + "step": 16128 + }, + { + "epoch": 1.7712497254557436, + "grad_norm": 2.2943294048309326, + "learning_rate": 1e-06, + "loss": 0.8181, + "mean_token_accuracy": 0.7384754419326782, + "num_tokens": 410099336.0, + "step": 16129 + }, + { + "epoch": 1.771359543158357, + "grad_norm": 2.0805952548980713, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.718090832233429, + "num_tokens": 410129175.0, + "step": 16130 + }, + { + "epoch": 1.7714693608609708, + "grad_norm": 2.2824580669403076, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7302699089050293, + "num_tokens": 410153541.0, + "step": 16131 + }, + { + "epoch": 1.7715791785635844, + "grad_norm": 2.235461711883545, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7129629254341125, + "num_tokens": 410179814.0, + "step": 16132 + }, + { + "epoch": 1.7716889962661981, + "grad_norm": 2.4241886138916016, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.7311875820159912, + "num_tokens": 410200379.0, + "step": 16133 + }, + { + "epoch": 1.771798813968812, + "grad_norm": 2.203655481338501, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.721611738204956, + "num_tokens": 410226913.0, + "step": 16134 + }, + { + "epoch": 1.7719086316714254, + "grad_norm": 2.3050262928009033, + "learning_rate": 1e-06, + "loss": 0.822, + "mean_token_accuracy": 0.7390953302383423, + "num_tokens": 410251877.0, + "step": 16135 + }, + { + "epoch": 1.772018449374039, + "grad_norm": 2.3950648307800293, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7165310382843018, + "num_tokens": 410274790.0, + "step": 16136 + }, + { + "epoch": 1.7721282670766527, + "grad_norm": 2.5708930492401123, + "learning_rate": 1e-06, + "loss": 0.8576, + "mean_token_accuracy": 0.730243444442749, + "num_tokens": 410295351.0, + "step": 16137 + }, + { + "epoch": 1.7722380847792665, + "grad_norm": 2.2276532649993896, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7133915424346924, + "num_tokens": 410322028.0, + "step": 16138 + }, + { + "epoch": 1.7723479024818802, + "grad_norm": 2.3579680919647217, + "learning_rate": 1e-06, + "loss": 0.806, + "mean_token_accuracy": 0.7425037622451782, + "num_tokens": 410344788.0, + "step": 16139 + }, + { + "epoch": 1.7724577201844938, + "grad_norm": 2.39665150642395, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7291883826255798, + "num_tokens": 410366817.0, + "step": 16140 + }, + { + "epoch": 1.7725675378871073, + "grad_norm": 2.26320743560791, + "learning_rate": 1e-06, + "loss": 0.8527, + "mean_token_accuracy": 0.7321268916130066, + "num_tokens": 410390450.0, + "step": 16141 + }, + { + "epoch": 1.772677355589721, + "grad_norm": 2.7255547046661377, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7296279072761536, + "num_tokens": 410407309.0, + "step": 16142 + }, + { + "epoch": 1.7727871732923348, + "grad_norm": 2.0940754413604736, + "learning_rate": 1e-06, + "loss": 0.9726, + "mean_token_accuracy": 0.7053824067115784, + "num_tokens": 410434621.0, + "step": 16143 + }, + { + "epoch": 1.7728969909949484, + "grad_norm": 2.369365930557251, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7133352756500244, + "num_tokens": 410457408.0, + "step": 16144 + }, + { + "epoch": 1.7730068086975619, + "grad_norm": 2.0945398807525635, + "learning_rate": 1e-06, + "loss": 0.8508, + "mean_token_accuracy": 0.7318258881568909, + "num_tokens": 410485114.0, + "step": 16145 + }, + { + "epoch": 1.7731166264001756, + "grad_norm": 2.5390186309814453, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7288682460784912, + "num_tokens": 410505690.0, + "step": 16146 + }, + { + "epoch": 1.7732264441027894, + "grad_norm": 2.28220534324646, + "learning_rate": 1e-06, + "loss": 0.9545, + "mean_token_accuracy": 0.7205102443695068, + "num_tokens": 410531096.0, + "step": 16147 + }, + { + "epoch": 1.7733362618054032, + "grad_norm": 1.9824926853179932, + "learning_rate": 1e-06, + "loss": 0.9972, + "mean_token_accuracy": 0.7000181674957275, + "num_tokens": 410562993.0, + "step": 16148 + }, + { + "epoch": 1.7734460795080167, + "grad_norm": 1.9541141986846924, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7080188989639282, + "num_tokens": 410596597.0, + "step": 16149 + }, + { + "epoch": 1.7735558972106302, + "grad_norm": 2.6199464797973633, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7352465391159058, + "num_tokens": 410616978.0, + "step": 16150 + }, + { + "epoch": 1.773665714913244, + "grad_norm": 2.134812593460083, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7341718077659607, + "num_tokens": 410645893.0, + "step": 16151 + }, + { + "epoch": 1.7737755326158577, + "grad_norm": 2.045581817626953, + "learning_rate": 1e-06, + "loss": 0.8423, + "mean_token_accuracy": 0.7352637052536011, + "num_tokens": 410674178.0, + "step": 16152 + }, + { + "epoch": 1.7738853503184715, + "grad_norm": 2.438455820083618, + "learning_rate": 1e-06, + "loss": 0.9192, + "mean_token_accuracy": 0.71173495054245, + "num_tokens": 410698199.0, + "step": 16153 + }, + { + "epoch": 1.773995168021085, + "grad_norm": 2.244234085083008, + "learning_rate": 1e-06, + "loss": 0.8721, + "mean_token_accuracy": 0.7252758741378784, + "num_tokens": 410721546.0, + "step": 16154 + }, + { + "epoch": 1.7741049857236986, + "grad_norm": 2.0272388458251953, + "learning_rate": 1e-06, + "loss": 1.0099, + "mean_token_accuracy": 0.6881780028343201, + "num_tokens": 410754600.0, + "step": 16155 + }, + { + "epoch": 1.7742148034263123, + "grad_norm": 2.2845864295959473, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7052195072174072, + "num_tokens": 410779941.0, + "step": 16156 + }, + { + "epoch": 1.774324621128926, + "grad_norm": 2.249985694885254, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7109845876693726, + "num_tokens": 410804100.0, + "step": 16157 + }, + { + "epoch": 1.7744344388315396, + "grad_norm": 2.0038156509399414, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7231109142303467, + "num_tokens": 410835307.0, + "step": 16158 + }, + { + "epoch": 1.7745442565341532, + "grad_norm": 2.047895669937134, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7245150804519653, + "num_tokens": 410866586.0, + "step": 16159 + }, + { + "epoch": 1.774654074236767, + "grad_norm": 2.3003172874450684, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7265622615814209, + "num_tokens": 410890633.0, + "step": 16160 + }, + { + "epoch": 1.7747638919393807, + "grad_norm": 1.9521268606185913, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7172867655754089, + "num_tokens": 410924111.0, + "step": 16161 + }, + { + "epoch": 1.7748737096419944, + "grad_norm": 2.634309768676758, + "learning_rate": 1e-06, + "loss": 0.8115, + "mean_token_accuracy": 0.7447454929351807, + "num_tokens": 410942904.0, + "step": 16162 + }, + { + "epoch": 1.774983527344608, + "grad_norm": 2.124310255050659, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.719304084777832, + "num_tokens": 410971631.0, + "step": 16163 + }, + { + "epoch": 1.7750933450472215, + "grad_norm": 2.235405206680298, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.726839542388916, + "num_tokens": 410996981.0, + "step": 16164 + }, + { + "epoch": 1.7752031627498353, + "grad_norm": 1.7537363767623901, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.71415114402771, + "num_tokens": 411032833.0, + "step": 16165 + }, + { + "epoch": 1.775312980452449, + "grad_norm": 2.377061367034912, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7099452614784241, + "num_tokens": 411057770.0, + "step": 16166 + }, + { + "epoch": 1.7754227981550625, + "grad_norm": 2.558777332305908, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7349154949188232, + "num_tokens": 411077758.0, + "step": 16167 + }, + { + "epoch": 1.7755326158576763, + "grad_norm": 2.374501943588257, + "learning_rate": 1e-06, + "loss": 0.9241, + "mean_token_accuracy": 0.7098811268806458, + "num_tokens": 411102642.0, + "step": 16168 + }, + { + "epoch": 1.7756424335602898, + "grad_norm": 2.2592954635620117, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7477620244026184, + "num_tokens": 411126336.0, + "step": 16169 + }, + { + "epoch": 1.7757522512629036, + "grad_norm": 2.2874338626861572, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7234424948692322, + "num_tokens": 411150759.0, + "step": 16170 + }, + { + "epoch": 1.7758620689655173, + "grad_norm": 2.1446826457977295, + "learning_rate": 1e-06, + "loss": 0.9562, + "mean_token_accuracy": 0.702007532119751, + "num_tokens": 411180471.0, + "step": 16171 + }, + { + "epoch": 1.7759718866681309, + "grad_norm": 2.0582189559936523, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.7250468730926514, + "num_tokens": 411208950.0, + "step": 16172 + }, + { + "epoch": 1.7760817043707444, + "grad_norm": 2.3586201667785645, + "learning_rate": 1e-06, + "loss": 0.8626, + "mean_token_accuracy": 0.7300267219543457, + "num_tokens": 411230877.0, + "step": 16173 + }, + { + "epoch": 1.7761915220733582, + "grad_norm": 2.1331629753112793, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7020569443702698, + "num_tokens": 411258023.0, + "step": 16174 + }, + { + "epoch": 1.776301339775972, + "grad_norm": 2.503488063812256, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7225633859634399, + "num_tokens": 411278638.0, + "step": 16175 + }, + { + "epoch": 1.7764111574785857, + "grad_norm": 2.293187379837036, + "learning_rate": 1e-06, + "loss": 0.8523, + "mean_token_accuracy": 0.7280729413032532, + "num_tokens": 411302670.0, + "step": 16176 + }, + { + "epoch": 1.7765209751811992, + "grad_norm": 2.2099037170410156, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7031171321868896, + "num_tokens": 411330308.0, + "step": 16177 + }, + { + "epoch": 1.7766307928838128, + "grad_norm": 2.334254503250122, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7233143448829651, + "num_tokens": 411353620.0, + "step": 16178 + }, + { + "epoch": 1.7767406105864265, + "grad_norm": 2.2067863941192627, + "learning_rate": 1e-06, + "loss": 0.9932, + "mean_token_accuracy": 0.6969712376594543, + "num_tokens": 411381022.0, + "step": 16179 + }, + { + "epoch": 1.7768504282890403, + "grad_norm": 2.1439945697784424, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7316604852676392, + "num_tokens": 411407928.0, + "step": 16180 + }, + { + "epoch": 1.7769602459916538, + "grad_norm": 2.2921032905578613, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7161898612976074, + "num_tokens": 411433103.0, + "step": 16181 + }, + { + "epoch": 1.7770700636942676, + "grad_norm": 2.1949779987335205, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.723139762878418, + "num_tokens": 411460010.0, + "step": 16182 + }, + { + "epoch": 1.777179881396881, + "grad_norm": 2.0697073936462402, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7106395959854126, + "num_tokens": 411487991.0, + "step": 16183 + }, + { + "epoch": 1.7772896990994949, + "grad_norm": 2.2104134559631348, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7165603637695312, + "num_tokens": 411512729.0, + "step": 16184 + }, + { + "epoch": 1.7773995168021086, + "grad_norm": 1.8876304626464844, + "learning_rate": 1e-06, + "loss": 0.9526, + "mean_token_accuracy": 0.7023372650146484, + "num_tokens": 411548485.0, + "step": 16185 + }, + { + "epoch": 1.7775093345047221, + "grad_norm": 2.0264759063720703, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7083758115768433, + "num_tokens": 411577929.0, + "step": 16186 + }, + { + "epoch": 1.7776191522073357, + "grad_norm": 2.054619789123535, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7315006852149963, + "num_tokens": 411605161.0, + "step": 16187 + }, + { + "epoch": 1.7777289699099494, + "grad_norm": 2.7443666458129883, + "learning_rate": 1e-06, + "loss": 0.8092, + "mean_token_accuracy": 0.7454939484596252, + "num_tokens": 411622328.0, + "step": 16188 + }, + { + "epoch": 1.7778387876125632, + "grad_norm": 2.45102858543396, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7240632772445679, + "num_tokens": 411642749.0, + "step": 16189 + }, + { + "epoch": 1.777948605315177, + "grad_norm": 2.352525472640991, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7215471267700195, + "num_tokens": 411665505.0, + "step": 16190 + }, + { + "epoch": 1.7780584230177905, + "grad_norm": 2.2132296562194824, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.734641969203949, + "num_tokens": 411694015.0, + "step": 16191 + }, + { + "epoch": 1.778168240720404, + "grad_norm": 2.529076099395752, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7271426916122437, + "num_tokens": 411716482.0, + "step": 16192 + }, + { + "epoch": 1.7782780584230178, + "grad_norm": 2.1756820678710938, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6969327926635742, + "num_tokens": 411744940.0, + "step": 16193 + }, + { + "epoch": 1.7783878761256315, + "grad_norm": 2.283557415008545, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7095972895622253, + "num_tokens": 411770242.0, + "step": 16194 + }, + { + "epoch": 1.778497693828245, + "grad_norm": 2.1724843978881836, + "learning_rate": 1e-06, + "loss": 0.9542, + "mean_token_accuracy": 0.7048029899597168, + "num_tokens": 411796688.0, + "step": 16195 + }, + { + "epoch": 1.7786075115308586, + "grad_norm": 2.051893949508667, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.7374011278152466, + "num_tokens": 411823562.0, + "step": 16196 + }, + { + "epoch": 1.7787173292334724, + "grad_norm": 2.06778621673584, + "learning_rate": 1e-06, + "loss": 0.968, + "mean_token_accuracy": 0.6949600577354431, + "num_tokens": 411854691.0, + "step": 16197 + }, + { + "epoch": 1.7788271469360861, + "grad_norm": 2.3120625019073486, + "learning_rate": 1e-06, + "loss": 0.8614, + "mean_token_accuracy": 0.7261977195739746, + "num_tokens": 411878535.0, + "step": 16198 + }, + { + "epoch": 1.7789369646386999, + "grad_norm": 2.198866128921509, + "learning_rate": 1e-06, + "loss": 0.9713, + "mean_token_accuracy": 0.7003898024559021, + "num_tokens": 411907070.0, + "step": 16199 + }, + { + "epoch": 1.7790467823413134, + "grad_norm": 2.2546091079711914, + "learning_rate": 1e-06, + "loss": 0.9249, + "mean_token_accuracy": 0.7098150253295898, + "num_tokens": 411931292.0, + "step": 16200 + }, + { + "epoch": 1.779156600043927, + "grad_norm": 2.605539083480835, + "learning_rate": 1e-06, + "loss": 0.8209, + "mean_token_accuracy": 0.7385950088500977, + "num_tokens": 411949890.0, + "step": 16201 + }, + { + "epoch": 1.7792664177465407, + "grad_norm": 2.466796875, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7330394983291626, + "num_tokens": 411970985.0, + "step": 16202 + }, + { + "epoch": 1.7793762354491545, + "grad_norm": 2.272636890411377, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7148395776748657, + "num_tokens": 411996744.0, + "step": 16203 + }, + { + "epoch": 1.7794860531517682, + "grad_norm": 2.1896302700042725, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7211154699325562, + "num_tokens": 412021366.0, + "step": 16204 + }, + { + "epoch": 1.7795958708543818, + "grad_norm": 2.0012598037719727, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7261062264442444, + "num_tokens": 412052431.0, + "step": 16205 + }, + { + "epoch": 1.7797056885569953, + "grad_norm": 2.204777956008911, + "learning_rate": 1e-06, + "loss": 0.8943, + "mean_token_accuracy": 0.7290809154510498, + "num_tokens": 412077761.0, + "step": 16206 + }, + { + "epoch": 1.779815506259609, + "grad_norm": 2.0843698978424072, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7085875868797302, + "num_tokens": 412107512.0, + "step": 16207 + }, + { + "epoch": 1.7799253239622228, + "grad_norm": 2.4614486694335938, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.718052864074707, + "num_tokens": 412129962.0, + "step": 16208 + }, + { + "epoch": 1.7800351416648363, + "grad_norm": 2.404607057571411, + "learning_rate": 1e-06, + "loss": 0.782, + "mean_token_accuracy": 0.7562417387962341, + "num_tokens": 412152629.0, + "step": 16209 + }, + { + "epoch": 1.7801449593674499, + "grad_norm": 2.3166444301605225, + "learning_rate": 1e-06, + "loss": 0.9621, + "mean_token_accuracy": 0.7169520258903503, + "num_tokens": 412176893.0, + "step": 16210 + }, + { + "epoch": 1.7802547770700636, + "grad_norm": 2.232459306716919, + "learning_rate": 1e-06, + "loss": 0.8704, + "mean_token_accuracy": 0.7245959043502808, + "num_tokens": 412203549.0, + "step": 16211 + }, + { + "epoch": 1.7803645947726774, + "grad_norm": 2.2733616828918457, + "learning_rate": 1e-06, + "loss": 0.8515, + "mean_token_accuracy": 0.7389477491378784, + "num_tokens": 412227411.0, + "step": 16212 + }, + { + "epoch": 1.7804744124752911, + "grad_norm": 2.284376621246338, + "learning_rate": 1e-06, + "loss": 0.9256, + "mean_token_accuracy": 0.7141949534416199, + "num_tokens": 412250923.0, + "step": 16213 + }, + { + "epoch": 1.7805842301779047, + "grad_norm": 2.040607452392578, + "learning_rate": 1e-06, + "loss": 0.9482, + "mean_token_accuracy": 0.7022278308868408, + "num_tokens": 412282832.0, + "step": 16214 + }, + { + "epoch": 1.7806940478805182, + "grad_norm": 2.338378667831421, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7146660089492798, + "num_tokens": 412307631.0, + "step": 16215 + }, + { + "epoch": 1.780803865583132, + "grad_norm": 2.811579465866089, + "learning_rate": 1e-06, + "loss": 0.825, + "mean_token_accuracy": 0.7361021041870117, + "num_tokens": 412325632.0, + "step": 16216 + }, + { + "epoch": 1.7809136832857457, + "grad_norm": 2.4140326976776123, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.7001096606254578, + "num_tokens": 412348265.0, + "step": 16217 + }, + { + "epoch": 1.7810235009883595, + "grad_norm": 2.272716999053955, + "learning_rate": 1e-06, + "loss": 0.8579, + "mean_token_accuracy": 0.7314817905426025, + "num_tokens": 412372468.0, + "step": 16218 + }, + { + "epoch": 1.781133318690973, + "grad_norm": 2.11649751663208, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7149530649185181, + "num_tokens": 412401981.0, + "step": 16219 + }, + { + "epoch": 1.7812431363935866, + "grad_norm": 2.069974899291992, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.7013494968414307, + "num_tokens": 412432258.0, + "step": 16220 + }, + { + "epoch": 1.7813529540962003, + "grad_norm": 2.4074809551239014, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7142319083213806, + "num_tokens": 412454431.0, + "step": 16221 + }, + { + "epoch": 1.781462771798814, + "grad_norm": 2.0809733867645264, + "learning_rate": 1e-06, + "loss": 1.018, + "mean_token_accuracy": 0.6904841065406799, + "num_tokens": 412483684.0, + "step": 16222 + }, + { + "epoch": 1.7815725895014276, + "grad_norm": 2.3259334564208984, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7152137756347656, + "num_tokens": 412506094.0, + "step": 16223 + }, + { + "epoch": 1.7816824072040411, + "grad_norm": 2.1689202785491943, + "learning_rate": 1e-06, + "loss": 0.8396, + "mean_token_accuracy": 0.7321798205375671, + "num_tokens": 412531093.0, + "step": 16224 + }, + { + "epoch": 1.781792224906655, + "grad_norm": 2.248509645462036, + "learning_rate": 1e-06, + "loss": 0.9737, + "mean_token_accuracy": 0.7025343179702759, + "num_tokens": 412556886.0, + "step": 16225 + }, + { + "epoch": 1.7819020426092687, + "grad_norm": 2.782449722290039, + "learning_rate": 1e-06, + "loss": 0.8517, + "mean_token_accuracy": 0.7290993928909302, + "num_tokens": 412574628.0, + "step": 16226 + }, + { + "epoch": 1.7820118603118824, + "grad_norm": 2.1077675819396973, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7253019213676453, + "num_tokens": 412603004.0, + "step": 16227 + }, + { + "epoch": 1.782121678014496, + "grad_norm": 2.2130463123321533, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7125579714775085, + "num_tokens": 412629697.0, + "step": 16228 + }, + { + "epoch": 1.7822314957171095, + "grad_norm": 2.122265338897705, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7353084087371826, + "num_tokens": 412656264.0, + "step": 16229 + }, + { + "epoch": 1.7823413134197232, + "grad_norm": 2.412247896194458, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7340457439422607, + "num_tokens": 412679504.0, + "step": 16230 + }, + { + "epoch": 1.782451131122337, + "grad_norm": 2.368042469024658, + "learning_rate": 1e-06, + "loss": 0.9072, + "mean_token_accuracy": 0.7213678359985352, + "num_tokens": 412702623.0, + "step": 16231 + }, + { + "epoch": 1.7825609488249505, + "grad_norm": 2.529698133468628, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.705109179019928, + "num_tokens": 412724482.0, + "step": 16232 + }, + { + "epoch": 1.7826707665275643, + "grad_norm": 2.1946752071380615, + "learning_rate": 1e-06, + "loss": 0.9073, + "mean_token_accuracy": 0.7189109325408936, + "num_tokens": 412750245.0, + "step": 16233 + }, + { + "epoch": 1.7827805842301778, + "grad_norm": 2.457078695297241, + "learning_rate": 1e-06, + "loss": 0.8492, + "mean_token_accuracy": 0.73061203956604, + "num_tokens": 412773256.0, + "step": 16234 + }, + { + "epoch": 1.7828904019327916, + "grad_norm": 2.213134527206421, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7209871411323547, + "num_tokens": 412799587.0, + "step": 16235 + }, + { + "epoch": 1.7830002196354053, + "grad_norm": 2.1416032314300537, + "learning_rate": 1e-06, + "loss": 0.9641, + "mean_token_accuracy": 0.7006077170372009, + "num_tokens": 412827695.0, + "step": 16236 + }, + { + "epoch": 1.7831100373380189, + "grad_norm": 2.1178386211395264, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7129537463188171, + "num_tokens": 412855061.0, + "step": 16237 + }, + { + "epoch": 1.7832198550406324, + "grad_norm": 2.1318325996398926, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7240349650382996, + "num_tokens": 412881480.0, + "step": 16238 + }, + { + "epoch": 1.7833296727432462, + "grad_norm": 2.270157814025879, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7035623788833618, + "num_tokens": 412906313.0, + "step": 16239 + }, + { + "epoch": 1.78343949044586, + "grad_norm": 2.2249011993408203, + "learning_rate": 1e-06, + "loss": 1.0214, + "mean_token_accuracy": 0.7007613182067871, + "num_tokens": 412932799.0, + "step": 16240 + }, + { + "epoch": 1.7835493081484737, + "grad_norm": 2.520707845687866, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7154480814933777, + "num_tokens": 412955585.0, + "step": 16241 + }, + { + "epoch": 1.7836591258510872, + "grad_norm": 2.2098946571350098, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.729263186454773, + "num_tokens": 412982196.0, + "step": 16242 + }, + { + "epoch": 1.7837689435537007, + "grad_norm": 2.28545880317688, + "learning_rate": 1e-06, + "loss": 0.8388, + "mean_token_accuracy": 0.7358511090278625, + "num_tokens": 413007621.0, + "step": 16243 + }, + { + "epoch": 1.7838787612563145, + "grad_norm": 2.485301971435547, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7199323177337646, + "num_tokens": 413029551.0, + "step": 16244 + }, + { + "epoch": 1.7839885789589283, + "grad_norm": 2.2802376747131348, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7390432357788086, + "num_tokens": 413052914.0, + "step": 16245 + }, + { + "epoch": 1.7840983966615418, + "grad_norm": 1.9777387380599976, + "learning_rate": 1e-06, + "loss": 1.0137, + "mean_token_accuracy": 0.6906867027282715, + "num_tokens": 413085277.0, + "step": 16246 + }, + { + "epoch": 1.7842082143641556, + "grad_norm": 2.079335927963257, + "learning_rate": 1e-06, + "loss": 0.909, + "mean_token_accuracy": 0.7171844244003296, + "num_tokens": 413114000.0, + "step": 16247 + }, + { + "epoch": 1.784318032066769, + "grad_norm": 2.051640272140503, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7098636627197266, + "num_tokens": 413145268.0, + "step": 16248 + }, + { + "epoch": 1.7844278497693828, + "grad_norm": 2.1867713928222656, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7312781810760498, + "num_tokens": 413174178.0, + "step": 16249 + }, + { + "epoch": 1.7845376674719966, + "grad_norm": 2.5311450958251953, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7260795831680298, + "num_tokens": 413195215.0, + "step": 16250 + }, + { + "epoch": 1.7846474851746101, + "grad_norm": 2.2285385131835938, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7343778610229492, + "num_tokens": 413221647.0, + "step": 16251 + }, + { + "epoch": 1.7847573028772237, + "grad_norm": 2.2364683151245117, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7224078178405762, + "num_tokens": 413247071.0, + "step": 16252 + }, + { + "epoch": 1.7848671205798374, + "grad_norm": 2.486849546432495, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7401319742202759, + "num_tokens": 413268162.0, + "step": 16253 + }, + { + "epoch": 1.7849769382824512, + "grad_norm": 2.3852829933166504, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7352517247200012, + "num_tokens": 413289595.0, + "step": 16254 + }, + { + "epoch": 1.785086755985065, + "grad_norm": 2.305699110031128, + "learning_rate": 1e-06, + "loss": 0.7898, + "mean_token_accuracy": 0.7527073621749878, + "num_tokens": 413311260.0, + "step": 16255 + }, + { + "epoch": 1.7851965736876785, + "grad_norm": 2.3286876678466797, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7073981761932373, + "num_tokens": 413336583.0, + "step": 16256 + }, + { + "epoch": 1.785306391390292, + "grad_norm": 2.1774516105651855, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.730704665184021, + "num_tokens": 413361732.0, + "step": 16257 + }, + { + "epoch": 1.7854162090929058, + "grad_norm": 2.0611236095428467, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7338844537734985, + "num_tokens": 413390407.0, + "step": 16258 + }, + { + "epoch": 1.7855260267955195, + "grad_norm": 2.2050838470458984, + "learning_rate": 1e-06, + "loss": 0.9337, + "mean_token_accuracy": 0.7205705046653748, + "num_tokens": 413418364.0, + "step": 16259 + }, + { + "epoch": 1.785635844498133, + "grad_norm": 2.2277748584747314, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7244523763656616, + "num_tokens": 413443664.0, + "step": 16260 + }, + { + "epoch": 1.7857456622007466, + "grad_norm": 2.294654607772827, + "learning_rate": 1e-06, + "loss": 0.8877, + "mean_token_accuracy": 0.7260180115699768, + "num_tokens": 413468949.0, + "step": 16261 + }, + { + "epoch": 1.7858554799033604, + "grad_norm": 2.3014755249023438, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.711625337600708, + "num_tokens": 413495002.0, + "step": 16262 + }, + { + "epoch": 1.7859652976059741, + "grad_norm": 2.4208641052246094, + "learning_rate": 1e-06, + "loss": 0.8797, + "mean_token_accuracy": 0.718845546245575, + "num_tokens": 413516898.0, + "step": 16263 + }, + { + "epoch": 1.7860751153085879, + "grad_norm": 2.3496437072753906, + "learning_rate": 1e-06, + "loss": 0.7768, + "mean_token_accuracy": 0.7519888877868652, + "num_tokens": 413539744.0, + "step": 16264 + }, + { + "epoch": 1.7861849330112014, + "grad_norm": 2.0636837482452393, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7331326603889465, + "num_tokens": 413567893.0, + "step": 16265 + }, + { + "epoch": 1.786294750713815, + "grad_norm": 2.5255775451660156, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7332290410995483, + "num_tokens": 413589905.0, + "step": 16266 + }, + { + "epoch": 1.7864045684164287, + "grad_norm": 2.410649061203003, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7277987003326416, + "num_tokens": 413613289.0, + "step": 16267 + }, + { + "epoch": 1.7865143861190425, + "grad_norm": 2.2211050987243652, + "learning_rate": 1e-06, + "loss": 0.9354, + "mean_token_accuracy": 0.7129709124565125, + "num_tokens": 413638923.0, + "step": 16268 + }, + { + "epoch": 1.7866242038216562, + "grad_norm": 2.18172025680542, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7118764519691467, + "num_tokens": 413666673.0, + "step": 16269 + }, + { + "epoch": 1.7867340215242697, + "grad_norm": 2.6714372634887695, + "learning_rate": 1e-06, + "loss": 0.8335, + "mean_token_accuracy": 0.7410876750946045, + "num_tokens": 413684322.0, + "step": 16270 + }, + { + "epoch": 1.7868438392268833, + "grad_norm": 2.7260525226593018, + "learning_rate": 1e-06, + "loss": 0.872, + "mean_token_accuracy": 0.7293668985366821, + "num_tokens": 413703541.0, + "step": 16271 + }, + { + "epoch": 1.786953656929497, + "grad_norm": 2.201814651489258, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.697726309299469, + "num_tokens": 413728429.0, + "step": 16272 + }, + { + "epoch": 1.7870634746321108, + "grad_norm": 2.154794692993164, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7174850702285767, + "num_tokens": 413754909.0, + "step": 16273 + }, + { + "epoch": 1.7871732923347243, + "grad_norm": 2.35343599319458, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7283484935760498, + "num_tokens": 413777072.0, + "step": 16274 + }, + { + "epoch": 1.7872831100373379, + "grad_norm": 2.248319387435913, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7059338688850403, + "num_tokens": 413803228.0, + "step": 16275 + }, + { + "epoch": 1.7873929277399516, + "grad_norm": 2.197329521179199, + "learning_rate": 1e-06, + "loss": 0.9103, + "mean_token_accuracy": 0.7196437120437622, + "num_tokens": 413830543.0, + "step": 16276 + }, + { + "epoch": 1.7875027454425654, + "grad_norm": 2.304966449737549, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7094815969467163, + "num_tokens": 413853810.0, + "step": 16277 + }, + { + "epoch": 1.7876125631451791, + "grad_norm": 2.303973913192749, + "learning_rate": 1e-06, + "loss": 0.877, + "mean_token_accuracy": 0.728153645992279, + "num_tokens": 413875049.0, + "step": 16278 + }, + { + "epoch": 1.7877223808477927, + "grad_norm": 1.9981640577316284, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.728741466999054, + "num_tokens": 413904147.0, + "step": 16279 + }, + { + "epoch": 1.7878321985504062, + "grad_norm": 2.1037886142730713, + "learning_rate": 1e-06, + "loss": 0.9858, + "mean_token_accuracy": 0.6945765018463135, + "num_tokens": 413935485.0, + "step": 16280 + }, + { + "epoch": 1.78794201625302, + "grad_norm": 2.4282023906707764, + "learning_rate": 1e-06, + "loss": 0.9834, + "mean_token_accuracy": 0.69814133644104, + "num_tokens": 413959355.0, + "step": 16281 + }, + { + "epoch": 1.7880518339556337, + "grad_norm": 2.359550714492798, + "learning_rate": 1e-06, + "loss": 0.7628, + "mean_token_accuracy": 0.7547410726547241, + "num_tokens": 413980995.0, + "step": 16282 + }, + { + "epoch": 1.7881616516582475, + "grad_norm": 2.096689462661743, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7173048257827759, + "num_tokens": 414009171.0, + "step": 16283 + }, + { + "epoch": 1.788271469360861, + "grad_norm": 2.3793952465057373, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7341157793998718, + "num_tokens": 414029880.0, + "step": 16284 + }, + { + "epoch": 1.7883812870634745, + "grad_norm": 2.3110783100128174, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7004311084747314, + "num_tokens": 414056994.0, + "step": 16285 + }, + { + "epoch": 1.7884911047660883, + "grad_norm": 2.2943904399871826, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7221426367759705, + "num_tokens": 414082852.0, + "step": 16286 + }, + { + "epoch": 1.788600922468702, + "grad_norm": 1.9226107597351074, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7160429954528809, + "num_tokens": 414116877.0, + "step": 16287 + }, + { + "epoch": 1.7887107401713156, + "grad_norm": 2.449321985244751, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7310117483139038, + "num_tokens": 414139119.0, + "step": 16288 + }, + { + "epoch": 1.7888205578739291, + "grad_norm": 2.1075806617736816, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7132498621940613, + "num_tokens": 414165911.0, + "step": 16289 + }, + { + "epoch": 1.7889303755765429, + "grad_norm": 2.266927480697632, + "learning_rate": 1e-06, + "loss": 0.9356, + "mean_token_accuracy": 0.7124356627464294, + "num_tokens": 414190841.0, + "step": 16290 + }, + { + "epoch": 1.7890401932791566, + "grad_norm": 2.0316061973571777, + "learning_rate": 1e-06, + "loss": 0.9868, + "mean_token_accuracy": 0.6953608989715576, + "num_tokens": 414223994.0, + "step": 16291 + }, + { + "epoch": 1.7891500109817704, + "grad_norm": 2.375988006591797, + "learning_rate": 1e-06, + "loss": 0.9182, + "mean_token_accuracy": 0.7158241271972656, + "num_tokens": 414246284.0, + "step": 16292 + }, + { + "epoch": 1.789259828684384, + "grad_norm": 2.391956329345703, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7071664333343506, + "num_tokens": 414270384.0, + "step": 16293 + }, + { + "epoch": 1.7893696463869975, + "grad_norm": 2.462226629257202, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7288965582847595, + "num_tokens": 414291003.0, + "step": 16294 + }, + { + "epoch": 1.7894794640896112, + "grad_norm": 2.642289161682129, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.7073839902877808, + "num_tokens": 414312253.0, + "step": 16295 + }, + { + "epoch": 1.789589281792225, + "grad_norm": 2.0600764751434326, + "learning_rate": 1e-06, + "loss": 0.9781, + "mean_token_accuracy": 0.7027223110198975, + "num_tokens": 414340996.0, + "step": 16296 + }, + { + "epoch": 1.7896990994948385, + "grad_norm": 2.2964978218078613, + "learning_rate": 1e-06, + "loss": 0.8412, + "mean_token_accuracy": 0.7316162586212158, + "num_tokens": 414365472.0, + "step": 16297 + }, + { + "epoch": 1.7898089171974523, + "grad_norm": 2.2273340225219727, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6918153762817383, + "num_tokens": 414394679.0, + "step": 16298 + }, + { + "epoch": 1.7899187349000658, + "grad_norm": 2.101227283477783, + "learning_rate": 1e-06, + "loss": 0.9732, + "mean_token_accuracy": 0.7088186144828796, + "num_tokens": 414422632.0, + "step": 16299 + }, + { + "epoch": 1.7900285526026796, + "grad_norm": 2.310328722000122, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7108335494995117, + "num_tokens": 414447239.0, + "step": 16300 + }, + { + "epoch": 1.7901383703052933, + "grad_norm": 2.120142936706543, + "learning_rate": 1e-06, + "loss": 0.86, + "mean_token_accuracy": 0.7258199453353882, + "num_tokens": 414473852.0, + "step": 16301 + }, + { + "epoch": 1.7902481880079069, + "grad_norm": 2.086745500564575, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7281103134155273, + "num_tokens": 414502566.0, + "step": 16302 + }, + { + "epoch": 1.7903580057105204, + "grad_norm": 2.683460235595703, + "learning_rate": 1e-06, + "loss": 0.8333, + "mean_token_accuracy": 0.7365230321884155, + "num_tokens": 414520239.0, + "step": 16303 + }, + { + "epoch": 1.7904678234131342, + "grad_norm": 2.1249711513519287, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7037948369979858, + "num_tokens": 414550145.0, + "step": 16304 + }, + { + "epoch": 1.790577641115748, + "grad_norm": 2.1778202056884766, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7229890823364258, + "num_tokens": 414576791.0, + "step": 16305 + }, + { + "epoch": 1.7906874588183617, + "grad_norm": 2.44929575920105, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7364312410354614, + "num_tokens": 414599145.0, + "step": 16306 + }, + { + "epoch": 1.7907972765209752, + "grad_norm": 2.0368812084198, + "learning_rate": 1e-06, + "loss": 0.9541, + "mean_token_accuracy": 0.7142429351806641, + "num_tokens": 414628481.0, + "step": 16307 + }, + { + "epoch": 1.7909070942235887, + "grad_norm": 2.191840410232544, + "learning_rate": 1e-06, + "loss": 0.9364, + "mean_token_accuracy": 0.7100924849510193, + "num_tokens": 414658700.0, + "step": 16308 + }, + { + "epoch": 1.7910169119262025, + "grad_norm": 2.3494696617126465, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7221941947937012, + "num_tokens": 414682017.0, + "step": 16309 + }, + { + "epoch": 1.7911267296288162, + "grad_norm": 2.3133456707000732, + "learning_rate": 1e-06, + "loss": 0.9396, + "mean_token_accuracy": 0.7097724676132202, + "num_tokens": 414706640.0, + "step": 16310 + }, + { + "epoch": 1.7912365473314298, + "grad_norm": 2.6216232776641846, + "learning_rate": 1e-06, + "loss": 0.8219, + "mean_token_accuracy": 0.7352229356765747, + "num_tokens": 414726040.0, + "step": 16311 + }, + { + "epoch": 1.7913463650340435, + "grad_norm": 2.280682325363159, + "learning_rate": 1e-06, + "loss": 0.8635, + "mean_token_accuracy": 0.7289018630981445, + "num_tokens": 414750444.0, + "step": 16312 + }, + { + "epoch": 1.791456182736657, + "grad_norm": 2.2715797424316406, + "learning_rate": 1e-06, + "loss": 0.9481, + "mean_token_accuracy": 0.7049990892410278, + "num_tokens": 414773968.0, + "step": 16313 + }, + { + "epoch": 1.7915660004392708, + "grad_norm": 2.428600788116455, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7277884483337402, + "num_tokens": 414796490.0, + "step": 16314 + }, + { + "epoch": 1.7916758181418846, + "grad_norm": 2.3276124000549316, + "learning_rate": 1e-06, + "loss": 0.8479, + "mean_token_accuracy": 0.7288784384727478, + "num_tokens": 414819592.0, + "step": 16315 + }, + { + "epoch": 1.7917856358444981, + "grad_norm": 2.858809471130371, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7355210185050964, + "num_tokens": 414836160.0, + "step": 16316 + }, + { + "epoch": 1.7918954535471117, + "grad_norm": 2.464249849319458, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7218664884567261, + "num_tokens": 414857596.0, + "step": 16317 + }, + { + "epoch": 1.7920052712497254, + "grad_norm": 2.1641061305999756, + "learning_rate": 1e-06, + "loss": 0.9748, + "mean_token_accuracy": 0.6968232989311218, + "num_tokens": 414884563.0, + "step": 16318 + }, + { + "epoch": 1.7921150889523392, + "grad_norm": 1.9877665042877197, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.71644127368927, + "num_tokens": 414916224.0, + "step": 16319 + }, + { + "epoch": 1.792224906654953, + "grad_norm": 2.2306318283081055, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7232645750045776, + "num_tokens": 414944058.0, + "step": 16320 + }, + { + "epoch": 1.7923347243575665, + "grad_norm": 2.0652949810028076, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7126857042312622, + "num_tokens": 414972163.0, + "step": 16321 + }, + { + "epoch": 1.79244454206018, + "grad_norm": 2.257814645767212, + "learning_rate": 1e-06, + "loss": 1.0038, + "mean_token_accuracy": 0.6982510089874268, + "num_tokens": 414999659.0, + "step": 16322 + }, + { + "epoch": 1.7925543597627938, + "grad_norm": 2.426666021347046, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7284481525421143, + "num_tokens": 415022689.0, + "step": 16323 + }, + { + "epoch": 1.7926641774654075, + "grad_norm": 2.097504138946533, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.740053653717041, + "num_tokens": 415048972.0, + "step": 16324 + }, + { + "epoch": 1.792773995168021, + "grad_norm": 2.229320764541626, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7260728478431702, + "num_tokens": 415075652.0, + "step": 16325 + }, + { + "epoch": 1.7928838128706346, + "grad_norm": 2.3691840171813965, + "learning_rate": 1e-06, + "loss": 0.9959, + "mean_token_accuracy": 0.6940346360206604, + "num_tokens": 415102556.0, + "step": 16326 + }, + { + "epoch": 1.7929936305732483, + "grad_norm": 2.2965993881225586, + "learning_rate": 1e-06, + "loss": 0.9518, + "mean_token_accuracy": 0.7116443514823914, + "num_tokens": 415128293.0, + "step": 16327 + }, + { + "epoch": 1.793103448275862, + "grad_norm": 2.037778615951538, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.732837975025177, + "num_tokens": 415156322.0, + "step": 16328 + }, + { + "epoch": 1.7932132659784759, + "grad_norm": 2.656116485595703, + "learning_rate": 1e-06, + "loss": 0.8004, + "mean_token_accuracy": 0.7420413494110107, + "num_tokens": 415175065.0, + "step": 16329 + }, + { + "epoch": 1.7933230836810894, + "grad_norm": 2.288322925567627, + "learning_rate": 1e-06, + "loss": 0.9219, + "mean_token_accuracy": 0.7137864828109741, + "num_tokens": 415199598.0, + "step": 16330 + }, + { + "epoch": 1.793432901383703, + "grad_norm": 1.8777590990066528, + "learning_rate": 1e-06, + "loss": 0.9761, + "mean_token_accuracy": 0.7011829614639282, + "num_tokens": 415235486.0, + "step": 16331 + }, + { + "epoch": 1.7935427190863167, + "grad_norm": 2.6012179851531982, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.724897027015686, + "num_tokens": 415258036.0, + "step": 16332 + }, + { + "epoch": 1.7936525367889304, + "grad_norm": 2.3933966159820557, + "learning_rate": 1e-06, + "loss": 0.8603, + "mean_token_accuracy": 0.7309711575508118, + "num_tokens": 415281233.0, + "step": 16333 + }, + { + "epoch": 1.7937623544915442, + "grad_norm": 2.296394109725952, + "learning_rate": 1e-06, + "loss": 0.8925, + "mean_token_accuracy": 0.7244178056716919, + "num_tokens": 415307282.0, + "step": 16334 + }, + { + "epoch": 1.7938721721941577, + "grad_norm": 2.3064212799072266, + "learning_rate": 1e-06, + "loss": 0.8893, + "mean_token_accuracy": 0.7286427021026611, + "num_tokens": 415330451.0, + "step": 16335 + }, + { + "epoch": 1.7939819898967713, + "grad_norm": 2.2631826400756836, + "learning_rate": 1e-06, + "loss": 0.7952, + "mean_token_accuracy": 0.7496680617332458, + "num_tokens": 415353696.0, + "step": 16336 + }, + { + "epoch": 1.794091807599385, + "grad_norm": 2.3802030086517334, + "learning_rate": 1e-06, + "loss": 0.8245, + "mean_token_accuracy": 0.7323088645935059, + "num_tokens": 415375814.0, + "step": 16337 + }, + { + "epoch": 1.7942016253019988, + "grad_norm": 2.0407469272613525, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7229554653167725, + "num_tokens": 415403427.0, + "step": 16338 + }, + { + "epoch": 1.7943114430046123, + "grad_norm": 1.9943310022354126, + "learning_rate": 1e-06, + "loss": 1.0098, + "mean_token_accuracy": 0.6981326937675476, + "num_tokens": 415435863.0, + "step": 16339 + }, + { + "epoch": 1.7944212607072259, + "grad_norm": 2.3759877681732178, + "learning_rate": 1e-06, + "loss": 0.867, + "mean_token_accuracy": 0.7281688451766968, + "num_tokens": 415458801.0, + "step": 16340 + }, + { + "epoch": 1.7945310784098396, + "grad_norm": 2.5903210639953613, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7174170017242432, + "num_tokens": 415479101.0, + "step": 16341 + }, + { + "epoch": 1.7946408961124534, + "grad_norm": 2.240147829055786, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7161407470703125, + "num_tokens": 415505860.0, + "step": 16342 + }, + { + "epoch": 1.7947507138150671, + "grad_norm": 2.0797131061553955, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7246361374855042, + "num_tokens": 415535258.0, + "step": 16343 + }, + { + "epoch": 1.7948605315176807, + "grad_norm": 2.0796303749084473, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7258505821228027, + "num_tokens": 415564638.0, + "step": 16344 + }, + { + "epoch": 1.7949703492202942, + "grad_norm": 2.0822768211364746, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7152935266494751, + "num_tokens": 415597008.0, + "step": 16345 + }, + { + "epoch": 1.795080166922908, + "grad_norm": 3.049487829208374, + "learning_rate": 1e-06, + "loss": 0.8294, + "mean_token_accuracy": 0.7323240041732788, + "num_tokens": 415612394.0, + "step": 16346 + }, + { + "epoch": 1.7951899846255217, + "grad_norm": 2.2596371173858643, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7350286841392517, + "num_tokens": 415637050.0, + "step": 16347 + }, + { + "epoch": 1.7952998023281352, + "grad_norm": 2.164644718170166, + "learning_rate": 1e-06, + "loss": 0.9944, + "mean_token_accuracy": 0.6945044994354248, + "num_tokens": 415663410.0, + "step": 16348 + }, + { + "epoch": 1.795409620030749, + "grad_norm": 2.303696870803833, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7240767478942871, + "num_tokens": 415688202.0, + "step": 16349 + }, + { + "epoch": 1.7955194377333625, + "grad_norm": 2.441140651702881, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7127664685249329, + "num_tokens": 415711565.0, + "step": 16350 + }, + { + "epoch": 1.7956292554359763, + "grad_norm": 2.332345724105835, + "learning_rate": 1e-06, + "loss": 0.9916, + "mean_token_accuracy": 0.6953034996986389, + "num_tokens": 415737583.0, + "step": 16351 + }, + { + "epoch": 1.79573907313859, + "grad_norm": 2.2069191932678223, + "learning_rate": 1e-06, + "loss": 0.7845, + "mean_token_accuracy": 0.7556838989257812, + "num_tokens": 415761930.0, + "step": 16352 + }, + { + "epoch": 1.7958488908412036, + "grad_norm": 2.441561460494995, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7375147342681885, + "num_tokens": 415787014.0, + "step": 16353 + }, + { + "epoch": 1.7959587085438171, + "grad_norm": 2.495119094848633, + "learning_rate": 1e-06, + "loss": 0.9207, + "mean_token_accuracy": 0.7160557508468628, + "num_tokens": 415810348.0, + "step": 16354 + }, + { + "epoch": 1.7960685262464309, + "grad_norm": 2.0779614448547363, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7378430962562561, + "num_tokens": 415839671.0, + "step": 16355 + }, + { + "epoch": 1.7961783439490446, + "grad_norm": 1.990387201309204, + "learning_rate": 1e-06, + "loss": 0.9344, + "mean_token_accuracy": 0.7146216630935669, + "num_tokens": 415871622.0, + "step": 16356 + }, + { + "epoch": 1.7962881616516584, + "grad_norm": 2.6051673889160156, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7159932851791382, + "num_tokens": 415892491.0, + "step": 16357 + }, + { + "epoch": 1.796397979354272, + "grad_norm": 1.8605507612228394, + "learning_rate": 1e-06, + "loss": 0.932, + "mean_token_accuracy": 0.7162225246429443, + "num_tokens": 415927191.0, + "step": 16358 + }, + { + "epoch": 1.7965077970568855, + "grad_norm": 2.1077845096588135, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6973545551300049, + "num_tokens": 415956625.0, + "step": 16359 + }, + { + "epoch": 1.7966176147594992, + "grad_norm": 2.1904594898223877, + "learning_rate": 1e-06, + "loss": 0.9237, + "mean_token_accuracy": 0.7095688581466675, + "num_tokens": 415981957.0, + "step": 16360 + }, + { + "epoch": 1.796727432462113, + "grad_norm": 2.296279191970825, + "learning_rate": 1e-06, + "loss": 0.8182, + "mean_token_accuracy": 0.7465807199478149, + "num_tokens": 416005499.0, + "step": 16361 + }, + { + "epoch": 1.7968372501647265, + "grad_norm": 2.3883392810821533, + "learning_rate": 1e-06, + "loss": 0.8914, + "mean_token_accuracy": 0.7128365635871887, + "num_tokens": 416028016.0, + "step": 16362 + }, + { + "epoch": 1.7969470678673403, + "grad_norm": 2.460784912109375, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7197798490524292, + "num_tokens": 416051233.0, + "step": 16363 + }, + { + "epoch": 1.7970568855699538, + "grad_norm": 2.227989912033081, + "learning_rate": 1e-06, + "loss": 0.8461, + "mean_token_accuracy": 0.7342827320098877, + "num_tokens": 416076455.0, + "step": 16364 + }, + { + "epoch": 1.7971667032725676, + "grad_norm": 1.9593485593795776, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7094192504882812, + "num_tokens": 416109029.0, + "step": 16365 + }, + { + "epoch": 1.7972765209751813, + "grad_norm": 2.3192527294158936, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7322739362716675, + "num_tokens": 416133416.0, + "step": 16366 + }, + { + "epoch": 1.7973863386777948, + "grad_norm": 2.1855812072753906, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7204275131225586, + "num_tokens": 416158418.0, + "step": 16367 + }, + { + "epoch": 1.7974961563804084, + "grad_norm": 2.208007574081421, + "learning_rate": 1e-06, + "loss": 0.9365, + "mean_token_accuracy": 0.7086923122406006, + "num_tokens": 416189288.0, + "step": 16368 + }, + { + "epoch": 1.7976059740830221, + "grad_norm": 2.1686649322509766, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.729442298412323, + "num_tokens": 416216019.0, + "step": 16369 + }, + { + "epoch": 1.797715791785636, + "grad_norm": 2.326052188873291, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7316337823867798, + "num_tokens": 416239355.0, + "step": 16370 + }, + { + "epoch": 1.7978256094882497, + "grad_norm": 2.0924203395843506, + "learning_rate": 1e-06, + "loss": 0.8428, + "mean_token_accuracy": 0.7243059873580933, + "num_tokens": 416266851.0, + "step": 16371 + }, + { + "epoch": 1.7979354271908632, + "grad_norm": 2.565605878829956, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7105990052223206, + "num_tokens": 416286750.0, + "step": 16372 + }, + { + "epoch": 1.7980452448934767, + "grad_norm": 1.9483532905578613, + "learning_rate": 1e-06, + "loss": 0.8553, + "mean_token_accuracy": 0.7262958288192749, + "num_tokens": 416318297.0, + "step": 16373 + }, + { + "epoch": 1.7981550625960905, + "grad_norm": 2.672548294067383, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7170374989509583, + "num_tokens": 416338693.0, + "step": 16374 + }, + { + "epoch": 1.7982648802987042, + "grad_norm": 2.191960096359253, + "learning_rate": 1e-06, + "loss": 0.7964, + "mean_token_accuracy": 0.7532060742378235, + "num_tokens": 416362909.0, + "step": 16375 + }, + { + "epoch": 1.7983746980013178, + "grad_norm": 2.1623847484588623, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.7398128509521484, + "num_tokens": 416389106.0, + "step": 16376 + }, + { + "epoch": 1.7984845157039313, + "grad_norm": 2.305835723876953, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7230812907218933, + "num_tokens": 416412609.0, + "step": 16377 + }, + { + "epoch": 1.798594333406545, + "grad_norm": 2.4101314544677734, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.726409912109375, + "num_tokens": 416435527.0, + "step": 16378 + }, + { + "epoch": 1.7987041511091588, + "grad_norm": 2.0200493335723877, + "learning_rate": 1e-06, + "loss": 0.9825, + "mean_token_accuracy": 0.6960668563842773, + "num_tokens": 416466196.0, + "step": 16379 + }, + { + "epoch": 1.7988139688117726, + "grad_norm": 2.1236796379089355, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7111471891403198, + "num_tokens": 416494011.0, + "step": 16380 + }, + { + "epoch": 1.7989237865143861, + "grad_norm": 2.003734588623047, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7243472337722778, + "num_tokens": 416526847.0, + "step": 16381 + }, + { + "epoch": 1.7990336042169996, + "grad_norm": 2.1395576000213623, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.729059100151062, + "num_tokens": 416554016.0, + "step": 16382 + }, + { + "epoch": 1.7991434219196134, + "grad_norm": 2.0969769954681396, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7110017538070679, + "num_tokens": 416585078.0, + "step": 16383 + }, + { + "epoch": 1.7992532396222272, + "grad_norm": 2.373631715774536, + "learning_rate": 1e-06, + "loss": 0.8348, + "mean_token_accuracy": 0.7440674901008606, + "num_tokens": 416608061.0, + "step": 16384 + }, + { + "epoch": 1.799363057324841, + "grad_norm": 2.3311312198638916, + "learning_rate": 1e-06, + "loss": 0.9407, + "mean_token_accuracy": 0.7027238607406616, + "num_tokens": 416631884.0, + "step": 16385 + }, + { + "epoch": 1.7994728750274545, + "grad_norm": 1.9647705554962158, + "learning_rate": 1e-06, + "loss": 0.9492, + "mean_token_accuracy": 0.7076902389526367, + "num_tokens": 416664041.0, + "step": 16386 + }, + { + "epoch": 1.799582692730068, + "grad_norm": 2.4760897159576416, + "learning_rate": 1e-06, + "loss": 0.9838, + "mean_token_accuracy": 0.6967489719390869, + "num_tokens": 416687907.0, + "step": 16387 + }, + { + "epoch": 1.7996925104326817, + "grad_norm": 2.5410540103912354, + "learning_rate": 1e-06, + "loss": 0.7814, + "mean_token_accuracy": 0.7458846569061279, + "num_tokens": 416707461.0, + "step": 16388 + }, + { + "epoch": 1.7998023281352955, + "grad_norm": 2.4749913215637207, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7266682386398315, + "num_tokens": 416727600.0, + "step": 16389 + }, + { + "epoch": 1.799912145837909, + "grad_norm": 2.162971258163452, + "learning_rate": 1e-06, + "loss": 0.9374, + "mean_token_accuracy": 0.7067281603813171, + "num_tokens": 416753803.0, + "step": 16390 + }, + { + "epoch": 1.8000219635405226, + "grad_norm": 2.100353956222534, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7434175610542297, + "num_tokens": 416779038.0, + "step": 16391 + }, + { + "epoch": 1.8001317812431363, + "grad_norm": 2.205315113067627, + "learning_rate": 1e-06, + "loss": 0.92, + "mean_token_accuracy": 0.713001012802124, + "num_tokens": 416805078.0, + "step": 16392 + }, + { + "epoch": 1.80024159894575, + "grad_norm": 2.295855760574341, + "learning_rate": 1e-06, + "loss": 0.9882, + "mean_token_accuracy": 0.6940155029296875, + "num_tokens": 416828367.0, + "step": 16393 + }, + { + "epoch": 1.8003514166483638, + "grad_norm": 1.9383132457733154, + "learning_rate": 1e-06, + "loss": 0.9582, + "mean_token_accuracy": 0.6984649896621704, + "num_tokens": 416860099.0, + "step": 16394 + }, + { + "epoch": 1.8004612343509774, + "grad_norm": 2.036130666732788, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7187962532043457, + "num_tokens": 416889505.0, + "step": 16395 + }, + { + "epoch": 1.800571052053591, + "grad_norm": 2.3927581310272217, + "learning_rate": 1e-06, + "loss": 0.8944, + "mean_token_accuracy": 0.718286395072937, + "num_tokens": 416914592.0, + "step": 16396 + }, + { + "epoch": 1.8006808697562047, + "grad_norm": 2.2052717208862305, + "learning_rate": 1e-06, + "loss": 0.8881, + "mean_token_accuracy": 0.7206571102142334, + "num_tokens": 416941399.0, + "step": 16397 + }, + { + "epoch": 1.8007906874588184, + "grad_norm": 2.3953800201416016, + "learning_rate": 1e-06, + "loss": 0.8573, + "mean_token_accuracy": 0.7318841218948364, + "num_tokens": 416964334.0, + "step": 16398 + }, + { + "epoch": 1.8009005051614322, + "grad_norm": 2.2939956188201904, + "learning_rate": 1e-06, + "loss": 0.815, + "mean_token_accuracy": 0.7385621070861816, + "num_tokens": 416988120.0, + "step": 16399 + }, + { + "epoch": 1.8010103228640457, + "grad_norm": 2.1221752166748047, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7099525928497314, + "num_tokens": 417014809.0, + "step": 16400 + }, + { + "epoch": 1.8011201405666593, + "grad_norm": 2.583132743835449, + "learning_rate": 1e-06, + "loss": 0.7976, + "mean_token_accuracy": 0.7464072704315186, + "num_tokens": 417034424.0, + "step": 16401 + }, + { + "epoch": 1.801229958269273, + "grad_norm": 2.253599166870117, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7230688333511353, + "num_tokens": 417061024.0, + "step": 16402 + }, + { + "epoch": 1.8013397759718868, + "grad_norm": 2.2446815967559814, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7366162538528442, + "num_tokens": 417085371.0, + "step": 16403 + }, + { + "epoch": 1.8014495936745003, + "grad_norm": 2.5303196907043457, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7274208068847656, + "num_tokens": 417105492.0, + "step": 16404 + }, + { + "epoch": 1.8015594113771138, + "grad_norm": 2.3847784996032715, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7202863693237305, + "num_tokens": 417128577.0, + "step": 16405 + }, + { + "epoch": 1.8016692290797276, + "grad_norm": 2.0431575775146484, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7235808372497559, + "num_tokens": 417157578.0, + "step": 16406 + }, + { + "epoch": 1.8017790467823414, + "grad_norm": 2.2389943599700928, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.708646297454834, + "num_tokens": 417185843.0, + "step": 16407 + }, + { + "epoch": 1.801888864484955, + "grad_norm": 2.376297950744629, + "learning_rate": 1e-06, + "loss": 0.949, + "mean_token_accuracy": 0.7160426378250122, + "num_tokens": 417210463.0, + "step": 16408 + }, + { + "epoch": 1.8019986821875686, + "grad_norm": 2.2549777030944824, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.7008944153785706, + "num_tokens": 417239067.0, + "step": 16409 + }, + { + "epoch": 1.8021084998901822, + "grad_norm": 2.0970213413238525, + "learning_rate": 1e-06, + "loss": 0.8811, + "mean_token_accuracy": 0.7289016246795654, + "num_tokens": 417267601.0, + "step": 16410 + }, + { + "epoch": 1.802218317592796, + "grad_norm": 2.300758123397827, + "learning_rate": 1e-06, + "loss": 0.8596, + "mean_token_accuracy": 0.7334950566291809, + "num_tokens": 417292794.0, + "step": 16411 + }, + { + "epoch": 1.8023281352954097, + "grad_norm": 2.0450642108917236, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7300912737846375, + "num_tokens": 417321411.0, + "step": 16412 + }, + { + "epoch": 1.8024379529980232, + "grad_norm": 2.089860200881958, + "learning_rate": 1e-06, + "loss": 0.8837, + "mean_token_accuracy": 0.7258976101875305, + "num_tokens": 417349953.0, + "step": 16413 + }, + { + "epoch": 1.802547770700637, + "grad_norm": 2.333754301071167, + "learning_rate": 1e-06, + "loss": 0.9692, + "mean_token_accuracy": 0.6978163719177246, + "num_tokens": 417374907.0, + "step": 16414 + }, + { + "epoch": 1.8026575884032505, + "grad_norm": 1.9333889484405518, + "learning_rate": 1e-06, + "loss": 1.0348, + "mean_token_accuracy": 0.6850355267524719, + "num_tokens": 417412106.0, + "step": 16415 + }, + { + "epoch": 1.8027674061058643, + "grad_norm": 2.048060178756714, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7209022045135498, + "num_tokens": 417441129.0, + "step": 16416 + }, + { + "epoch": 1.802877223808478, + "grad_norm": 1.9055920839309692, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7131088972091675, + "num_tokens": 417475254.0, + "step": 16417 + }, + { + "epoch": 1.8029870415110916, + "grad_norm": 2.25168514251709, + "learning_rate": 1e-06, + "loss": 0.9467, + "mean_token_accuracy": 0.7074175477027893, + "num_tokens": 417500145.0, + "step": 16418 + }, + { + "epoch": 1.803096859213705, + "grad_norm": 2.094855308532715, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7350982427597046, + "num_tokens": 417526229.0, + "step": 16419 + }, + { + "epoch": 1.8032066769163189, + "grad_norm": 2.3031797409057617, + "learning_rate": 1e-06, + "loss": 0.9819, + "mean_token_accuracy": 0.6984309554100037, + "num_tokens": 417550660.0, + "step": 16420 + }, + { + "epoch": 1.8033164946189326, + "grad_norm": 2.543541431427002, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.6991153359413147, + "num_tokens": 417572246.0, + "step": 16421 + }, + { + "epoch": 1.8034263123215464, + "grad_norm": 2.273522138595581, + "learning_rate": 1e-06, + "loss": 0.9568, + "mean_token_accuracy": 0.7107089161872864, + "num_tokens": 417596283.0, + "step": 16422 + }, + { + "epoch": 1.80353613002416, + "grad_norm": 2.321275472640991, + "learning_rate": 1e-06, + "loss": 0.8854, + "mean_token_accuracy": 0.7162650227546692, + "num_tokens": 417619516.0, + "step": 16423 + }, + { + "epoch": 1.8036459477267734, + "grad_norm": 2.0702269077301025, + "learning_rate": 1e-06, + "loss": 1.0498, + "mean_token_accuracy": 0.6830134987831116, + "num_tokens": 417649604.0, + "step": 16424 + }, + { + "epoch": 1.8037557654293872, + "grad_norm": 2.215796709060669, + "learning_rate": 1e-06, + "loss": 0.9778, + "mean_token_accuracy": 0.6993640065193176, + "num_tokens": 417677704.0, + "step": 16425 + }, + { + "epoch": 1.803865583132001, + "grad_norm": 2.4002742767333984, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7116602659225464, + "num_tokens": 417700748.0, + "step": 16426 + }, + { + "epoch": 1.8039754008346145, + "grad_norm": 2.1574018001556396, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7313253283500671, + "num_tokens": 417726715.0, + "step": 16427 + }, + { + "epoch": 1.8040852185372283, + "grad_norm": 1.9979426860809326, + "learning_rate": 1e-06, + "loss": 0.8215, + "mean_token_accuracy": 0.7415257692337036, + "num_tokens": 417754003.0, + "step": 16428 + }, + { + "epoch": 1.8041950362398418, + "grad_norm": 2.564084053039551, + "learning_rate": 1e-06, + "loss": 0.7256, + "mean_token_accuracy": 0.7657349109649658, + "num_tokens": 417772826.0, + "step": 16429 + }, + { + "epoch": 1.8043048539424555, + "grad_norm": 2.2111170291900635, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7197327613830566, + "num_tokens": 417799182.0, + "step": 16430 + }, + { + "epoch": 1.8044146716450693, + "grad_norm": 2.017888069152832, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.7019129395484924, + "num_tokens": 417832190.0, + "step": 16431 + }, + { + "epoch": 1.8045244893476828, + "grad_norm": 2.4020040035247803, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7296475172042847, + "num_tokens": 417854098.0, + "step": 16432 + }, + { + "epoch": 1.8046343070502964, + "grad_norm": 2.021013021469116, + "learning_rate": 1e-06, + "loss": 0.8088, + "mean_token_accuracy": 0.7336748242378235, + "num_tokens": 417883456.0, + "step": 16433 + }, + { + "epoch": 1.8047441247529101, + "grad_norm": 2.151578903198242, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7217316627502441, + "num_tokens": 417910213.0, + "step": 16434 + }, + { + "epoch": 1.8048539424555239, + "grad_norm": 2.0600805282592773, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7138686180114746, + "num_tokens": 417940595.0, + "step": 16435 + }, + { + "epoch": 1.8049637601581376, + "grad_norm": 2.6776974201202393, + "learning_rate": 1e-06, + "loss": 0.7903, + "mean_token_accuracy": 0.7497403621673584, + "num_tokens": 417959150.0, + "step": 16436 + }, + { + "epoch": 1.8050735778607512, + "grad_norm": 2.2348413467407227, + "learning_rate": 1e-06, + "loss": 0.8908, + "mean_token_accuracy": 0.7193440198898315, + "num_tokens": 417983274.0, + "step": 16437 + }, + { + "epoch": 1.8051833955633647, + "grad_norm": 2.0267555713653564, + "learning_rate": 1e-06, + "loss": 0.9736, + "mean_token_accuracy": 0.6982713341712952, + "num_tokens": 418015141.0, + "step": 16438 + }, + { + "epoch": 1.8052932132659785, + "grad_norm": 2.3099567890167236, + "learning_rate": 1e-06, + "loss": 0.8915, + "mean_token_accuracy": 0.7284232378005981, + "num_tokens": 418038922.0, + "step": 16439 + }, + { + "epoch": 1.8054030309685922, + "grad_norm": 2.2035887241363525, + "learning_rate": 1e-06, + "loss": 1.0478, + "mean_token_accuracy": 0.6813782453536987, + "num_tokens": 418067402.0, + "step": 16440 + }, + { + "epoch": 1.8055128486712058, + "grad_norm": 2.3516297340393066, + "learning_rate": 1e-06, + "loss": 0.9822, + "mean_token_accuracy": 0.6976516246795654, + "num_tokens": 418092238.0, + "step": 16441 + }, + { + "epoch": 1.8056226663738193, + "grad_norm": 2.3333139419555664, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7260252833366394, + "num_tokens": 418117345.0, + "step": 16442 + }, + { + "epoch": 1.805732484076433, + "grad_norm": 2.192307710647583, + "learning_rate": 1e-06, + "loss": 0.9759, + "mean_token_accuracy": 0.7038035988807678, + "num_tokens": 418149803.0, + "step": 16443 + }, + { + "epoch": 1.8058423017790468, + "grad_norm": 2.252518653869629, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7145698070526123, + "num_tokens": 418177128.0, + "step": 16444 + }, + { + "epoch": 1.8059521194816606, + "grad_norm": 2.628645181655884, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7273279428482056, + "num_tokens": 418197634.0, + "step": 16445 + }, + { + "epoch": 1.806061937184274, + "grad_norm": 2.2145586013793945, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7280621528625488, + "num_tokens": 418223487.0, + "step": 16446 + }, + { + "epoch": 1.8061717548868876, + "grad_norm": 2.1968791484832764, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7264270782470703, + "num_tokens": 418250182.0, + "step": 16447 + }, + { + "epoch": 1.8062815725895014, + "grad_norm": 2.40453839302063, + "learning_rate": 1e-06, + "loss": 0.8199, + "mean_token_accuracy": 0.7403379082679749, + "num_tokens": 418272673.0, + "step": 16448 + }, + { + "epoch": 1.8063913902921152, + "grad_norm": 2.0139074325561523, + "learning_rate": 1e-06, + "loss": 0.9434, + "mean_token_accuracy": 0.7082894444465637, + "num_tokens": 418302634.0, + "step": 16449 + }, + { + "epoch": 1.806501207994729, + "grad_norm": 2.3590853214263916, + "learning_rate": 1e-06, + "loss": 0.8198, + "mean_token_accuracy": 0.7407249212265015, + "num_tokens": 418326246.0, + "step": 16450 + }, + { + "epoch": 1.8066110256973424, + "grad_norm": 2.4505462646484375, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7058694958686829, + "num_tokens": 418348934.0, + "step": 16451 + }, + { + "epoch": 1.806720843399956, + "grad_norm": 2.2775471210479736, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7214652299880981, + "num_tokens": 418373636.0, + "step": 16452 + }, + { + "epoch": 1.8068306611025697, + "grad_norm": 2.0143840312957764, + "learning_rate": 1e-06, + "loss": 0.8327, + "mean_token_accuracy": 0.7393825054168701, + "num_tokens": 418401952.0, + "step": 16453 + }, + { + "epoch": 1.8069404788051835, + "grad_norm": 2.2005558013916016, + "learning_rate": 1e-06, + "loss": 0.9105, + "mean_token_accuracy": 0.7272646427154541, + "num_tokens": 418427023.0, + "step": 16454 + }, + { + "epoch": 1.807050296507797, + "grad_norm": 2.527435541152954, + "learning_rate": 1e-06, + "loss": 0.8663, + "mean_token_accuracy": 0.7286669611930847, + "num_tokens": 418448225.0, + "step": 16455 + }, + { + "epoch": 1.8071601142104106, + "grad_norm": 2.062465190887451, + "learning_rate": 1e-06, + "loss": 0.9161, + "mean_token_accuracy": 0.7137572765350342, + "num_tokens": 418477763.0, + "step": 16456 + }, + { + "epoch": 1.8072699319130243, + "grad_norm": 2.2424919605255127, + "learning_rate": 1e-06, + "loss": 0.9162, + "mean_token_accuracy": 0.7146754264831543, + "num_tokens": 418502904.0, + "step": 16457 + }, + { + "epoch": 1.807379749615638, + "grad_norm": 2.1167242527008057, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7041302919387817, + "num_tokens": 418531493.0, + "step": 16458 + }, + { + "epoch": 1.8074895673182518, + "grad_norm": 1.9594990015029907, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7417991161346436, + "num_tokens": 418561396.0, + "step": 16459 + }, + { + "epoch": 1.8075993850208654, + "grad_norm": 2.573277473449707, + "learning_rate": 1e-06, + "loss": 0.7716, + "mean_token_accuracy": 0.7524349093437195, + "num_tokens": 418580668.0, + "step": 16460 + }, + { + "epoch": 1.807709202723479, + "grad_norm": 2.267712354660034, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.719075083732605, + "num_tokens": 418607798.0, + "step": 16461 + }, + { + "epoch": 1.8078190204260927, + "grad_norm": 2.4135117530822754, + "learning_rate": 1e-06, + "loss": 0.8287, + "mean_token_accuracy": 0.7420114874839783, + "num_tokens": 418629802.0, + "step": 16462 + }, + { + "epoch": 1.8079288381287064, + "grad_norm": 1.9271302223205566, + "learning_rate": 1e-06, + "loss": 0.9449, + "mean_token_accuracy": 0.7219057679176331, + "num_tokens": 418662281.0, + "step": 16463 + }, + { + "epoch": 1.8080386558313202, + "grad_norm": 2.2799205780029297, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7155023813247681, + "num_tokens": 418685093.0, + "step": 16464 + }, + { + "epoch": 1.8081484735339337, + "grad_norm": 2.138303518295288, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7239006757736206, + "num_tokens": 418710832.0, + "step": 16465 + }, + { + "epoch": 1.8082582912365472, + "grad_norm": 2.1562001705169678, + "learning_rate": 1e-06, + "loss": 0.9639, + "mean_token_accuracy": 0.7076557874679565, + "num_tokens": 418738095.0, + "step": 16466 + }, + { + "epoch": 1.808368108939161, + "grad_norm": 1.903860092163086, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7162642478942871, + "num_tokens": 418772703.0, + "step": 16467 + }, + { + "epoch": 1.8084779266417748, + "grad_norm": 2.073704481124878, + "learning_rate": 1e-06, + "loss": 0.9657, + "mean_token_accuracy": 0.7029699683189392, + "num_tokens": 418804281.0, + "step": 16468 + }, + { + "epoch": 1.8085877443443883, + "grad_norm": 2.077536106109619, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7194917798042297, + "num_tokens": 418833713.0, + "step": 16469 + }, + { + "epoch": 1.8086975620470018, + "grad_norm": 3.230179786682129, + "learning_rate": 1e-06, + "loss": 0.8931, + "mean_token_accuracy": 0.7251608967781067, + "num_tokens": 418858705.0, + "step": 16470 + }, + { + "epoch": 1.8088073797496156, + "grad_norm": 1.9121884107589722, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7111914157867432, + "num_tokens": 418891696.0, + "step": 16471 + }, + { + "epoch": 1.8089171974522293, + "grad_norm": 2.3195669651031494, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7008058428764343, + "num_tokens": 418916138.0, + "step": 16472 + }, + { + "epoch": 1.809027015154843, + "grad_norm": 2.3633172512054443, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7231346368789673, + "num_tokens": 418940893.0, + "step": 16473 + }, + { + "epoch": 1.8091368328574566, + "grad_norm": 2.394848346710205, + "learning_rate": 1e-06, + "loss": 0.9926, + "mean_token_accuracy": 0.6969573497772217, + "num_tokens": 418965406.0, + "step": 16474 + }, + { + "epoch": 1.8092466505600702, + "grad_norm": 2.238168239593506, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7176170349121094, + "num_tokens": 418989530.0, + "step": 16475 + }, + { + "epoch": 1.809356468262684, + "grad_norm": 2.2543792724609375, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7050133347511292, + "num_tokens": 419013877.0, + "step": 16476 + }, + { + "epoch": 1.8094662859652977, + "grad_norm": 1.9020715951919556, + "learning_rate": 1e-06, + "loss": 1.005, + "mean_token_accuracy": 0.6899361610412598, + "num_tokens": 419047835.0, + "step": 16477 + }, + { + "epoch": 1.8095761036679112, + "grad_norm": 2.495720863342285, + "learning_rate": 1e-06, + "loss": 0.9187, + "mean_token_accuracy": 0.7205866575241089, + "num_tokens": 419069727.0, + "step": 16478 + }, + { + "epoch": 1.809685921370525, + "grad_norm": 2.209131956100464, + "learning_rate": 1e-06, + "loss": 0.8902, + "mean_token_accuracy": 0.7229050397872925, + "num_tokens": 419093735.0, + "step": 16479 + }, + { + "epoch": 1.8097957390731385, + "grad_norm": 2.4640681743621826, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7392063140869141, + "num_tokens": 419116298.0, + "step": 16480 + }, + { + "epoch": 1.8099055567757523, + "grad_norm": 2.5160136222839355, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.720653772354126, + "num_tokens": 419139138.0, + "step": 16481 + }, + { + "epoch": 1.810015374478366, + "grad_norm": 2.5567378997802734, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7231299877166748, + "num_tokens": 419159612.0, + "step": 16482 + }, + { + "epoch": 1.8101251921809796, + "grad_norm": 2.0793726444244385, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7300251722335815, + "num_tokens": 419189142.0, + "step": 16483 + }, + { + "epoch": 1.810235009883593, + "grad_norm": 2.4403932094573975, + "learning_rate": 1e-06, + "loss": 0.8788, + "mean_token_accuracy": 0.7307813763618469, + "num_tokens": 419212095.0, + "step": 16484 + }, + { + "epoch": 1.8103448275862069, + "grad_norm": 2.2539496421813965, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7162614464759827, + "num_tokens": 419237291.0, + "step": 16485 + }, + { + "epoch": 1.8104546452888206, + "grad_norm": 2.400268316268921, + "learning_rate": 1e-06, + "loss": 0.8325, + "mean_token_accuracy": 0.7336243987083435, + "num_tokens": 419260307.0, + "step": 16486 + }, + { + "epoch": 1.8105644629914344, + "grad_norm": 2.295344591140747, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7100263833999634, + "num_tokens": 419284931.0, + "step": 16487 + }, + { + "epoch": 1.810674280694048, + "grad_norm": 2.5439705848693848, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7245057821273804, + "num_tokens": 419305307.0, + "step": 16488 + }, + { + "epoch": 1.8107840983966614, + "grad_norm": 1.893723487854004, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7166840434074402, + "num_tokens": 419340173.0, + "step": 16489 + }, + { + "epoch": 1.8108939160992752, + "grad_norm": 2.108769655227661, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.710729718208313, + "num_tokens": 419368415.0, + "step": 16490 + }, + { + "epoch": 1.811003733801889, + "grad_norm": 2.000880241394043, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7164555788040161, + "num_tokens": 419396821.0, + "step": 16491 + }, + { + "epoch": 1.8111135515045025, + "grad_norm": 2.290818929672241, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7120414972305298, + "num_tokens": 419420998.0, + "step": 16492 + }, + { + "epoch": 1.8112233692071162, + "grad_norm": 2.219540596008301, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7244181632995605, + "num_tokens": 419447005.0, + "step": 16493 + }, + { + "epoch": 1.8113331869097298, + "grad_norm": 2.821028709411621, + "learning_rate": 1e-06, + "loss": 0.8532, + "mean_token_accuracy": 0.7319316864013672, + "num_tokens": 419465181.0, + "step": 16494 + }, + { + "epoch": 1.8114430046123435, + "grad_norm": 2.4057555198669434, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7113785743713379, + "num_tokens": 419488433.0, + "step": 16495 + }, + { + "epoch": 1.8115528223149573, + "grad_norm": 2.051325559616089, + "learning_rate": 1e-06, + "loss": 0.8717, + "mean_token_accuracy": 0.7311833500862122, + "num_tokens": 419518535.0, + "step": 16496 + }, + { + "epoch": 1.8116626400175708, + "grad_norm": 2.5318942070007324, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7215720415115356, + "num_tokens": 419538711.0, + "step": 16497 + }, + { + "epoch": 1.8117724577201844, + "grad_norm": 2.4500088691711426, + "learning_rate": 1e-06, + "loss": 0.8548, + "mean_token_accuracy": 0.7367861866950989, + "num_tokens": 419559886.0, + "step": 16498 + }, + { + "epoch": 1.8118822754227981, + "grad_norm": 1.9956830739974976, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.721818208694458, + "num_tokens": 419590710.0, + "step": 16499 + }, + { + "epoch": 1.8119920931254119, + "grad_norm": 2.231419086456299, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7285928726196289, + "num_tokens": 419615281.0, + "step": 16500 + }, + { + "epoch": 1.8121019108280256, + "grad_norm": 2.0932552814483643, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7178923487663269, + "num_tokens": 419642800.0, + "step": 16501 + }, + { + "epoch": 1.8122117285306392, + "grad_norm": 2.3847973346710205, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7181022763252258, + "num_tokens": 419667192.0, + "step": 16502 + }, + { + "epoch": 1.8123215462332527, + "grad_norm": 1.8910139799118042, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.7067499160766602, + "num_tokens": 419701999.0, + "step": 16503 + }, + { + "epoch": 1.8124313639358665, + "grad_norm": 2.402184009552002, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7255858778953552, + "num_tokens": 419725575.0, + "step": 16504 + }, + { + "epoch": 1.8125411816384802, + "grad_norm": 2.1915009021759033, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7004851698875427, + "num_tokens": 419755776.0, + "step": 16505 + }, + { + "epoch": 1.8126509993410937, + "grad_norm": 2.4573862552642822, + "learning_rate": 1e-06, + "loss": 0.9587, + "mean_token_accuracy": 0.7036210298538208, + "num_tokens": 419780635.0, + "step": 16506 + }, + { + "epoch": 1.8127608170437073, + "grad_norm": 2.3218417167663574, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7050700187683105, + "num_tokens": 419805685.0, + "step": 16507 + }, + { + "epoch": 1.812870634746321, + "grad_norm": 2.664236307144165, + "learning_rate": 1e-06, + "loss": 0.8852, + "mean_token_accuracy": 0.719397783279419, + "num_tokens": 419825851.0, + "step": 16508 + }, + { + "epoch": 1.8129804524489348, + "grad_norm": 2.1490862369537354, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.7144573926925659, + "num_tokens": 419852895.0, + "step": 16509 + }, + { + "epoch": 1.8130902701515486, + "grad_norm": 1.9439152479171753, + "learning_rate": 1e-06, + "loss": 0.9095, + "mean_token_accuracy": 0.7190546989440918, + "num_tokens": 419887466.0, + "step": 16510 + }, + { + "epoch": 1.813200087854162, + "grad_norm": 2.480388879776001, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7191659212112427, + "num_tokens": 419909879.0, + "step": 16511 + }, + { + "epoch": 1.8133099055567756, + "grad_norm": 2.4240214824676514, + "learning_rate": 1e-06, + "loss": 0.7186, + "mean_token_accuracy": 0.7731118202209473, + "num_tokens": 419930207.0, + "step": 16512 + }, + { + "epoch": 1.8134197232593894, + "grad_norm": 2.3511531352996826, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7149407863616943, + "num_tokens": 419954517.0, + "step": 16513 + }, + { + "epoch": 1.8135295409620031, + "grad_norm": 2.258359909057617, + "learning_rate": 1e-06, + "loss": 0.9139, + "mean_token_accuracy": 0.7121946811676025, + "num_tokens": 419980251.0, + "step": 16514 + }, + { + "epoch": 1.813639358664617, + "grad_norm": 2.0610530376434326, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7332823276519775, + "num_tokens": 420011450.0, + "step": 16515 + }, + { + "epoch": 1.8137491763672304, + "grad_norm": 2.289567232131958, + "learning_rate": 1e-06, + "loss": 0.9045, + "mean_token_accuracy": 0.7223007082939148, + "num_tokens": 420035445.0, + "step": 16516 + }, + { + "epoch": 1.813858994069844, + "grad_norm": 2.6544461250305176, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7230116724967957, + "num_tokens": 420055555.0, + "step": 16517 + }, + { + "epoch": 1.8139688117724577, + "grad_norm": 2.1383633613586426, + "learning_rate": 1e-06, + "loss": 0.868, + "mean_token_accuracy": 0.728966236114502, + "num_tokens": 420081527.0, + "step": 16518 + }, + { + "epoch": 1.8140786294750715, + "grad_norm": 2.196197032928467, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7104202508926392, + "num_tokens": 420108083.0, + "step": 16519 + }, + { + "epoch": 1.814188447177685, + "grad_norm": 2.3702757358551025, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7286165356636047, + "num_tokens": 420131270.0, + "step": 16520 + }, + { + "epoch": 1.8142982648802986, + "grad_norm": 2.221341848373413, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.7441195249557495, + "num_tokens": 420154831.0, + "step": 16521 + }, + { + "epoch": 1.8144080825829123, + "grad_norm": 2.103957176208496, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.715345025062561, + "num_tokens": 420183229.0, + "step": 16522 + }, + { + "epoch": 1.814517900285526, + "grad_norm": 2.3994853496551514, + "learning_rate": 1e-06, + "loss": 0.8095, + "mean_token_accuracy": 0.7414199709892273, + "num_tokens": 420204673.0, + "step": 16523 + }, + { + "epoch": 1.8146277179881398, + "grad_norm": 2.5958306789398193, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7228527665138245, + "num_tokens": 420225896.0, + "step": 16524 + }, + { + "epoch": 1.8147375356907534, + "grad_norm": 2.495783805847168, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7279442548751831, + "num_tokens": 420245705.0, + "step": 16525 + }, + { + "epoch": 1.814847353393367, + "grad_norm": 2.5836246013641357, + "learning_rate": 1e-06, + "loss": 0.8653, + "mean_token_accuracy": 0.7282989621162415, + "num_tokens": 420266531.0, + "step": 16526 + }, + { + "epoch": 1.8149571710959806, + "grad_norm": 2.2226903438568115, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7024157047271729, + "num_tokens": 420294577.0, + "step": 16527 + }, + { + "epoch": 1.8150669887985944, + "grad_norm": 2.55194354057312, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.7211182117462158, + "num_tokens": 420314853.0, + "step": 16528 + }, + { + "epoch": 1.815176806501208, + "grad_norm": 2.0938353538513184, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7141529321670532, + "num_tokens": 420343177.0, + "step": 16529 + }, + { + "epoch": 1.8152866242038217, + "grad_norm": 2.240522623062134, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7328125238418579, + "num_tokens": 420369347.0, + "step": 16530 + }, + { + "epoch": 1.8153964419064352, + "grad_norm": 2.557819366455078, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7103561162948608, + "num_tokens": 420390969.0, + "step": 16531 + }, + { + "epoch": 1.815506259609049, + "grad_norm": 1.964105248451233, + "learning_rate": 1e-06, + "loss": 0.9575, + "mean_token_accuracy": 0.7110764980316162, + "num_tokens": 420422337.0, + "step": 16532 + }, + { + "epoch": 1.8156160773116627, + "grad_norm": 2.220086097717285, + "learning_rate": 1e-06, + "loss": 0.8924, + "mean_token_accuracy": 0.7228119373321533, + "num_tokens": 420447471.0, + "step": 16533 + }, + { + "epoch": 1.8157258950142763, + "grad_norm": 2.056241989135742, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7069116830825806, + "num_tokens": 420476042.0, + "step": 16534 + }, + { + "epoch": 1.8158357127168898, + "grad_norm": 2.122819423675537, + "learning_rate": 1e-06, + "loss": 0.9088, + "mean_token_accuracy": 0.7206154465675354, + "num_tokens": 420503912.0, + "step": 16535 + }, + { + "epoch": 1.8159455304195036, + "grad_norm": 2.3769264221191406, + "learning_rate": 1e-06, + "loss": 1.0173, + "mean_token_accuracy": 0.6862994432449341, + "num_tokens": 420528625.0, + "step": 16536 + }, + { + "epoch": 1.8160553481221173, + "grad_norm": 2.1683027744293213, + "learning_rate": 1e-06, + "loss": 0.9393, + "mean_token_accuracy": 0.7095677256584167, + "num_tokens": 420558090.0, + "step": 16537 + }, + { + "epoch": 1.816165165824731, + "grad_norm": 2.3143601417541504, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7192956805229187, + "num_tokens": 420582259.0, + "step": 16538 + }, + { + "epoch": 1.8162749835273446, + "grad_norm": 2.093245267868042, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.71280437707901, + "num_tokens": 420612373.0, + "step": 16539 + }, + { + "epoch": 1.8163848012299582, + "grad_norm": 2.421085834503174, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7278845310211182, + "num_tokens": 420634050.0, + "step": 16540 + }, + { + "epoch": 1.816494618932572, + "grad_norm": 2.0372135639190674, + "learning_rate": 1e-06, + "loss": 0.9158, + "mean_token_accuracy": 0.71595299243927, + "num_tokens": 420667502.0, + "step": 16541 + }, + { + "epoch": 1.8166044366351857, + "grad_norm": 2.4851040840148926, + "learning_rate": 1e-06, + "loss": 0.8322, + "mean_token_accuracy": 0.732129693031311, + "num_tokens": 420688712.0, + "step": 16542 + }, + { + "epoch": 1.8167142543377992, + "grad_norm": 2.318012237548828, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7194907069206238, + "num_tokens": 420714125.0, + "step": 16543 + }, + { + "epoch": 1.816824072040413, + "grad_norm": 1.9271745681762695, + "learning_rate": 1e-06, + "loss": 0.9936, + "mean_token_accuracy": 0.6940785050392151, + "num_tokens": 420748413.0, + "step": 16544 + }, + { + "epoch": 1.8169338897430265, + "grad_norm": 2.0542213916778564, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.704602062702179, + "num_tokens": 420779282.0, + "step": 16545 + }, + { + "epoch": 1.8170437074456403, + "grad_norm": 2.487335443496704, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.728894829750061, + "num_tokens": 420800504.0, + "step": 16546 + }, + { + "epoch": 1.817153525148254, + "grad_norm": 2.0867106914520264, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7156091928482056, + "num_tokens": 420828347.0, + "step": 16547 + }, + { + "epoch": 1.8172633428508675, + "grad_norm": 2.4937326908111572, + "learning_rate": 1e-06, + "loss": 0.9402, + "mean_token_accuracy": 0.7051270008087158, + "num_tokens": 420851491.0, + "step": 16548 + }, + { + "epoch": 1.817373160553481, + "grad_norm": 2.2806191444396973, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7155579924583435, + "num_tokens": 420875236.0, + "step": 16549 + }, + { + "epoch": 1.8174829782560948, + "grad_norm": 2.495562791824341, + "learning_rate": 1e-06, + "loss": 0.8757, + "mean_token_accuracy": 0.7253895998001099, + "num_tokens": 420897576.0, + "step": 16550 + }, + { + "epoch": 1.8175927959587086, + "grad_norm": 2.0494275093078613, + "learning_rate": 1e-06, + "loss": 0.9594, + "mean_token_accuracy": 0.7092921137809753, + "num_tokens": 420927906.0, + "step": 16551 + }, + { + "epoch": 1.8177026136613224, + "grad_norm": 2.390472173690796, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7284533977508545, + "num_tokens": 420951189.0, + "step": 16552 + }, + { + "epoch": 1.8178124313639359, + "grad_norm": 1.996235728263855, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7309346199035645, + "num_tokens": 420981646.0, + "step": 16553 + }, + { + "epoch": 1.8179222490665494, + "grad_norm": 2.2072887420654297, + "learning_rate": 1e-06, + "loss": 0.7917, + "mean_token_accuracy": 0.7537915110588074, + "num_tokens": 421004744.0, + "step": 16554 + }, + { + "epoch": 1.8180320667691632, + "grad_norm": 2.5181453227996826, + "learning_rate": 1e-06, + "loss": 0.9473, + "mean_token_accuracy": 0.7013837695121765, + "num_tokens": 421030001.0, + "step": 16555 + }, + { + "epoch": 1.818141884471777, + "grad_norm": 2.3126907348632812, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.7069613933563232, + "num_tokens": 421055017.0, + "step": 16556 + }, + { + "epoch": 1.8182517021743905, + "grad_norm": 2.463782787322998, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7274152040481567, + "num_tokens": 421075082.0, + "step": 16557 + }, + { + "epoch": 1.818361519877004, + "grad_norm": 2.16182541847229, + "learning_rate": 1e-06, + "loss": 0.9901, + "mean_token_accuracy": 0.6922080516815186, + "num_tokens": 421103675.0, + "step": 16558 + }, + { + "epoch": 1.8184713375796178, + "grad_norm": 2.2391374111175537, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7279252409934998, + "num_tokens": 421129234.0, + "step": 16559 + }, + { + "epoch": 1.8185811552822315, + "grad_norm": 2.267547607421875, + "learning_rate": 1e-06, + "loss": 0.9309, + "mean_token_accuracy": 0.7057973146438599, + "num_tokens": 421155506.0, + "step": 16560 + }, + { + "epoch": 1.8186909729848453, + "grad_norm": 2.257888078689575, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7216760516166687, + "num_tokens": 421178889.0, + "step": 16561 + }, + { + "epoch": 1.8188007906874588, + "grad_norm": 2.355311393737793, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7138181328773499, + "num_tokens": 421202263.0, + "step": 16562 + }, + { + "epoch": 1.8189106083900723, + "grad_norm": 2.542492151260376, + "learning_rate": 1e-06, + "loss": 0.9397, + "mean_token_accuracy": 0.7083078622817993, + "num_tokens": 421225126.0, + "step": 16563 + }, + { + "epoch": 1.819020426092686, + "grad_norm": 2.066511392593384, + "learning_rate": 1e-06, + "loss": 0.951, + "mean_token_accuracy": 0.714767575263977, + "num_tokens": 421254537.0, + "step": 16564 + }, + { + "epoch": 1.8191302437952999, + "grad_norm": 2.269887924194336, + "learning_rate": 1e-06, + "loss": 0.8897, + "mean_token_accuracy": 0.7262856960296631, + "num_tokens": 421278013.0, + "step": 16565 + }, + { + "epoch": 1.8192400614979136, + "grad_norm": 2.484483003616333, + "learning_rate": 1e-06, + "loss": 0.7973, + "mean_token_accuracy": 0.7482627630233765, + "num_tokens": 421299530.0, + "step": 16566 + }, + { + "epoch": 1.8193498792005272, + "grad_norm": 2.053860664367676, + "learning_rate": 1e-06, + "loss": 0.8559, + "mean_token_accuracy": 0.7271257638931274, + "num_tokens": 421328791.0, + "step": 16567 + }, + { + "epoch": 1.8194596969031407, + "grad_norm": 2.3151228427886963, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7213952541351318, + "num_tokens": 421350925.0, + "step": 16568 + }, + { + "epoch": 1.8195695146057544, + "grad_norm": 2.267341136932373, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7226206064224243, + "num_tokens": 421374164.0, + "step": 16569 + }, + { + "epoch": 1.8196793323083682, + "grad_norm": 2.545342206954956, + "learning_rate": 1e-06, + "loss": 0.955, + "mean_token_accuracy": 0.7039206027984619, + "num_tokens": 421395548.0, + "step": 16570 + }, + { + "epoch": 1.8197891500109817, + "grad_norm": 2.8978404998779297, + "learning_rate": 1e-06, + "loss": 0.9198, + "mean_token_accuracy": 0.7167679071426392, + "num_tokens": 421412620.0, + "step": 16571 + }, + { + "epoch": 1.8198989677135953, + "grad_norm": 2.306011915206909, + "learning_rate": 1e-06, + "loss": 0.9879, + "mean_token_accuracy": 0.693963348865509, + "num_tokens": 421437802.0, + "step": 16572 + }, + { + "epoch": 1.820008785416209, + "grad_norm": 2.4540798664093018, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7321540713310242, + "num_tokens": 421459455.0, + "step": 16573 + }, + { + "epoch": 1.8201186031188228, + "grad_norm": 2.2329628467559814, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7085132598876953, + "num_tokens": 421485691.0, + "step": 16574 + }, + { + "epoch": 1.8202284208214365, + "grad_norm": 2.155463695526123, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.72742760181427, + "num_tokens": 421511143.0, + "step": 16575 + }, + { + "epoch": 1.82033823852405, + "grad_norm": 2.5243067741394043, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7157816290855408, + "num_tokens": 421531146.0, + "step": 16576 + }, + { + "epoch": 1.8204480562266636, + "grad_norm": 2.400421619415283, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.7205474376678467, + "num_tokens": 421556196.0, + "step": 16577 + }, + { + "epoch": 1.8205578739292774, + "grad_norm": 2.41049861907959, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.7033172845840454, + "num_tokens": 421579195.0, + "step": 16578 + }, + { + "epoch": 1.8206676916318911, + "grad_norm": 2.074099540710449, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.701749861240387, + "num_tokens": 421607998.0, + "step": 16579 + }, + { + "epoch": 1.8207775093345049, + "grad_norm": 2.308901071548462, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7332558631896973, + "num_tokens": 421631146.0, + "step": 16580 + }, + { + "epoch": 1.8208873270371184, + "grad_norm": 2.5247554779052734, + "learning_rate": 1e-06, + "loss": 0.7778, + "mean_token_accuracy": 0.7486259937286377, + "num_tokens": 421650956.0, + "step": 16581 + }, + { + "epoch": 1.820997144739732, + "grad_norm": 2.572362184524536, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7286763191223145, + "num_tokens": 421671314.0, + "step": 16582 + }, + { + "epoch": 1.8211069624423457, + "grad_norm": 2.151697874069214, + "learning_rate": 1e-06, + "loss": 0.8406, + "mean_token_accuracy": 0.7343155145645142, + "num_tokens": 421699084.0, + "step": 16583 + }, + { + "epoch": 1.8212167801449595, + "grad_norm": 2.0707621574401855, + "learning_rate": 1e-06, + "loss": 1.0649, + "mean_token_accuracy": 0.6823205351829529, + "num_tokens": 421728629.0, + "step": 16584 + }, + { + "epoch": 1.821326597847573, + "grad_norm": 2.453239679336548, + "learning_rate": 1e-06, + "loss": 0.7622, + "mean_token_accuracy": 0.752250075340271, + "num_tokens": 421748722.0, + "step": 16585 + }, + { + "epoch": 1.8214364155501865, + "grad_norm": 2.1443989276885986, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.718122124671936, + "num_tokens": 421777373.0, + "step": 16586 + }, + { + "epoch": 1.8215462332528003, + "grad_norm": 2.2818546295166016, + "learning_rate": 1e-06, + "loss": 0.8146, + "mean_token_accuracy": 0.7389734387397766, + "num_tokens": 421802000.0, + "step": 16587 + }, + { + "epoch": 1.821656050955414, + "grad_norm": 1.9314372539520264, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7076756954193115, + "num_tokens": 421836453.0, + "step": 16588 + }, + { + "epoch": 1.8217658686580278, + "grad_norm": 2.0606932640075684, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7189385890960693, + "num_tokens": 421865254.0, + "step": 16589 + }, + { + "epoch": 1.8218756863606413, + "grad_norm": 2.2310333251953125, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7263192534446716, + "num_tokens": 421889683.0, + "step": 16590 + }, + { + "epoch": 1.8219855040632549, + "grad_norm": 1.980394959449768, + "learning_rate": 1e-06, + "loss": 0.9401, + "mean_token_accuracy": 0.7086384892463684, + "num_tokens": 421918752.0, + "step": 16591 + }, + { + "epoch": 1.8220953217658686, + "grad_norm": 2.192908763885498, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7204840183258057, + "num_tokens": 421945227.0, + "step": 16592 + }, + { + "epoch": 1.8222051394684824, + "grad_norm": 1.927281141281128, + "learning_rate": 1e-06, + "loss": 0.9348, + "mean_token_accuracy": 0.7194445133209229, + "num_tokens": 421978169.0, + "step": 16593 + }, + { + "epoch": 1.822314957171096, + "grad_norm": 2.0320687294006348, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7147622108459473, + "num_tokens": 422009588.0, + "step": 16594 + }, + { + "epoch": 1.8224247748737097, + "grad_norm": 2.371222734451294, + "learning_rate": 1e-06, + "loss": 0.8442, + "mean_token_accuracy": 0.7302857637405396, + "num_tokens": 422032855.0, + "step": 16595 + }, + { + "epoch": 1.8225345925763232, + "grad_norm": 2.2315258979797363, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.7168796062469482, + "num_tokens": 422059128.0, + "step": 16596 + }, + { + "epoch": 1.822644410278937, + "grad_norm": 2.1997287273406982, + "learning_rate": 1e-06, + "loss": 1.0948, + "mean_token_accuracy": 0.6844965815544128, + "num_tokens": 422087924.0, + "step": 16597 + }, + { + "epoch": 1.8227542279815507, + "grad_norm": 2.210702419281006, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7123639583587646, + "num_tokens": 422113834.0, + "step": 16598 + }, + { + "epoch": 1.8228640456841643, + "grad_norm": 2.075789451599121, + "learning_rate": 1e-06, + "loss": 0.9222, + "mean_token_accuracy": 0.7235662937164307, + "num_tokens": 422143361.0, + "step": 16599 + }, + { + "epoch": 1.8229738633867778, + "grad_norm": 2.197317123413086, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7155377864837646, + "num_tokens": 422168492.0, + "step": 16600 + }, + { + "epoch": 1.8230836810893916, + "grad_norm": 2.141545534133911, + "learning_rate": 1e-06, + "loss": 0.7943, + "mean_token_accuracy": 0.752705991268158, + "num_tokens": 422196451.0, + "step": 16601 + }, + { + "epoch": 1.8231934987920053, + "grad_norm": 2.617088556289673, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.7140522599220276, + "num_tokens": 422217091.0, + "step": 16602 + }, + { + "epoch": 1.823303316494619, + "grad_norm": 2.0928385257720947, + "learning_rate": 1e-06, + "loss": 0.8956, + "mean_token_accuracy": 0.7258934378623962, + "num_tokens": 422244659.0, + "step": 16603 + }, + { + "epoch": 1.8234131341972326, + "grad_norm": 2.179961681365967, + "learning_rate": 1e-06, + "loss": 0.8887, + "mean_token_accuracy": 0.7254266142845154, + "num_tokens": 422270155.0, + "step": 16604 + }, + { + "epoch": 1.8235229518998461, + "grad_norm": 2.1301698684692383, + "learning_rate": 1e-06, + "loss": 0.9495, + "mean_token_accuracy": 0.7059483528137207, + "num_tokens": 422297567.0, + "step": 16605 + }, + { + "epoch": 1.82363276960246, + "grad_norm": 2.2333173751831055, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7284449338912964, + "num_tokens": 422322126.0, + "step": 16606 + }, + { + "epoch": 1.8237425873050737, + "grad_norm": 2.3800644874572754, + "learning_rate": 1e-06, + "loss": 0.9234, + "mean_token_accuracy": 0.7113596200942993, + "num_tokens": 422345175.0, + "step": 16607 + }, + { + "epoch": 1.8238524050076872, + "grad_norm": 2.112192153930664, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7068756222724915, + "num_tokens": 422371778.0, + "step": 16608 + }, + { + "epoch": 1.823962222710301, + "grad_norm": 2.4711265563964844, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7152692079544067, + "num_tokens": 422393103.0, + "step": 16609 + }, + { + "epoch": 1.8240720404129145, + "grad_norm": 2.2320303916931152, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.720784068107605, + "num_tokens": 422418514.0, + "step": 16610 + }, + { + "epoch": 1.8241818581155282, + "grad_norm": 2.630568742752075, + "learning_rate": 1e-06, + "loss": 1.0029, + "mean_token_accuracy": 0.693381667137146, + "num_tokens": 422439294.0, + "step": 16611 + }, + { + "epoch": 1.824291675818142, + "grad_norm": 2.2689404487609863, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7279482483863831, + "num_tokens": 422463042.0, + "step": 16612 + }, + { + "epoch": 1.8244014935207555, + "grad_norm": 2.1418557167053223, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7265973091125488, + "num_tokens": 422491104.0, + "step": 16613 + }, + { + "epoch": 1.824511311223369, + "grad_norm": 2.1142444610595703, + "learning_rate": 1e-06, + "loss": 0.9003, + "mean_token_accuracy": 0.7202540636062622, + "num_tokens": 422518914.0, + "step": 16614 + }, + { + "epoch": 1.8246211289259828, + "grad_norm": 2.128527879714966, + "learning_rate": 1e-06, + "loss": 0.7766, + "mean_token_accuracy": 0.7451085448265076, + "num_tokens": 422545771.0, + "step": 16615 + }, + { + "epoch": 1.8247309466285966, + "grad_norm": 2.136286497116089, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.732668936252594, + "num_tokens": 422571125.0, + "step": 16616 + }, + { + "epoch": 1.8248407643312103, + "grad_norm": 2.177668809890747, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7175836563110352, + "num_tokens": 422598780.0, + "step": 16617 + }, + { + "epoch": 1.8249505820338239, + "grad_norm": 2.146036386489868, + "learning_rate": 1e-06, + "loss": 0.9875, + "mean_token_accuracy": 0.7029129266738892, + "num_tokens": 422630851.0, + "step": 16618 + }, + { + "epoch": 1.8250603997364374, + "grad_norm": 2.1076037883758545, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7166631817817688, + "num_tokens": 422658273.0, + "step": 16619 + }, + { + "epoch": 1.8251702174390512, + "grad_norm": 2.209502696990967, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7244457006454468, + "num_tokens": 422685751.0, + "step": 16620 + }, + { + "epoch": 1.825280035141665, + "grad_norm": 2.3227016925811768, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7101955413818359, + "num_tokens": 422709207.0, + "step": 16621 + }, + { + "epoch": 1.8253898528442785, + "grad_norm": 2.2034006118774414, + "learning_rate": 1e-06, + "loss": 0.8692, + "mean_token_accuracy": 0.7325113415718079, + "num_tokens": 422735012.0, + "step": 16622 + }, + { + "epoch": 1.825499670546892, + "grad_norm": 2.4782445430755615, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7172408103942871, + "num_tokens": 422758360.0, + "step": 16623 + }, + { + "epoch": 1.8256094882495058, + "grad_norm": 2.499781608581543, + "learning_rate": 1e-06, + "loss": 0.7418, + "mean_token_accuracy": 0.7596275806427002, + "num_tokens": 422777539.0, + "step": 16624 + }, + { + "epoch": 1.8257193059521195, + "grad_norm": 2.210116147994995, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7245457768440247, + "num_tokens": 422804256.0, + "step": 16625 + }, + { + "epoch": 1.8258291236547333, + "grad_norm": 2.3248300552368164, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.704567551612854, + "num_tokens": 422829969.0, + "step": 16626 + }, + { + "epoch": 1.8259389413573468, + "grad_norm": 2.145697832107544, + "learning_rate": 1e-06, + "loss": 0.9229, + "mean_token_accuracy": 0.7113979458808899, + "num_tokens": 422857732.0, + "step": 16627 + }, + { + "epoch": 1.8260487590599603, + "grad_norm": 2.16428279876709, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7325997948646545, + "num_tokens": 422884546.0, + "step": 16628 + }, + { + "epoch": 1.826158576762574, + "grad_norm": 2.151825428009033, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7105666399002075, + "num_tokens": 422911804.0, + "step": 16629 + }, + { + "epoch": 1.8262683944651878, + "grad_norm": 2.1869049072265625, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7155179977416992, + "num_tokens": 422936692.0, + "step": 16630 + }, + { + "epoch": 1.8263782121678016, + "grad_norm": 2.2068898677825928, + "learning_rate": 1e-06, + "loss": 0.9553, + "mean_token_accuracy": 0.7044826149940491, + "num_tokens": 422965931.0, + "step": 16631 + }, + { + "epoch": 1.8264880298704151, + "grad_norm": 2.3336164951324463, + "learning_rate": 1e-06, + "loss": 0.9387, + "mean_token_accuracy": 0.7056583166122437, + "num_tokens": 422990169.0, + "step": 16632 + }, + { + "epoch": 1.8265978475730287, + "grad_norm": 2.067551851272583, + "learning_rate": 1e-06, + "loss": 0.9078, + "mean_token_accuracy": 0.715991735458374, + "num_tokens": 423018080.0, + "step": 16633 + }, + { + "epoch": 1.8267076652756424, + "grad_norm": 2.0626280307769775, + "learning_rate": 1e-06, + "loss": 0.8846, + "mean_token_accuracy": 0.7217892408370972, + "num_tokens": 423048241.0, + "step": 16634 + }, + { + "epoch": 1.8268174829782562, + "grad_norm": 2.43595027923584, + "learning_rate": 1e-06, + "loss": 0.9288, + "mean_token_accuracy": 0.718032717704773, + "num_tokens": 423071910.0, + "step": 16635 + }, + { + "epoch": 1.8269273006808697, + "grad_norm": 2.5535881519317627, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7249428033828735, + "num_tokens": 423092534.0, + "step": 16636 + }, + { + "epoch": 1.8270371183834833, + "grad_norm": 2.296584367752075, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.7044262886047363, + "num_tokens": 423116801.0, + "step": 16637 + }, + { + "epoch": 1.827146936086097, + "grad_norm": 1.9752168655395508, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7111706137657166, + "num_tokens": 423147916.0, + "step": 16638 + }, + { + "epoch": 1.8272567537887108, + "grad_norm": 2.045382261276245, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7304842472076416, + "num_tokens": 423175439.0, + "step": 16639 + }, + { + "epoch": 1.8273665714913245, + "grad_norm": 2.004523754119873, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7063136696815491, + "num_tokens": 423206770.0, + "step": 16640 + }, + { + "epoch": 1.827476389193938, + "grad_norm": 2.5898988246917725, + "learning_rate": 1e-06, + "loss": 0.7865, + "mean_token_accuracy": 0.744606077671051, + "num_tokens": 423223875.0, + "step": 16641 + }, + { + "epoch": 1.8275862068965516, + "grad_norm": 2.262112855911255, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.7348976135253906, + "num_tokens": 423247989.0, + "step": 16642 + }, + { + "epoch": 1.8276960245991654, + "grad_norm": 2.3343005180358887, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.704340934753418, + "num_tokens": 423273563.0, + "step": 16643 + }, + { + "epoch": 1.8278058423017791, + "grad_norm": 1.9531561136245728, + "learning_rate": 1e-06, + "loss": 0.9891, + "mean_token_accuracy": 0.6903907656669617, + "num_tokens": 423307005.0, + "step": 16644 + }, + { + "epoch": 1.8279156600043929, + "grad_norm": 2.4607620239257812, + "learning_rate": 1e-06, + "loss": 0.7514, + "mean_token_accuracy": 0.7586638331413269, + "num_tokens": 423327911.0, + "step": 16645 + }, + { + "epoch": 1.8280254777070064, + "grad_norm": 2.1040658950805664, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7385846376419067, + "num_tokens": 423355049.0, + "step": 16646 + }, + { + "epoch": 1.82813529540962, + "grad_norm": 2.4190726280212402, + "learning_rate": 1e-06, + "loss": 0.8945, + "mean_token_accuracy": 0.7117702960968018, + "num_tokens": 423376943.0, + "step": 16647 + }, + { + "epoch": 1.8282451131122337, + "grad_norm": 2.318655490875244, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7087966203689575, + "num_tokens": 423397963.0, + "step": 16648 + }, + { + "epoch": 1.8283549308148475, + "grad_norm": 2.206366777420044, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7272647619247437, + "num_tokens": 423423212.0, + "step": 16649 + }, + { + "epoch": 1.828464748517461, + "grad_norm": 2.450383186340332, + "learning_rate": 1e-06, + "loss": 0.9718, + "mean_token_accuracy": 0.6991711854934692, + "num_tokens": 423445275.0, + "step": 16650 + }, + { + "epoch": 1.8285745662200745, + "grad_norm": 2.142414093017578, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7191801071166992, + "num_tokens": 423473355.0, + "step": 16651 + }, + { + "epoch": 1.8286843839226883, + "grad_norm": 2.0323445796966553, + "learning_rate": 1e-06, + "loss": 0.8976, + "mean_token_accuracy": 0.7191808223724365, + "num_tokens": 423501927.0, + "step": 16652 + }, + { + "epoch": 1.828794201625302, + "grad_norm": 2.187507390975952, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7336688041687012, + "num_tokens": 423527200.0, + "step": 16653 + }, + { + "epoch": 1.8289040193279158, + "grad_norm": 2.3951404094696045, + "learning_rate": 1e-06, + "loss": 0.9313, + "mean_token_accuracy": 0.713976263999939, + "num_tokens": 423551495.0, + "step": 16654 + }, + { + "epoch": 1.8290138370305293, + "grad_norm": 2.1409404277801514, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7209877371788025, + "num_tokens": 423580046.0, + "step": 16655 + }, + { + "epoch": 1.8291236547331429, + "grad_norm": 1.9712520837783813, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7132400274276733, + "num_tokens": 423610688.0, + "step": 16656 + }, + { + "epoch": 1.8292334724357566, + "grad_norm": 2.494483470916748, + "learning_rate": 1e-06, + "loss": 0.9019, + "mean_token_accuracy": 0.7203649878501892, + "num_tokens": 423630896.0, + "step": 16657 + }, + { + "epoch": 1.8293432901383704, + "grad_norm": 2.4205851554870605, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7044154405593872, + "num_tokens": 423656655.0, + "step": 16658 + }, + { + "epoch": 1.829453107840984, + "grad_norm": 2.856006622314453, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7207449078559875, + "num_tokens": 423674597.0, + "step": 16659 + }, + { + "epoch": 1.8295629255435977, + "grad_norm": 2.336127281188965, + "learning_rate": 1e-06, + "loss": 0.911, + "mean_token_accuracy": 0.72064608335495, + "num_tokens": 423702155.0, + "step": 16660 + }, + { + "epoch": 1.8296727432462112, + "grad_norm": 2.4954495429992676, + "learning_rate": 1e-06, + "loss": 0.9082, + "mean_token_accuracy": 0.7301862239837646, + "num_tokens": 423724666.0, + "step": 16661 + }, + { + "epoch": 1.829782560948825, + "grad_norm": 2.0620572566986084, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7163110375404358, + "num_tokens": 423752983.0, + "step": 16662 + }, + { + "epoch": 1.8298923786514387, + "grad_norm": 2.254220724105835, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.727840781211853, + "num_tokens": 423775896.0, + "step": 16663 + }, + { + "epoch": 1.8300021963540523, + "grad_norm": 2.3304286003112793, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7058308124542236, + "num_tokens": 423799132.0, + "step": 16664 + }, + { + "epoch": 1.8301120140566658, + "grad_norm": 2.5051381587982178, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.7164982557296753, + "num_tokens": 423822892.0, + "step": 16665 + }, + { + "epoch": 1.8302218317592795, + "grad_norm": 2.273742198944092, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7136842012405396, + "num_tokens": 423847908.0, + "step": 16666 + }, + { + "epoch": 1.8303316494618933, + "grad_norm": 2.0313055515289307, + "learning_rate": 1e-06, + "loss": 0.8339, + "mean_token_accuracy": 0.7422651648521423, + "num_tokens": 423878421.0, + "step": 16667 + }, + { + "epoch": 1.830441467164507, + "grad_norm": 2.262087821960449, + "learning_rate": 1e-06, + "loss": 0.7694, + "mean_token_accuracy": 0.7564741373062134, + "num_tokens": 423902388.0, + "step": 16668 + }, + { + "epoch": 1.8305512848671206, + "grad_norm": 2.222196102142334, + "learning_rate": 1e-06, + "loss": 0.8572, + "mean_token_accuracy": 0.7297008633613586, + "num_tokens": 423926487.0, + "step": 16669 + }, + { + "epoch": 1.8306611025697341, + "grad_norm": 2.518256902694702, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7258884906768799, + "num_tokens": 423946694.0, + "step": 16670 + }, + { + "epoch": 1.830770920272348, + "grad_norm": 2.012765884399414, + "learning_rate": 1e-06, + "loss": 0.8856, + "mean_token_accuracy": 0.7229477763175964, + "num_tokens": 423974984.0, + "step": 16671 + }, + { + "epoch": 1.8308807379749616, + "grad_norm": 2.125072956085205, + "learning_rate": 1e-06, + "loss": 0.894, + "mean_token_accuracy": 0.7195615768432617, + "num_tokens": 424005549.0, + "step": 16672 + }, + { + "epoch": 1.8309905556775752, + "grad_norm": 1.9685434103012085, + "learning_rate": 1e-06, + "loss": 0.9576, + "mean_token_accuracy": 0.6957279443740845, + "num_tokens": 424037020.0, + "step": 16673 + }, + { + "epoch": 1.831100373380189, + "grad_norm": 2.5778183937072754, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7302655577659607, + "num_tokens": 424058469.0, + "step": 16674 + }, + { + "epoch": 1.8312101910828025, + "grad_norm": 2.4531149864196777, + "learning_rate": 1e-06, + "loss": 0.777, + "mean_token_accuracy": 0.7479977011680603, + "num_tokens": 424081097.0, + "step": 16675 + }, + { + "epoch": 1.8313200087854162, + "grad_norm": 2.505523204803467, + "learning_rate": 1e-06, + "loss": 0.9457, + "mean_token_accuracy": 0.7104822993278503, + "num_tokens": 424105213.0, + "step": 16676 + }, + { + "epoch": 1.83142982648803, + "grad_norm": 2.320136070251465, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7240688800811768, + "num_tokens": 424130551.0, + "step": 16677 + }, + { + "epoch": 1.8315396441906435, + "grad_norm": 2.565516948699951, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.7109099626541138, + "num_tokens": 424152393.0, + "step": 16678 + }, + { + "epoch": 1.831649461893257, + "grad_norm": 2.359959840774536, + "learning_rate": 1e-06, + "loss": 0.9059, + "mean_token_accuracy": 0.7143306732177734, + "num_tokens": 424175949.0, + "step": 16679 + }, + { + "epoch": 1.8317592795958708, + "grad_norm": 2.052536725997925, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7331005334854126, + "num_tokens": 424205318.0, + "step": 16680 + }, + { + "epoch": 1.8318690972984846, + "grad_norm": 2.226130723953247, + "learning_rate": 1e-06, + "loss": 0.9195, + "mean_token_accuracy": 0.7136983871459961, + "num_tokens": 424230637.0, + "step": 16681 + }, + { + "epoch": 1.8319789150010983, + "grad_norm": 2.3626761436462402, + "learning_rate": 1e-06, + "loss": 0.7949, + "mean_token_accuracy": 0.7396529316902161, + "num_tokens": 424254363.0, + "step": 16682 + }, + { + "epoch": 1.8320887327037119, + "grad_norm": 2.29331636428833, + "learning_rate": 1e-06, + "loss": 0.8511, + "mean_token_accuracy": 0.7271643280982971, + "num_tokens": 424276117.0, + "step": 16683 + }, + { + "epoch": 1.8321985504063254, + "grad_norm": 2.1703548431396484, + "learning_rate": 1e-06, + "loss": 0.9952, + "mean_token_accuracy": 0.6948313117027283, + "num_tokens": 424304320.0, + "step": 16684 + }, + { + "epoch": 1.8323083681089392, + "grad_norm": 2.4303648471832275, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.724251389503479, + "num_tokens": 424326519.0, + "step": 16685 + }, + { + "epoch": 1.832418185811553, + "grad_norm": 2.3017501831054688, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7228434085845947, + "num_tokens": 424349561.0, + "step": 16686 + }, + { + "epoch": 1.8325280035141664, + "grad_norm": 2.0995092391967773, + "learning_rate": 1e-06, + "loss": 0.921, + "mean_token_accuracy": 0.7088781595230103, + "num_tokens": 424378001.0, + "step": 16687 + }, + { + "epoch": 1.83263782121678, + "grad_norm": 2.408478021621704, + "learning_rate": 1e-06, + "loss": 0.8212, + "mean_token_accuracy": 0.73903489112854, + "num_tokens": 424400033.0, + "step": 16688 + }, + { + "epoch": 1.8327476389193937, + "grad_norm": 2.3563108444213867, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7276545763015747, + "num_tokens": 424421851.0, + "step": 16689 + }, + { + "epoch": 1.8328574566220075, + "grad_norm": 2.3377554416656494, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.717902660369873, + "num_tokens": 424448288.0, + "step": 16690 + }, + { + "epoch": 1.8329672743246213, + "grad_norm": 2.2492988109588623, + "learning_rate": 1e-06, + "loss": 0.9597, + "mean_token_accuracy": 0.7043389678001404, + "num_tokens": 424473434.0, + "step": 16691 + }, + { + "epoch": 1.8330770920272348, + "grad_norm": 2.0345818996429443, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7125043272972107, + "num_tokens": 424501942.0, + "step": 16692 + }, + { + "epoch": 1.8331869097298483, + "grad_norm": 2.1277689933776855, + "learning_rate": 1e-06, + "loss": 0.8808, + "mean_token_accuracy": 0.7313426733016968, + "num_tokens": 424530173.0, + "step": 16693 + }, + { + "epoch": 1.833296727432462, + "grad_norm": 1.929675817489624, + "learning_rate": 1e-06, + "loss": 0.9403, + "mean_token_accuracy": 0.7089527249336243, + "num_tokens": 424563142.0, + "step": 16694 + }, + { + "epoch": 1.8334065451350758, + "grad_norm": 2.1125292778015137, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7322812080383301, + "num_tokens": 424589016.0, + "step": 16695 + }, + { + "epoch": 1.8335163628376896, + "grad_norm": 2.654966354370117, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7271016836166382, + "num_tokens": 424608594.0, + "step": 16696 + }, + { + "epoch": 1.8336261805403031, + "grad_norm": 2.0415735244750977, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7000826597213745, + "num_tokens": 424638982.0, + "step": 16697 + }, + { + "epoch": 1.8337359982429167, + "grad_norm": 2.419025421142578, + "learning_rate": 1e-06, + "loss": 0.81, + "mean_token_accuracy": 0.7422261834144592, + "num_tokens": 424661765.0, + "step": 16698 + }, + { + "epoch": 1.8338458159455304, + "grad_norm": 2.171970844268799, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7213217616081238, + "num_tokens": 424690932.0, + "step": 16699 + }, + { + "epoch": 1.8339556336481442, + "grad_norm": 2.279895544052124, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7261534929275513, + "num_tokens": 424717079.0, + "step": 16700 + }, + { + "epoch": 1.8340654513507577, + "grad_norm": 2.166782855987549, + "learning_rate": 1e-06, + "loss": 0.8929, + "mean_token_accuracy": 0.7181106805801392, + "num_tokens": 424742406.0, + "step": 16701 + }, + { + "epoch": 1.8341752690533712, + "grad_norm": 2.6411354541778564, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7222561240196228, + "num_tokens": 424763043.0, + "step": 16702 + }, + { + "epoch": 1.834285086755985, + "grad_norm": 2.258638858795166, + "learning_rate": 1e-06, + "loss": 0.8969, + "mean_token_accuracy": 0.7145934104919434, + "num_tokens": 424788822.0, + "step": 16703 + }, + { + "epoch": 1.8343949044585988, + "grad_norm": 2.062938690185547, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7276771664619446, + "num_tokens": 424816645.0, + "step": 16704 + }, + { + "epoch": 1.8345047221612125, + "grad_norm": 2.0172805786132812, + "learning_rate": 1e-06, + "loss": 0.9157, + "mean_token_accuracy": 0.7115931510925293, + "num_tokens": 424848450.0, + "step": 16705 + }, + { + "epoch": 1.834614539863826, + "grad_norm": 2.2129712104797363, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7208067178726196, + "num_tokens": 424871679.0, + "step": 16706 + }, + { + "epoch": 1.8347243575664396, + "grad_norm": 2.3807642459869385, + "learning_rate": 1e-06, + "loss": 0.8535, + "mean_token_accuracy": 0.7332295179367065, + "num_tokens": 424894475.0, + "step": 16707 + }, + { + "epoch": 1.8348341752690533, + "grad_norm": 2.2017650604248047, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7118432521820068, + "num_tokens": 424922501.0, + "step": 16708 + }, + { + "epoch": 1.834943992971667, + "grad_norm": 2.2119436264038086, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.7049775123596191, + "num_tokens": 424949098.0, + "step": 16709 + }, + { + "epoch": 1.8350538106742806, + "grad_norm": 2.2560007572174072, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7118855714797974, + "num_tokens": 424973991.0, + "step": 16710 + }, + { + "epoch": 1.8351636283768944, + "grad_norm": 2.4803638458251953, + "learning_rate": 1e-06, + "loss": 0.8311, + "mean_token_accuracy": 0.7367060780525208, + "num_tokens": 424994873.0, + "step": 16711 + }, + { + "epoch": 1.835273446079508, + "grad_norm": 2.40746808052063, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7245122790336609, + "num_tokens": 425018131.0, + "step": 16712 + }, + { + "epoch": 1.8353832637821217, + "grad_norm": 2.4088897705078125, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.722770094871521, + "num_tokens": 425041358.0, + "step": 16713 + }, + { + "epoch": 1.8354930814847354, + "grad_norm": 2.4754552841186523, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7163185477256775, + "num_tokens": 425063188.0, + "step": 16714 + }, + { + "epoch": 1.835602899187349, + "grad_norm": 2.0494883060455322, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.7228156328201294, + "num_tokens": 425091547.0, + "step": 16715 + }, + { + "epoch": 1.8357127168899625, + "grad_norm": 2.1881778240203857, + "learning_rate": 1e-06, + "loss": 0.925, + "mean_token_accuracy": 0.7156513929367065, + "num_tokens": 425117726.0, + "step": 16716 + }, + { + "epoch": 1.8358225345925763, + "grad_norm": 2.0722649097442627, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7222267389297485, + "num_tokens": 425147520.0, + "step": 16717 + }, + { + "epoch": 1.83593235229519, + "grad_norm": 2.165252923965454, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7329926490783691, + "num_tokens": 425175399.0, + "step": 16718 + }, + { + "epoch": 1.8360421699978038, + "grad_norm": 1.9937740564346313, + "learning_rate": 1e-06, + "loss": 0.9869, + "mean_token_accuracy": 0.7015218138694763, + "num_tokens": 425210577.0, + "step": 16719 + }, + { + "epoch": 1.8361519877004173, + "grad_norm": 2.2038300037384033, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7379475831985474, + "num_tokens": 425234711.0, + "step": 16720 + }, + { + "epoch": 1.8362618054030309, + "grad_norm": 2.214797019958496, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.73663330078125, + "num_tokens": 425260202.0, + "step": 16721 + }, + { + "epoch": 1.8363716231056446, + "grad_norm": 2.087404489517212, + "learning_rate": 1e-06, + "loss": 0.938, + "mean_token_accuracy": 0.7060776352882385, + "num_tokens": 425288669.0, + "step": 16722 + }, + { + "epoch": 1.8364814408082584, + "grad_norm": 2.1670022010803223, + "learning_rate": 1e-06, + "loss": 0.9735, + "mean_token_accuracy": 0.6988780498504639, + "num_tokens": 425315622.0, + "step": 16723 + }, + { + "epoch": 1.836591258510872, + "grad_norm": 2.1711161136627197, + "learning_rate": 1e-06, + "loss": 0.9084, + "mean_token_accuracy": 0.7145421504974365, + "num_tokens": 425345060.0, + "step": 16724 + }, + { + "epoch": 1.8367010762134857, + "grad_norm": 2.4961330890655518, + "learning_rate": 1e-06, + "loss": 0.8234, + "mean_token_accuracy": 0.7448020577430725, + "num_tokens": 425366942.0, + "step": 16725 + }, + { + "epoch": 1.8368108939160992, + "grad_norm": 2.644655466079712, + "learning_rate": 1e-06, + "loss": 0.8543, + "mean_token_accuracy": 0.7333358526229858, + "num_tokens": 425386429.0, + "step": 16726 + }, + { + "epoch": 1.836920711618713, + "grad_norm": 2.3133838176727295, + "learning_rate": 1e-06, + "loss": 0.8992, + "mean_token_accuracy": 0.7265020608901978, + "num_tokens": 425410127.0, + "step": 16727 + }, + { + "epoch": 1.8370305293213267, + "grad_norm": 2.5976085662841797, + "learning_rate": 1e-06, + "loss": 0.757, + "mean_token_accuracy": 0.7552709579467773, + "num_tokens": 425429599.0, + "step": 16728 + }, + { + "epoch": 1.8371403470239402, + "grad_norm": 2.3890297412872314, + "learning_rate": 1e-06, + "loss": 0.8305, + "mean_token_accuracy": 0.7409197688102722, + "num_tokens": 425453092.0, + "step": 16729 + }, + { + "epoch": 1.8372501647265538, + "grad_norm": 2.043764591217041, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7252585887908936, + "num_tokens": 425485014.0, + "step": 16730 + }, + { + "epoch": 1.8373599824291675, + "grad_norm": 2.5526509284973145, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7259401082992554, + "num_tokens": 425505996.0, + "step": 16731 + }, + { + "epoch": 1.8374698001317813, + "grad_norm": 2.1348612308502197, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.7111883163452148, + "num_tokens": 425533027.0, + "step": 16732 + }, + { + "epoch": 1.837579617834395, + "grad_norm": 2.192497968673706, + "learning_rate": 1e-06, + "loss": 0.8758, + "mean_token_accuracy": 0.7201558351516724, + "num_tokens": 425557838.0, + "step": 16733 + }, + { + "epoch": 1.8376894355370086, + "grad_norm": 2.5402798652648926, + "learning_rate": 1e-06, + "loss": 0.9345, + "mean_token_accuracy": 0.709505558013916, + "num_tokens": 425579747.0, + "step": 16734 + }, + { + "epoch": 1.8377992532396221, + "grad_norm": 2.2054052352905273, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7237619161605835, + "num_tokens": 425603650.0, + "step": 16735 + }, + { + "epoch": 1.8379090709422359, + "grad_norm": 2.283395767211914, + "learning_rate": 1e-06, + "loss": 0.8669, + "mean_token_accuracy": 0.7257583141326904, + "num_tokens": 425628541.0, + "step": 16736 + }, + { + "epoch": 1.8380188886448496, + "grad_norm": 2.0708096027374268, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7155978679656982, + "num_tokens": 425658472.0, + "step": 16737 + }, + { + "epoch": 1.8381287063474632, + "grad_norm": 2.7107722759246826, + "learning_rate": 1e-06, + "loss": 0.9233, + "mean_token_accuracy": 0.7301809787750244, + "num_tokens": 425678580.0, + "step": 16738 + }, + { + "epoch": 1.838238524050077, + "grad_norm": 2.0379250049591064, + "learning_rate": 1e-06, + "loss": 0.8766, + "mean_token_accuracy": 0.7199333310127258, + "num_tokens": 425708136.0, + "step": 16739 + }, + { + "epoch": 1.8383483417526905, + "grad_norm": 2.201756477355957, + "learning_rate": 1e-06, + "loss": 0.9453, + "mean_token_accuracy": 0.7129842042922974, + "num_tokens": 425735435.0, + "step": 16740 + }, + { + "epoch": 1.8384581594553042, + "grad_norm": 2.129706382751465, + "learning_rate": 1e-06, + "loss": 0.897, + "mean_token_accuracy": 0.72532719373703, + "num_tokens": 425763319.0, + "step": 16741 + }, + { + "epoch": 1.838567977157918, + "grad_norm": 2.353588581085205, + "learning_rate": 1e-06, + "loss": 0.8044, + "mean_token_accuracy": 0.7408095598220825, + "num_tokens": 425784584.0, + "step": 16742 + }, + { + "epoch": 1.8386777948605315, + "grad_norm": 2.45623517036438, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7054765224456787, + "num_tokens": 425807561.0, + "step": 16743 + }, + { + "epoch": 1.838787612563145, + "grad_norm": 2.443244457244873, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7251322269439697, + "num_tokens": 425829304.0, + "step": 16744 + }, + { + "epoch": 1.8388974302657588, + "grad_norm": 3.060425043106079, + "learning_rate": 1e-06, + "loss": 0.7649, + "mean_token_accuracy": 0.7551056146621704, + "num_tokens": 425844565.0, + "step": 16745 + }, + { + "epoch": 1.8390072479683726, + "grad_norm": 2.2808656692504883, + "learning_rate": 1e-06, + "loss": 0.9998, + "mean_token_accuracy": 0.6955916881561279, + "num_tokens": 425873234.0, + "step": 16746 + }, + { + "epoch": 1.8391170656709863, + "grad_norm": 2.155433177947998, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7258902788162231, + "num_tokens": 425899146.0, + "step": 16747 + }, + { + "epoch": 1.8392268833735999, + "grad_norm": 2.0895159244537354, + "learning_rate": 1e-06, + "loss": 0.8962, + "mean_token_accuracy": 0.7282928228378296, + "num_tokens": 425929580.0, + "step": 16748 + }, + { + "epoch": 1.8393367010762134, + "grad_norm": 2.2013206481933594, + "learning_rate": 1e-06, + "loss": 0.8502, + "mean_token_accuracy": 0.7295185327529907, + "num_tokens": 425955102.0, + "step": 16749 + }, + { + "epoch": 1.8394465187788271, + "grad_norm": 2.501115083694458, + "learning_rate": 1e-06, + "loss": 0.8393, + "mean_token_accuracy": 0.7379703521728516, + "num_tokens": 425974771.0, + "step": 16750 + }, + { + "epoch": 1.839556336481441, + "grad_norm": 2.091702461242676, + "learning_rate": 1e-06, + "loss": 0.9876, + "mean_token_accuracy": 0.6975836753845215, + "num_tokens": 426005352.0, + "step": 16751 + }, + { + "epoch": 1.8396661541840544, + "grad_norm": 2.132211685180664, + "learning_rate": 1e-06, + "loss": 0.9522, + "mean_token_accuracy": 0.7050647735595703, + "num_tokens": 426032784.0, + "step": 16752 + }, + { + "epoch": 1.839775971886668, + "grad_norm": 2.273460626602173, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.71335369348526, + "num_tokens": 426058663.0, + "step": 16753 + }, + { + "epoch": 1.8398857895892817, + "grad_norm": 2.2493960857391357, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7279888987541199, + "num_tokens": 426084722.0, + "step": 16754 + }, + { + "epoch": 1.8399956072918955, + "grad_norm": 2.6137611865997314, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7404270768165588, + "num_tokens": 426103740.0, + "step": 16755 + }, + { + "epoch": 1.8401054249945092, + "grad_norm": 2.3148252964019775, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7331953048706055, + "num_tokens": 426126362.0, + "step": 16756 + }, + { + "epoch": 1.8402152426971228, + "grad_norm": 2.08254075050354, + "learning_rate": 1e-06, + "loss": 0.8752, + "mean_token_accuracy": 0.725342869758606, + "num_tokens": 426152458.0, + "step": 16757 + }, + { + "epoch": 1.8403250603997363, + "grad_norm": 2.265429973602295, + "learning_rate": 1e-06, + "loss": 0.9382, + "mean_token_accuracy": 0.715071439743042, + "num_tokens": 426179632.0, + "step": 16758 + }, + { + "epoch": 1.84043487810235, + "grad_norm": 2.3462166786193848, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7208709716796875, + "num_tokens": 426203811.0, + "step": 16759 + }, + { + "epoch": 1.8405446958049638, + "grad_norm": 2.357717752456665, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7091076374053955, + "num_tokens": 426226970.0, + "step": 16760 + }, + { + "epoch": 1.8406545135075776, + "grad_norm": 1.92904531955719, + "learning_rate": 1e-06, + "loss": 0.9147, + "mean_token_accuracy": 0.7132945656776428, + "num_tokens": 426259835.0, + "step": 16761 + }, + { + "epoch": 1.8407643312101911, + "grad_norm": 2.264284610748291, + "learning_rate": 1e-06, + "loss": 0.9454, + "mean_token_accuracy": 0.7145562767982483, + "num_tokens": 426285606.0, + "step": 16762 + }, + { + "epoch": 1.8408741489128047, + "grad_norm": 2.1627936363220215, + "learning_rate": 1e-06, + "loss": 0.8888, + "mean_token_accuracy": 0.7210056185722351, + "num_tokens": 426312697.0, + "step": 16763 + }, + { + "epoch": 1.8409839666154184, + "grad_norm": 2.2344205379486084, + "learning_rate": 1e-06, + "loss": 0.9271, + "mean_token_accuracy": 0.7159309387207031, + "num_tokens": 426338993.0, + "step": 16764 + }, + { + "epoch": 1.8410937843180322, + "grad_norm": 2.5199718475341797, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7304695248603821, + "num_tokens": 426361121.0, + "step": 16765 + }, + { + "epoch": 1.8412036020206457, + "grad_norm": 2.1249492168426514, + "learning_rate": 1e-06, + "loss": 0.8672, + "mean_token_accuracy": 0.7227205634117126, + "num_tokens": 426387514.0, + "step": 16766 + }, + { + "epoch": 1.8413134197232592, + "grad_norm": 1.9700274467468262, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7324021458625793, + "num_tokens": 426419650.0, + "step": 16767 + }, + { + "epoch": 1.841423237425873, + "grad_norm": 2.407064914703369, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7253903746604919, + "num_tokens": 426440137.0, + "step": 16768 + }, + { + "epoch": 1.8415330551284868, + "grad_norm": 2.1124203205108643, + "learning_rate": 1e-06, + "loss": 0.9043, + "mean_token_accuracy": 0.7135927081108093, + "num_tokens": 426469591.0, + "step": 16769 + }, + { + "epoch": 1.8416428728311005, + "grad_norm": 2.066166400909424, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.709217369556427, + "num_tokens": 426501007.0, + "step": 16770 + }, + { + "epoch": 1.841752690533714, + "grad_norm": 2.2011406421661377, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7272650003433228, + "num_tokens": 426527357.0, + "step": 16771 + }, + { + "epoch": 1.8418625082363276, + "grad_norm": 2.4737701416015625, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7195200324058533, + "num_tokens": 426549321.0, + "step": 16772 + }, + { + "epoch": 1.8419723259389413, + "grad_norm": 2.392618179321289, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7072449922561646, + "num_tokens": 426575735.0, + "step": 16773 + }, + { + "epoch": 1.842082143641555, + "grad_norm": 2.080752372741699, + "learning_rate": 1e-06, + "loss": 0.7618, + "mean_token_accuracy": 0.7496068477630615, + "num_tokens": 426600771.0, + "step": 16774 + }, + { + "epoch": 1.8421919613441686, + "grad_norm": 1.9859445095062256, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.717465877532959, + "num_tokens": 426630947.0, + "step": 16775 + }, + { + "epoch": 1.8423017790467824, + "grad_norm": 2.4506938457489014, + "learning_rate": 1e-06, + "loss": 0.7456, + "mean_token_accuracy": 0.759246289730072, + "num_tokens": 426653608.0, + "step": 16776 + }, + { + "epoch": 1.842411596749396, + "grad_norm": 1.9664491415023804, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7271052002906799, + "num_tokens": 426684250.0, + "step": 16777 + }, + { + "epoch": 1.8425214144520097, + "grad_norm": 2.2040183544158936, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7316575050354004, + "num_tokens": 426709223.0, + "step": 16778 + }, + { + "epoch": 1.8426312321546234, + "grad_norm": 2.364772319793701, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7123332023620605, + "num_tokens": 426735053.0, + "step": 16779 + }, + { + "epoch": 1.842741049857237, + "grad_norm": 2.247497797012329, + "learning_rate": 1e-06, + "loss": 0.913, + "mean_token_accuracy": 0.7282489538192749, + "num_tokens": 426760508.0, + "step": 16780 + }, + { + "epoch": 1.8428508675598505, + "grad_norm": 2.0674026012420654, + "learning_rate": 1e-06, + "loss": 0.8562, + "mean_token_accuracy": 0.7289046049118042, + "num_tokens": 426788297.0, + "step": 16781 + }, + { + "epoch": 1.8429606852624643, + "grad_norm": 2.0855460166931152, + "learning_rate": 1e-06, + "loss": 0.9376, + "mean_token_accuracy": 0.703680157661438, + "num_tokens": 426815879.0, + "step": 16782 + }, + { + "epoch": 1.843070502965078, + "grad_norm": 2.2238762378692627, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7194637060165405, + "num_tokens": 426842337.0, + "step": 16783 + }, + { + "epoch": 1.8431803206676918, + "grad_norm": 2.423929452896118, + "learning_rate": 1e-06, + "loss": 0.8364, + "mean_token_accuracy": 0.735284149646759, + "num_tokens": 426864017.0, + "step": 16784 + }, + { + "epoch": 1.8432901383703053, + "grad_norm": 2.4466121196746826, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7302629351615906, + "num_tokens": 426887613.0, + "step": 16785 + }, + { + "epoch": 1.8433999560729188, + "grad_norm": 2.1357526779174805, + "learning_rate": 1e-06, + "loss": 0.887, + "mean_token_accuracy": 0.7222750186920166, + "num_tokens": 426914611.0, + "step": 16786 + }, + { + "epoch": 1.8435097737755326, + "grad_norm": 2.1446359157562256, + "learning_rate": 1e-06, + "loss": 0.958, + "mean_token_accuracy": 0.6979634761810303, + "num_tokens": 426942443.0, + "step": 16787 + }, + { + "epoch": 1.8436195914781464, + "grad_norm": 2.2582781314849854, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7303701639175415, + "num_tokens": 426967069.0, + "step": 16788 + }, + { + "epoch": 1.84372940918076, + "grad_norm": 2.2406225204467773, + "learning_rate": 1e-06, + "loss": 0.861, + "mean_token_accuracy": 0.7348908185958862, + "num_tokens": 426991774.0, + "step": 16789 + }, + { + "epoch": 1.8438392268833736, + "grad_norm": 2.3200433254241943, + "learning_rate": 1e-06, + "loss": 0.841, + "mean_token_accuracy": 0.7344548106193542, + "num_tokens": 427015876.0, + "step": 16790 + }, + { + "epoch": 1.8439490445859872, + "grad_norm": 2.337454080581665, + "learning_rate": 1e-06, + "loss": 0.8283, + "mean_token_accuracy": 0.7375922799110413, + "num_tokens": 427041399.0, + "step": 16791 + }, + { + "epoch": 1.844058862288601, + "grad_norm": 2.2655231952667236, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7274261713027954, + "num_tokens": 427064630.0, + "step": 16792 + }, + { + "epoch": 1.8441686799912147, + "grad_norm": 2.3933629989624023, + "learning_rate": 1e-06, + "loss": 0.9329, + "mean_token_accuracy": 0.710350751876831, + "num_tokens": 427088185.0, + "step": 16793 + }, + { + "epoch": 1.8442784976938282, + "grad_norm": 2.3852224349975586, + "learning_rate": 1e-06, + "loss": 0.8623, + "mean_token_accuracy": 0.7313498854637146, + "num_tokens": 427111671.0, + "step": 16794 + }, + { + "epoch": 1.8443883153964418, + "grad_norm": 2.0669941902160645, + "learning_rate": 1e-06, + "loss": 0.9316, + "mean_token_accuracy": 0.7078168988227844, + "num_tokens": 427141631.0, + "step": 16795 + }, + { + "epoch": 1.8444981330990555, + "grad_norm": 2.16835618019104, + "learning_rate": 1e-06, + "loss": 0.9577, + "mean_token_accuracy": 0.6996943950653076, + "num_tokens": 427168249.0, + "step": 16796 + }, + { + "epoch": 1.8446079508016693, + "grad_norm": 2.0108020305633545, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7134604454040527, + "num_tokens": 427197618.0, + "step": 16797 + }, + { + "epoch": 1.844717768504283, + "grad_norm": 2.1464719772338867, + "learning_rate": 1e-06, + "loss": 0.982, + "mean_token_accuracy": 0.7016451358795166, + "num_tokens": 427228139.0, + "step": 16798 + }, + { + "epoch": 1.8448275862068966, + "grad_norm": 2.4657607078552246, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7196177244186401, + "num_tokens": 427248881.0, + "step": 16799 + }, + { + "epoch": 1.84493740390951, + "grad_norm": 2.168577194213867, + "learning_rate": 1e-06, + "loss": 1.013, + "mean_token_accuracy": 0.6943548917770386, + "num_tokens": 427277324.0, + "step": 16800 + }, + { + "epoch": 1.8450472216121239, + "grad_norm": 2.468571186065674, + "learning_rate": 1e-06, + "loss": 0.9915, + "mean_token_accuracy": 0.6939507722854614, + "num_tokens": 427300038.0, + "step": 16801 + }, + { + "epoch": 1.8451570393147376, + "grad_norm": 2.432790994644165, + "learning_rate": 1e-06, + "loss": 0.9341, + "mean_token_accuracy": 0.7190145254135132, + "num_tokens": 427320909.0, + "step": 16802 + }, + { + "epoch": 1.8452668570173512, + "grad_norm": 2.444465398788452, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.716592013835907, + "num_tokens": 427343050.0, + "step": 16803 + }, + { + "epoch": 1.8453766747199647, + "grad_norm": 2.167635679244995, + "learning_rate": 1e-06, + "loss": 0.9235, + "mean_token_accuracy": 0.7238026261329651, + "num_tokens": 427369829.0, + "step": 16804 + }, + { + "epoch": 1.8454864924225785, + "grad_norm": 2.166110038757324, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7126779556274414, + "num_tokens": 427396043.0, + "step": 16805 + }, + { + "epoch": 1.8455963101251922, + "grad_norm": 2.0146126747131348, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7144606113433838, + "num_tokens": 427426544.0, + "step": 16806 + }, + { + "epoch": 1.845706127827806, + "grad_norm": 2.3793740272521973, + "learning_rate": 1e-06, + "loss": 0.8728, + "mean_token_accuracy": 0.7220041751861572, + "num_tokens": 427449549.0, + "step": 16807 + }, + { + "epoch": 1.8458159455304195, + "grad_norm": 2.287989854812622, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7320228219032288, + "num_tokens": 427475410.0, + "step": 16808 + }, + { + "epoch": 1.845925763233033, + "grad_norm": 2.0494906902313232, + "learning_rate": 1e-06, + "loss": 0.8843, + "mean_token_accuracy": 0.7252663969993591, + "num_tokens": 427503486.0, + "step": 16809 + }, + { + "epoch": 1.8460355809356468, + "grad_norm": 2.0824790000915527, + "learning_rate": 1e-06, + "loss": 0.9347, + "mean_token_accuracy": 0.7088624835014343, + "num_tokens": 427532563.0, + "step": 16810 + }, + { + "epoch": 1.8461453986382605, + "grad_norm": 2.7428643703460693, + "learning_rate": 1e-06, + "loss": 0.798, + "mean_token_accuracy": 0.7411676645278931, + "num_tokens": 427550832.0, + "step": 16811 + }, + { + "epoch": 1.8462552163408743, + "grad_norm": 2.10923433303833, + "learning_rate": 1e-06, + "loss": 0.9538, + "mean_token_accuracy": 0.7100495100021362, + "num_tokens": 427578160.0, + "step": 16812 + }, + { + "epoch": 1.8463650340434878, + "grad_norm": 2.3321266174316406, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7274906039237976, + "num_tokens": 427599037.0, + "step": 16813 + }, + { + "epoch": 1.8464748517461014, + "grad_norm": 2.3068063259124756, + "learning_rate": 1e-06, + "loss": 0.931, + "mean_token_accuracy": 0.7140408158302307, + "num_tokens": 427624159.0, + "step": 16814 + }, + { + "epoch": 1.8465846694487151, + "grad_norm": 2.2115540504455566, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7106039524078369, + "num_tokens": 427652438.0, + "step": 16815 + }, + { + "epoch": 1.846694487151329, + "grad_norm": 2.128464698791504, + "learning_rate": 1e-06, + "loss": 0.9052, + "mean_token_accuracy": 0.7275471687316895, + "num_tokens": 427678546.0, + "step": 16816 + }, + { + "epoch": 1.8468043048539424, + "grad_norm": 2.114532709121704, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7312626838684082, + "num_tokens": 427708232.0, + "step": 16817 + }, + { + "epoch": 1.846914122556556, + "grad_norm": 2.7699687480926514, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.729630708694458, + "num_tokens": 427725715.0, + "step": 16818 + }, + { + "epoch": 1.8470239402591697, + "grad_norm": 2.194671869277954, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7234644889831543, + "num_tokens": 427751258.0, + "step": 16819 + }, + { + "epoch": 1.8471337579617835, + "grad_norm": 2.5573911666870117, + "learning_rate": 1e-06, + "loss": 0.8357, + "mean_token_accuracy": 0.7312642931938171, + "num_tokens": 427772121.0, + "step": 16820 + }, + { + "epoch": 1.8472435756643972, + "grad_norm": 2.3312020301818848, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.70696622133255, + "num_tokens": 427797801.0, + "step": 16821 + }, + { + "epoch": 1.8473533933670108, + "grad_norm": 2.189532518386841, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7394143342971802, + "num_tokens": 427822209.0, + "step": 16822 + }, + { + "epoch": 1.8474632110696243, + "grad_norm": 1.9822317361831665, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.7227755188941956, + "num_tokens": 427851594.0, + "step": 16823 + }, + { + "epoch": 1.847573028772238, + "grad_norm": 2.1655235290527344, + "learning_rate": 1e-06, + "loss": 0.8059, + "mean_token_accuracy": 0.7340712547302246, + "num_tokens": 427877499.0, + "step": 16824 + }, + { + "epoch": 1.8476828464748518, + "grad_norm": 1.9652279615402222, + "learning_rate": 1e-06, + "loss": 0.9484, + "mean_token_accuracy": 0.7034623622894287, + "num_tokens": 427909448.0, + "step": 16825 + }, + { + "epoch": 1.8477926641774656, + "grad_norm": 2.06073260307312, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7384892702102661, + "num_tokens": 427938085.0, + "step": 16826 + }, + { + "epoch": 1.847902481880079, + "grad_norm": 2.1957547664642334, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7189902067184448, + "num_tokens": 427963314.0, + "step": 16827 + }, + { + "epoch": 1.8480122995826926, + "grad_norm": 2.5489563941955566, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7332457304000854, + "num_tokens": 427983059.0, + "step": 16828 + }, + { + "epoch": 1.8481221172853064, + "grad_norm": 2.094061851501465, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7334941029548645, + "num_tokens": 428009359.0, + "step": 16829 + }, + { + "epoch": 1.8482319349879202, + "grad_norm": 2.3271069526672363, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7380466461181641, + "num_tokens": 428032224.0, + "step": 16830 + }, + { + "epoch": 1.8483417526905337, + "grad_norm": 1.9714282751083374, + "learning_rate": 1e-06, + "loss": 0.9471, + "mean_token_accuracy": 0.7103521823883057, + "num_tokens": 428064263.0, + "step": 16831 + }, + { + "epoch": 1.8484515703931472, + "grad_norm": 2.2227580547332764, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7201237678527832, + "num_tokens": 428089310.0, + "step": 16832 + }, + { + "epoch": 1.848561388095761, + "grad_norm": 2.1018693447113037, + "learning_rate": 1e-06, + "loss": 1.0619, + "mean_token_accuracy": 0.678755521774292, + "num_tokens": 428119435.0, + "step": 16833 + }, + { + "epoch": 1.8486712057983747, + "grad_norm": 2.160858154296875, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7226879596710205, + "num_tokens": 428144837.0, + "step": 16834 + }, + { + "epoch": 1.8487810235009885, + "grad_norm": 2.2316646575927734, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7189653515815735, + "num_tokens": 428170022.0, + "step": 16835 + }, + { + "epoch": 1.848890841203602, + "grad_norm": 2.071535348892212, + "learning_rate": 1e-06, + "loss": 0.991, + "mean_token_accuracy": 0.6964569091796875, + "num_tokens": 428201223.0, + "step": 16836 + }, + { + "epoch": 1.8490006589062156, + "grad_norm": 2.202594518661499, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7013787031173706, + "num_tokens": 428228875.0, + "step": 16837 + }, + { + "epoch": 1.8491104766088293, + "grad_norm": 2.5642189979553223, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7318091988563538, + "num_tokens": 428249409.0, + "step": 16838 + }, + { + "epoch": 1.849220294311443, + "grad_norm": 2.3373303413391113, + "learning_rate": 1e-06, + "loss": 0.8667, + "mean_token_accuracy": 0.7247823476791382, + "num_tokens": 428274893.0, + "step": 16839 + }, + { + "epoch": 1.8493301120140566, + "grad_norm": 1.8985800743103027, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.7018097639083862, + "num_tokens": 428309556.0, + "step": 16840 + }, + { + "epoch": 1.8494399297166704, + "grad_norm": 2.558866500854492, + "learning_rate": 1e-06, + "loss": 0.7877, + "mean_token_accuracy": 0.7498233318328857, + "num_tokens": 428328895.0, + "step": 16841 + }, + { + "epoch": 1.849549747419284, + "grad_norm": 1.89095938205719, + "learning_rate": 1e-06, + "loss": 0.9798, + "mean_token_accuracy": 0.6995318531990051, + "num_tokens": 428365517.0, + "step": 16842 + }, + { + "epoch": 1.8496595651218977, + "grad_norm": 2.2093095779418945, + "learning_rate": 1e-06, + "loss": 0.8839, + "mean_token_accuracy": 0.7240382432937622, + "num_tokens": 428390347.0, + "step": 16843 + }, + { + "epoch": 1.8497693828245114, + "grad_norm": 2.556814193725586, + "learning_rate": 1e-06, + "loss": 0.8355, + "mean_token_accuracy": 0.7351531386375427, + "num_tokens": 428410233.0, + "step": 16844 + }, + { + "epoch": 1.849879200527125, + "grad_norm": 1.968643069267273, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7280387878417969, + "num_tokens": 428440650.0, + "step": 16845 + }, + { + "epoch": 1.8499890182297385, + "grad_norm": 2.1273601055145264, + "learning_rate": 1e-06, + "loss": 0.9892, + "mean_token_accuracy": 0.6978733539581299, + "num_tokens": 428468275.0, + "step": 16846 + }, + { + "epoch": 1.8500988359323522, + "grad_norm": 2.186998128890991, + "learning_rate": 1e-06, + "loss": 0.9711, + "mean_token_accuracy": 0.7006319761276245, + "num_tokens": 428497965.0, + "step": 16847 + }, + { + "epoch": 1.850208653634966, + "grad_norm": 2.247896194458008, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.7171049118041992, + "num_tokens": 428524268.0, + "step": 16848 + }, + { + "epoch": 1.8503184713375798, + "grad_norm": 2.4864089488983154, + "learning_rate": 1e-06, + "loss": 0.8677, + "mean_token_accuracy": 0.7259563207626343, + "num_tokens": 428545579.0, + "step": 16849 + }, + { + "epoch": 1.8504282890401933, + "grad_norm": 2.1238701343536377, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.72887122631073, + "num_tokens": 428574710.0, + "step": 16850 + }, + { + "epoch": 1.8505381067428068, + "grad_norm": 2.045031785964966, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7209975123405457, + "num_tokens": 428602803.0, + "step": 16851 + }, + { + "epoch": 1.8506479244454206, + "grad_norm": 2.4089620113372803, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7394212484359741, + "num_tokens": 428624251.0, + "step": 16852 + }, + { + "epoch": 1.8507577421480343, + "grad_norm": 2.31489634513855, + "learning_rate": 1e-06, + "loss": 0.8875, + "mean_token_accuracy": 0.7239111661911011, + "num_tokens": 428647073.0, + "step": 16853 + }, + { + "epoch": 1.8508675598506479, + "grad_norm": 2.2873451709747314, + "learning_rate": 1e-06, + "loss": 0.8848, + "mean_token_accuracy": 0.7226781845092773, + "num_tokens": 428672329.0, + "step": 16854 + }, + { + "epoch": 1.8509773775532616, + "grad_norm": 2.3756678104400635, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7075223922729492, + "num_tokens": 428696874.0, + "step": 16855 + }, + { + "epoch": 1.8510871952558752, + "grad_norm": 2.1708619594573975, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7178586721420288, + "num_tokens": 428724810.0, + "step": 16856 + }, + { + "epoch": 1.851197012958489, + "grad_norm": 2.142045021057129, + "learning_rate": 1e-06, + "loss": 0.8844, + "mean_token_accuracy": 0.7262426614761353, + "num_tokens": 428750774.0, + "step": 16857 + }, + { + "epoch": 1.8513068306611027, + "grad_norm": 2.232334613800049, + "learning_rate": 1e-06, + "loss": 0.9793, + "mean_token_accuracy": 0.6973488926887512, + "num_tokens": 428777387.0, + "step": 16858 + }, + { + "epoch": 1.8514166483637162, + "grad_norm": 2.469087839126587, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.7181740403175354, + "num_tokens": 428799535.0, + "step": 16859 + }, + { + "epoch": 1.8515264660663298, + "grad_norm": 2.3210654258728027, + "learning_rate": 1e-06, + "loss": 0.8263, + "mean_token_accuracy": 0.7375998497009277, + "num_tokens": 428822607.0, + "step": 16860 + }, + { + "epoch": 1.8516362837689435, + "grad_norm": 2.2316906452178955, + "learning_rate": 1e-06, + "loss": 0.828, + "mean_token_accuracy": 0.7384055852890015, + "num_tokens": 428847319.0, + "step": 16861 + }, + { + "epoch": 1.8517461014715573, + "grad_norm": 2.357738733291626, + "learning_rate": 1e-06, + "loss": 0.844, + "mean_token_accuracy": 0.7282854914665222, + "num_tokens": 428870917.0, + "step": 16862 + }, + { + "epoch": 1.851855919174171, + "grad_norm": 2.127078056335449, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7182108759880066, + "num_tokens": 428897540.0, + "step": 16863 + }, + { + "epoch": 1.8519657368767846, + "grad_norm": 2.362987756729126, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.7432316541671753, + "num_tokens": 428918926.0, + "step": 16864 + }, + { + "epoch": 1.852075554579398, + "grad_norm": 1.9140772819519043, + "learning_rate": 1e-06, + "loss": 0.9199, + "mean_token_accuracy": 0.7102488279342651, + "num_tokens": 428949654.0, + "step": 16865 + }, + { + "epoch": 1.8521853722820119, + "grad_norm": 2.259290933609009, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7074092030525208, + "num_tokens": 428975244.0, + "step": 16866 + }, + { + "epoch": 1.8522951899846256, + "grad_norm": 2.00302791595459, + "learning_rate": 1e-06, + "loss": 0.9304, + "mean_token_accuracy": 0.7155999541282654, + "num_tokens": 429007630.0, + "step": 16867 + }, + { + "epoch": 1.8524050076872391, + "grad_norm": 2.2138559818267822, + "learning_rate": 1e-06, + "loss": 0.9239, + "mean_token_accuracy": 0.7232306003570557, + "num_tokens": 429033397.0, + "step": 16868 + }, + { + "epoch": 1.8525148253898527, + "grad_norm": 2.181129217147827, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.718907356262207, + "num_tokens": 429060907.0, + "step": 16869 + }, + { + "epoch": 1.8526246430924664, + "grad_norm": 2.051970958709717, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.737497866153717, + "num_tokens": 429088652.0, + "step": 16870 + }, + { + "epoch": 1.8527344607950802, + "grad_norm": 2.0885348320007324, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7220137119293213, + "num_tokens": 429116567.0, + "step": 16871 + }, + { + "epoch": 1.852844278497694, + "grad_norm": 2.123206377029419, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7214542627334595, + "num_tokens": 429145963.0, + "step": 16872 + }, + { + "epoch": 1.8529540962003075, + "grad_norm": 2.2560508251190186, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7173402309417725, + "num_tokens": 429171509.0, + "step": 16873 + }, + { + "epoch": 1.853063913902921, + "grad_norm": 2.351667881011963, + "learning_rate": 1e-06, + "loss": 0.9049, + "mean_token_accuracy": 0.718596875667572, + "num_tokens": 429196396.0, + "step": 16874 + }, + { + "epoch": 1.8531737316055348, + "grad_norm": 2.2243599891662598, + "learning_rate": 1e-06, + "loss": 0.9243, + "mean_token_accuracy": 0.7168043851852417, + "num_tokens": 429222863.0, + "step": 16875 + }, + { + "epoch": 1.8532835493081485, + "grad_norm": 2.255483865737915, + "learning_rate": 1e-06, + "loss": 0.8616, + "mean_token_accuracy": 0.7309043407440186, + "num_tokens": 429247121.0, + "step": 16876 + }, + { + "epoch": 1.8533933670107623, + "grad_norm": 2.3550422191619873, + "learning_rate": 1e-06, + "loss": 0.9911, + "mean_token_accuracy": 0.7052977085113525, + "num_tokens": 429271624.0, + "step": 16877 + }, + { + "epoch": 1.8535031847133758, + "grad_norm": 2.1771955490112305, + "learning_rate": 1e-06, + "loss": 0.764, + "mean_token_accuracy": 0.7576291561126709, + "num_tokens": 429296449.0, + "step": 16878 + }, + { + "epoch": 1.8536130024159894, + "grad_norm": 1.9257967472076416, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.728797197341919, + "num_tokens": 429328271.0, + "step": 16879 + }, + { + "epoch": 1.8537228201186031, + "grad_norm": 2.431265354156494, + "learning_rate": 1e-06, + "loss": 0.8142, + "mean_token_accuracy": 0.7472357749938965, + "num_tokens": 429350709.0, + "step": 16880 + }, + { + "epoch": 1.8538326378212169, + "grad_norm": 2.2881526947021484, + "learning_rate": 1e-06, + "loss": 0.981, + "mean_token_accuracy": 0.6939936876296997, + "num_tokens": 429375223.0, + "step": 16881 + }, + { + "epoch": 1.8539424555238304, + "grad_norm": 2.132192850112915, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7165328860282898, + "num_tokens": 429402669.0, + "step": 16882 + }, + { + "epoch": 1.854052273226444, + "grad_norm": 2.03494930267334, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7159730195999146, + "num_tokens": 429431744.0, + "step": 16883 + }, + { + "epoch": 1.8541620909290577, + "grad_norm": 2.450571060180664, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7394663095474243, + "num_tokens": 429452441.0, + "step": 16884 + }, + { + "epoch": 1.8542719086316715, + "grad_norm": 2.4611804485321045, + "learning_rate": 1e-06, + "loss": 0.7884, + "mean_token_accuracy": 0.7438814043998718, + "num_tokens": 429472296.0, + "step": 16885 + }, + { + "epoch": 1.8543817263342852, + "grad_norm": 2.447244167327881, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7409906983375549, + "num_tokens": 429493274.0, + "step": 16886 + }, + { + "epoch": 1.8544915440368988, + "grad_norm": 2.296555280685425, + "learning_rate": 1e-06, + "loss": 0.9101, + "mean_token_accuracy": 0.7155662178993225, + "num_tokens": 429516596.0, + "step": 16887 + }, + { + "epoch": 1.8546013617395123, + "grad_norm": 2.2196738719940186, + "learning_rate": 1e-06, + "loss": 0.9254, + "mean_token_accuracy": 0.7098495960235596, + "num_tokens": 429544965.0, + "step": 16888 + }, + { + "epoch": 1.854711179442126, + "grad_norm": 2.1809709072113037, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7178643941879272, + "num_tokens": 429570716.0, + "step": 16889 + }, + { + "epoch": 1.8548209971447398, + "grad_norm": 2.5167598724365234, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7413522601127625, + "num_tokens": 429590960.0, + "step": 16890 + }, + { + "epoch": 1.8549308148473536, + "grad_norm": 2.1759347915649414, + "learning_rate": 1e-06, + "loss": 1.0229, + "mean_token_accuracy": 0.6963866949081421, + "num_tokens": 429620418.0, + "step": 16891 + }, + { + "epoch": 1.855040632549967, + "grad_norm": 2.4943814277648926, + "learning_rate": 1e-06, + "loss": 0.8646, + "mean_token_accuracy": 0.7350901365280151, + "num_tokens": 429639729.0, + "step": 16892 + }, + { + "epoch": 1.8551504502525806, + "grad_norm": 2.5137364864349365, + "learning_rate": 1e-06, + "loss": 0.9247, + "mean_token_accuracy": 0.7112574577331543, + "num_tokens": 429661411.0, + "step": 16893 + }, + { + "epoch": 1.8552602679551944, + "grad_norm": 2.1899073123931885, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6910701990127563, + "num_tokens": 429689707.0, + "step": 16894 + }, + { + "epoch": 1.8553700856578081, + "grad_norm": 2.527843475341797, + "learning_rate": 1e-06, + "loss": 0.853, + "mean_token_accuracy": 0.7290401458740234, + "num_tokens": 429711611.0, + "step": 16895 + }, + { + "epoch": 1.8554799033604217, + "grad_norm": 2.326869010925293, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7270873785018921, + "num_tokens": 429735974.0, + "step": 16896 + }, + { + "epoch": 1.8555897210630352, + "grad_norm": 2.250282049179077, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.709990382194519, + "num_tokens": 429762148.0, + "step": 16897 + }, + { + "epoch": 1.855699538765649, + "grad_norm": 2.4118845462799072, + "learning_rate": 1e-06, + "loss": 0.9108, + "mean_token_accuracy": 0.736609935760498, + "num_tokens": 429784358.0, + "step": 16898 + }, + { + "epoch": 1.8558093564682627, + "grad_norm": 2.264693260192871, + "learning_rate": 1e-06, + "loss": 1.0136, + "mean_token_accuracy": 0.6901218295097351, + "num_tokens": 429812743.0, + "step": 16899 + }, + { + "epoch": 1.8559191741708765, + "grad_norm": 2.5756378173828125, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7050321698188782, + "num_tokens": 429835730.0, + "step": 16900 + }, + { + "epoch": 1.85602899187349, + "grad_norm": 2.1767361164093018, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7285014390945435, + "num_tokens": 429863683.0, + "step": 16901 + }, + { + "epoch": 1.8561388095761036, + "grad_norm": 2.4825143814086914, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7214484810829163, + "num_tokens": 429886339.0, + "step": 16902 + }, + { + "epoch": 1.8562486272787173, + "grad_norm": 2.4816906452178955, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7018697261810303, + "num_tokens": 429910270.0, + "step": 16903 + }, + { + "epoch": 1.856358444981331, + "grad_norm": 2.071906089782715, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7117906808853149, + "num_tokens": 429938791.0, + "step": 16904 + }, + { + "epoch": 1.8564682626839446, + "grad_norm": 2.049365520477295, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.724741518497467, + "num_tokens": 429969340.0, + "step": 16905 + }, + { + "epoch": 1.8565780803865584, + "grad_norm": 2.239086151123047, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7109538316726685, + "num_tokens": 429995649.0, + "step": 16906 + }, + { + "epoch": 1.856687898089172, + "grad_norm": 2.3155713081359863, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7316220998764038, + "num_tokens": 430019979.0, + "step": 16907 + }, + { + "epoch": 1.8567977157917857, + "grad_norm": 2.2305781841278076, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7294223308563232, + "num_tokens": 430044988.0, + "step": 16908 + }, + { + "epoch": 1.8569075334943994, + "grad_norm": 2.645076274871826, + "learning_rate": 1e-06, + "loss": 0.9697, + "mean_token_accuracy": 0.7020165920257568, + "num_tokens": 430067372.0, + "step": 16909 + }, + { + "epoch": 1.857017351197013, + "grad_norm": 1.9949499368667603, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7160706520080566, + "num_tokens": 430099363.0, + "step": 16910 + }, + { + "epoch": 1.8571271688996265, + "grad_norm": 2.174034595489502, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.700995683670044, + "num_tokens": 430127546.0, + "step": 16911 + }, + { + "epoch": 1.8572369866022402, + "grad_norm": 2.3856759071350098, + "learning_rate": 1e-06, + "loss": 0.8151, + "mean_token_accuracy": 0.7375788688659668, + "num_tokens": 430149271.0, + "step": 16912 + }, + { + "epoch": 1.857346804304854, + "grad_norm": 2.2457046508789062, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.7046085596084595, + "num_tokens": 430175277.0, + "step": 16913 + }, + { + "epoch": 1.8574566220074678, + "grad_norm": 2.388430595397949, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.7184740304946899, + "num_tokens": 430197800.0, + "step": 16914 + }, + { + "epoch": 1.8575664397100813, + "grad_norm": 2.1856515407562256, + "learning_rate": 1e-06, + "loss": 0.8156, + "mean_token_accuracy": 0.7435722351074219, + "num_tokens": 430222679.0, + "step": 16915 + }, + { + "epoch": 1.8576762574126948, + "grad_norm": 2.4585940837860107, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7196303606033325, + "num_tokens": 430245818.0, + "step": 16916 + }, + { + "epoch": 1.8577860751153086, + "grad_norm": 2.1302952766418457, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6921144723892212, + "num_tokens": 430274793.0, + "step": 16917 + }, + { + "epoch": 1.8578958928179223, + "grad_norm": 2.0664682388305664, + "learning_rate": 1e-06, + "loss": 0.9476, + "mean_token_accuracy": 0.7060337066650391, + "num_tokens": 430306192.0, + "step": 16918 + }, + { + "epoch": 1.8580057105205359, + "grad_norm": 2.325059413909912, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7025564908981323, + "num_tokens": 430329374.0, + "step": 16919 + }, + { + "epoch": 1.8581155282231496, + "grad_norm": 2.591700553894043, + "learning_rate": 1e-06, + "loss": 0.826, + "mean_token_accuracy": 0.73451828956604, + "num_tokens": 430349443.0, + "step": 16920 + }, + { + "epoch": 1.8582253459257632, + "grad_norm": 2.102743148803711, + "learning_rate": 1e-06, + "loss": 0.9986, + "mean_token_accuracy": 0.6904357671737671, + "num_tokens": 430379370.0, + "step": 16921 + }, + { + "epoch": 1.858335163628377, + "grad_norm": 2.187727928161621, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7042719721794128, + "num_tokens": 430408123.0, + "step": 16922 + }, + { + "epoch": 1.8584449813309907, + "grad_norm": 2.850438117980957, + "learning_rate": 1e-06, + "loss": 0.8269, + "mean_token_accuracy": 0.7371109127998352, + "num_tokens": 430425479.0, + "step": 16923 + }, + { + "epoch": 1.8585547990336042, + "grad_norm": 2.254713773727417, + "learning_rate": 1e-06, + "loss": 0.8696, + "mean_token_accuracy": 0.7246666550636292, + "num_tokens": 430449838.0, + "step": 16924 + }, + { + "epoch": 1.8586646167362177, + "grad_norm": 2.0426113605499268, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7312225699424744, + "num_tokens": 430478066.0, + "step": 16925 + }, + { + "epoch": 1.8587744344388315, + "grad_norm": 1.9911208152770996, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7241764664649963, + "num_tokens": 430507955.0, + "step": 16926 + }, + { + "epoch": 1.8588842521414453, + "grad_norm": 2.1940724849700928, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7129012942314148, + "num_tokens": 430533354.0, + "step": 16927 + }, + { + "epoch": 1.858994069844059, + "grad_norm": 2.371218681335449, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7324916124343872, + "num_tokens": 430556248.0, + "step": 16928 + }, + { + "epoch": 1.8591038875466726, + "grad_norm": 2.123894691467285, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6857516169548035, + "num_tokens": 430586505.0, + "step": 16929 + }, + { + "epoch": 1.859213705249286, + "grad_norm": 2.31732439994812, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7051491141319275, + "num_tokens": 430611216.0, + "step": 16930 + }, + { + "epoch": 1.8593235229518998, + "grad_norm": 2.758474588394165, + "learning_rate": 1e-06, + "loss": 0.8154, + "mean_token_accuracy": 0.7471067905426025, + "num_tokens": 430630698.0, + "step": 16931 + }, + { + "epoch": 1.8594333406545136, + "grad_norm": 2.5953733921051025, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7253048419952393, + "num_tokens": 430652174.0, + "step": 16932 + }, + { + "epoch": 1.8595431583571271, + "grad_norm": 2.2521770000457764, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7200829982757568, + "num_tokens": 430678230.0, + "step": 16933 + }, + { + "epoch": 1.8596529760597407, + "grad_norm": 2.15229868888855, + "learning_rate": 1e-06, + "loss": 0.9485, + "mean_token_accuracy": 0.7118502855300903, + "num_tokens": 430705737.0, + "step": 16934 + }, + { + "epoch": 1.8597627937623544, + "grad_norm": 2.263606548309326, + "learning_rate": 1e-06, + "loss": 0.8163, + "mean_token_accuracy": 0.7384493350982666, + "num_tokens": 430728997.0, + "step": 16935 + }, + { + "epoch": 1.8598726114649682, + "grad_norm": 2.2370500564575195, + "learning_rate": 1e-06, + "loss": 0.9593, + "mean_token_accuracy": 0.6974021792411804, + "num_tokens": 430754795.0, + "step": 16936 + }, + { + "epoch": 1.859982429167582, + "grad_norm": 2.422790288925171, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7197847366333008, + "num_tokens": 430777548.0, + "step": 16937 + }, + { + "epoch": 1.8600922468701955, + "grad_norm": 2.50604510307312, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7229586839675903, + "num_tokens": 430799433.0, + "step": 16938 + }, + { + "epoch": 1.860202064572809, + "grad_norm": 2.514810800552368, + "learning_rate": 1e-06, + "loss": 0.7811, + "mean_token_accuracy": 0.7518661618232727, + "num_tokens": 430818954.0, + "step": 16939 + }, + { + "epoch": 1.8603118822754228, + "grad_norm": 2.0616300106048584, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.717376708984375, + "num_tokens": 430847379.0, + "step": 16940 + }, + { + "epoch": 1.8604216999780365, + "grad_norm": 2.181504487991333, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7174382209777832, + "num_tokens": 430872797.0, + "step": 16941 + }, + { + "epoch": 1.8605315176806503, + "grad_norm": 2.2344601154327393, + "learning_rate": 1e-06, + "loss": 0.8586, + "mean_token_accuracy": 0.7323186993598938, + "num_tokens": 430897236.0, + "step": 16942 + }, + { + "epoch": 1.8606413353832638, + "grad_norm": 1.9462025165557861, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.7241970300674438, + "num_tokens": 430928879.0, + "step": 16943 + }, + { + "epoch": 1.8607511530858774, + "grad_norm": 2.336143970489502, + "learning_rate": 1e-06, + "loss": 0.8958, + "mean_token_accuracy": 0.7307864427566528, + "num_tokens": 430952298.0, + "step": 16944 + }, + { + "epoch": 1.860860970788491, + "grad_norm": 2.2001793384552, + "learning_rate": 1e-06, + "loss": 0.9012, + "mean_token_accuracy": 0.719257652759552, + "num_tokens": 430979245.0, + "step": 16945 + }, + { + "epoch": 1.8609707884911049, + "grad_norm": 2.252448081970215, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7175967693328857, + "num_tokens": 431004823.0, + "step": 16946 + }, + { + "epoch": 1.8610806061937184, + "grad_norm": 2.2234580516815186, + "learning_rate": 1e-06, + "loss": 0.9942, + "mean_token_accuracy": 0.7035094499588013, + "num_tokens": 431032595.0, + "step": 16947 + }, + { + "epoch": 1.861190423896332, + "grad_norm": 2.2729716300964355, + "learning_rate": 1e-06, + "loss": 0.8815, + "mean_token_accuracy": 0.7240128517150879, + "num_tokens": 431057611.0, + "step": 16948 + }, + { + "epoch": 1.8613002415989457, + "grad_norm": 1.856536865234375, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7199726104736328, + "num_tokens": 431092491.0, + "step": 16949 + }, + { + "epoch": 1.8614100593015594, + "grad_norm": 1.924391508102417, + "learning_rate": 1e-06, + "loss": 0.9949, + "mean_token_accuracy": 0.6924200654029846, + "num_tokens": 431126611.0, + "step": 16950 + }, + { + "epoch": 1.8615198770041732, + "grad_norm": 2.270496129989624, + "learning_rate": 1e-06, + "loss": 0.9583, + "mean_token_accuracy": 0.7202068567276001, + "num_tokens": 431151415.0, + "step": 16951 + }, + { + "epoch": 1.8616296947067867, + "grad_norm": 2.4623255729675293, + "learning_rate": 1e-06, + "loss": 0.9027, + "mean_token_accuracy": 0.7251638174057007, + "num_tokens": 431173062.0, + "step": 16952 + }, + { + "epoch": 1.8617395124094003, + "grad_norm": 2.326305627822876, + "learning_rate": 1e-06, + "loss": 0.9394, + "mean_token_accuracy": 0.7039903402328491, + "num_tokens": 431196981.0, + "step": 16953 + }, + { + "epoch": 1.861849330112014, + "grad_norm": 2.1980628967285156, + "learning_rate": 1e-06, + "loss": 0.9211, + "mean_token_accuracy": 0.7135452628135681, + "num_tokens": 431222978.0, + "step": 16954 + }, + { + "epoch": 1.8619591478146278, + "grad_norm": 2.3080689907073975, + "learning_rate": 1e-06, + "loss": 0.8452, + "mean_token_accuracy": 0.7333716154098511, + "num_tokens": 431245939.0, + "step": 16955 + }, + { + "epoch": 1.8620689655172413, + "grad_norm": 2.1236281394958496, + "learning_rate": 1e-06, + "loss": 0.9499, + "mean_token_accuracy": 0.6971879601478577, + "num_tokens": 431274639.0, + "step": 16956 + }, + { + "epoch": 1.862178783219855, + "grad_norm": 2.0563387870788574, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7014486193656921, + "num_tokens": 431302513.0, + "step": 16957 + }, + { + "epoch": 1.8622886009224686, + "grad_norm": 2.2158780097961426, + "learning_rate": 1e-06, + "loss": 1.0308, + "mean_token_accuracy": 0.6964280009269714, + "num_tokens": 431331927.0, + "step": 16958 + }, + { + "epoch": 1.8623984186250824, + "grad_norm": 2.1179039478302, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7100368738174438, + "num_tokens": 431361498.0, + "step": 16959 + }, + { + "epoch": 1.8625082363276961, + "grad_norm": 1.8828786611557007, + "learning_rate": 1e-06, + "loss": 0.916, + "mean_token_accuracy": 0.7129608392715454, + "num_tokens": 431394235.0, + "step": 16960 + }, + { + "epoch": 1.8626180540303097, + "grad_norm": 2.1880061626434326, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7213943004608154, + "num_tokens": 431418993.0, + "step": 16961 + }, + { + "epoch": 1.8627278717329232, + "grad_norm": 2.461397886276245, + "learning_rate": 1e-06, + "loss": 0.8948, + "mean_token_accuracy": 0.7157445549964905, + "num_tokens": 431440995.0, + "step": 16962 + }, + { + "epoch": 1.862837689435537, + "grad_norm": 2.2093074321746826, + "learning_rate": 1e-06, + "loss": 0.9992, + "mean_token_accuracy": 0.6921455264091492, + "num_tokens": 431468072.0, + "step": 16963 + }, + { + "epoch": 1.8629475071381507, + "grad_norm": 2.540083408355713, + "learning_rate": 1e-06, + "loss": 0.9279, + "mean_token_accuracy": 0.7047544717788696, + "num_tokens": 431491003.0, + "step": 16964 + }, + { + "epoch": 1.8630573248407645, + "grad_norm": 2.068941354751587, + "learning_rate": 1e-06, + "loss": 0.9557, + "mean_token_accuracy": 0.7066977024078369, + "num_tokens": 431521805.0, + "step": 16965 + }, + { + "epoch": 1.863167142543378, + "grad_norm": 2.1251566410064697, + "learning_rate": 1e-06, + "loss": 0.8334, + "mean_token_accuracy": 0.7327911853790283, + "num_tokens": 431549045.0, + "step": 16966 + }, + { + "epoch": 1.8632769602459915, + "grad_norm": 2.6839985847473145, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7279860973358154, + "num_tokens": 431568177.0, + "step": 16967 + }, + { + "epoch": 1.8633867779486053, + "grad_norm": 2.2581470012664795, + "learning_rate": 1e-06, + "loss": 0.9796, + "mean_token_accuracy": 0.7086557745933533, + "num_tokens": 431595900.0, + "step": 16968 + }, + { + "epoch": 1.863496595651219, + "grad_norm": 1.988776683807373, + "learning_rate": 1e-06, + "loss": 0.9857, + "mean_token_accuracy": 0.6957513093948364, + "num_tokens": 431626680.0, + "step": 16969 + }, + { + "epoch": 1.8636064133538326, + "grad_norm": 2.0634188652038574, + "learning_rate": 1e-06, + "loss": 0.99, + "mean_token_accuracy": 0.6954675912857056, + "num_tokens": 431657000.0, + "step": 16970 + }, + { + "epoch": 1.8637162310564463, + "grad_norm": 2.4135684967041016, + "learning_rate": 1e-06, + "loss": 0.9248, + "mean_token_accuracy": 0.7184914350509644, + "num_tokens": 431679116.0, + "step": 16971 + }, + { + "epoch": 1.8638260487590599, + "grad_norm": 2.25258469581604, + "learning_rate": 1e-06, + "loss": 0.8753, + "mean_token_accuracy": 0.7295379042625427, + "num_tokens": 431703783.0, + "step": 16972 + }, + { + "epoch": 1.8639358664616736, + "grad_norm": 2.305070638656616, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7309848070144653, + "num_tokens": 431726754.0, + "step": 16973 + }, + { + "epoch": 1.8640456841642874, + "grad_norm": 2.093730926513672, + "learning_rate": 1e-06, + "loss": 0.9493, + "mean_token_accuracy": 0.7078056335449219, + "num_tokens": 431756291.0, + "step": 16974 + }, + { + "epoch": 1.864155501866901, + "grad_norm": 2.0690183639526367, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.7175619006156921, + "num_tokens": 431785205.0, + "step": 16975 + }, + { + "epoch": 1.8642653195695145, + "grad_norm": 2.0289647579193115, + "learning_rate": 1e-06, + "loss": 0.8772, + "mean_token_accuracy": 0.7206557989120483, + "num_tokens": 431812922.0, + "step": 16976 + }, + { + "epoch": 1.8643751372721282, + "grad_norm": 2.28654408454895, + "learning_rate": 1e-06, + "loss": 0.8847, + "mean_token_accuracy": 0.7239800691604614, + "num_tokens": 431836139.0, + "step": 16977 + }, + { + "epoch": 1.864484954974742, + "grad_norm": 2.2758700847625732, + "learning_rate": 1e-06, + "loss": 0.9209, + "mean_token_accuracy": 0.7130635976791382, + "num_tokens": 431862250.0, + "step": 16978 + }, + { + "epoch": 1.8645947726773557, + "grad_norm": 2.7180912494659424, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7419895529747009, + "num_tokens": 431880156.0, + "step": 16979 + }, + { + "epoch": 1.8647045903799693, + "grad_norm": 2.2145912647247314, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7041723132133484, + "num_tokens": 431906816.0, + "step": 16980 + }, + { + "epoch": 1.8648144080825828, + "grad_norm": 2.0718376636505127, + "learning_rate": 1e-06, + "loss": 0.8629, + "mean_token_accuracy": 0.7273851037025452, + "num_tokens": 431933456.0, + "step": 16981 + }, + { + "epoch": 1.8649242257851966, + "grad_norm": 2.3641951084136963, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7082318067550659, + "num_tokens": 431958411.0, + "step": 16982 + }, + { + "epoch": 1.8650340434878103, + "grad_norm": 2.450634479522705, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7091052532196045, + "num_tokens": 431980814.0, + "step": 16983 + }, + { + "epoch": 1.8651438611904239, + "grad_norm": 2.2606403827667236, + "learning_rate": 1e-06, + "loss": 0.8602, + "mean_token_accuracy": 0.7278862595558167, + "num_tokens": 432005547.0, + "step": 16984 + }, + { + "epoch": 1.8652536788930374, + "grad_norm": 2.2534589767456055, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7232578992843628, + "num_tokens": 432029561.0, + "step": 16985 + }, + { + "epoch": 1.8653634965956511, + "grad_norm": 2.1755032539367676, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.6988130211830139, + "num_tokens": 432058957.0, + "step": 16986 + }, + { + "epoch": 1.865473314298265, + "grad_norm": 2.1443030834198, + "learning_rate": 1e-06, + "loss": 0.9194, + "mean_token_accuracy": 0.7185773849487305, + "num_tokens": 432087412.0, + "step": 16987 + }, + { + "epoch": 1.8655831320008787, + "grad_norm": 2.336337089538574, + "learning_rate": 1e-06, + "loss": 0.9665, + "mean_token_accuracy": 0.7094453573226929, + "num_tokens": 432113246.0, + "step": 16988 + }, + { + "epoch": 1.8656929497034922, + "grad_norm": 2.3760011196136475, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.7191574573516846, + "num_tokens": 432137434.0, + "step": 16989 + }, + { + "epoch": 1.8658027674061057, + "grad_norm": 2.4806361198425293, + "learning_rate": 1e-06, + "loss": 0.8777, + "mean_token_accuracy": 0.7219223976135254, + "num_tokens": 432159103.0, + "step": 16990 + }, + { + "epoch": 1.8659125851087195, + "grad_norm": 2.356583833694458, + "learning_rate": 1e-06, + "loss": 0.8991, + "mean_token_accuracy": 0.7144026756286621, + "num_tokens": 432184342.0, + "step": 16991 + }, + { + "epoch": 1.8660224028113332, + "grad_norm": 2.1991183757781982, + "learning_rate": 1e-06, + "loss": 0.9923, + "mean_token_accuracy": 0.7021383047103882, + "num_tokens": 432212692.0, + "step": 16992 + }, + { + "epoch": 1.866132220513947, + "grad_norm": 2.5291080474853516, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7362966537475586, + "num_tokens": 432234044.0, + "step": 16993 + }, + { + "epoch": 1.8662420382165605, + "grad_norm": 2.1607067584991455, + "learning_rate": 1e-06, + "loss": 0.8803, + "mean_token_accuracy": 0.7280120849609375, + "num_tokens": 432260821.0, + "step": 16994 + }, + { + "epoch": 1.866351855919174, + "grad_norm": 2.5697009563446045, + "learning_rate": 1e-06, + "loss": 0.8102, + "mean_token_accuracy": 0.7445319890975952, + "num_tokens": 432280608.0, + "step": 16995 + }, + { + "epoch": 1.8664616736217878, + "grad_norm": 2.4485907554626465, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7282868027687073, + "num_tokens": 432303090.0, + "step": 16996 + }, + { + "epoch": 1.8665714913244016, + "grad_norm": 2.403599739074707, + "learning_rate": 1e-06, + "loss": 0.8642, + "mean_token_accuracy": 0.7403247356414795, + "num_tokens": 432324653.0, + "step": 16997 + }, + { + "epoch": 1.8666813090270151, + "grad_norm": 2.2034099102020264, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7182282209396362, + "num_tokens": 432351872.0, + "step": 16998 + }, + { + "epoch": 1.8667911267296287, + "grad_norm": 2.53352427482605, + "learning_rate": 1e-06, + "loss": 0.801, + "mean_token_accuracy": 0.7423320412635803, + "num_tokens": 432372139.0, + "step": 16999 + }, + { + "epoch": 1.8669009444322424, + "grad_norm": 2.071150541305542, + "learning_rate": 1e-06, + "loss": 0.7971, + "mean_token_accuracy": 0.7519415020942688, + "num_tokens": 432400252.0, + "step": 17000 + }, + { + "epoch": 1.8670107621348562, + "grad_norm": 2.0914623737335205, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7119458913803101, + "num_tokens": 432427390.0, + "step": 17001 + }, + { + "epoch": 1.86712057983747, + "grad_norm": 2.368791103363037, + "learning_rate": 1e-06, + "loss": 0.7938, + "mean_token_accuracy": 0.7515385150909424, + "num_tokens": 432448189.0, + "step": 17002 + }, + { + "epoch": 1.8672303975400835, + "grad_norm": 2.205732822418213, + "learning_rate": 1e-06, + "loss": 0.8475, + "mean_token_accuracy": 0.7397230863571167, + "num_tokens": 432472914.0, + "step": 17003 + }, + { + "epoch": 1.867340215242697, + "grad_norm": 2.201131582260132, + "learning_rate": 1e-06, + "loss": 0.8979, + "mean_token_accuracy": 0.7162035703659058, + "num_tokens": 432498050.0, + "step": 17004 + }, + { + "epoch": 1.8674500329453108, + "grad_norm": 2.1798038482666016, + "learning_rate": 1e-06, + "loss": 0.8239, + "mean_token_accuracy": 0.7382097244262695, + "num_tokens": 432522485.0, + "step": 17005 + }, + { + "epoch": 1.8675598506479245, + "grad_norm": 2.2263264656066895, + "learning_rate": 1e-06, + "loss": 0.9133, + "mean_token_accuracy": 0.7159280776977539, + "num_tokens": 432546342.0, + "step": 17006 + }, + { + "epoch": 1.8676696683505383, + "grad_norm": 2.41542387008667, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7202932834625244, + "num_tokens": 432570703.0, + "step": 17007 + }, + { + "epoch": 1.8677794860531518, + "grad_norm": 2.397878646850586, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7221613526344299, + "num_tokens": 432593507.0, + "step": 17008 + }, + { + "epoch": 1.8678893037557653, + "grad_norm": 2.569035291671753, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.717429518699646, + "num_tokens": 432613767.0, + "step": 17009 + }, + { + "epoch": 1.867999121458379, + "grad_norm": 2.34013032913208, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7144560813903809, + "num_tokens": 432637826.0, + "step": 17010 + }, + { + "epoch": 1.8681089391609929, + "grad_norm": 2.3051180839538574, + "learning_rate": 1e-06, + "loss": 0.8495, + "mean_token_accuracy": 0.7298062443733215, + "num_tokens": 432661728.0, + "step": 17011 + }, + { + "epoch": 1.8682187568636064, + "grad_norm": 2.2163150310516357, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7263022661209106, + "num_tokens": 432687996.0, + "step": 17012 + }, + { + "epoch": 1.86832857456622, + "grad_norm": 2.153489828109741, + "learning_rate": 1e-06, + "loss": 0.8993, + "mean_token_accuracy": 0.7195916175842285, + "num_tokens": 432715700.0, + "step": 17013 + }, + { + "epoch": 1.8684383922688337, + "grad_norm": 2.35377836227417, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7176758646965027, + "num_tokens": 432740447.0, + "step": 17014 + }, + { + "epoch": 1.8685482099714474, + "grad_norm": 2.4149510860443115, + "learning_rate": 1e-06, + "loss": 0.8985, + "mean_token_accuracy": 0.7091923952102661, + "num_tokens": 432763221.0, + "step": 17015 + }, + { + "epoch": 1.8686580276740612, + "grad_norm": 2.243558406829834, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7073787450790405, + "num_tokens": 432789446.0, + "step": 17016 + }, + { + "epoch": 1.8687678453766747, + "grad_norm": 2.0027036666870117, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7122390270233154, + "num_tokens": 432819830.0, + "step": 17017 + }, + { + "epoch": 1.8688776630792883, + "grad_norm": 2.1046013832092285, + "learning_rate": 1e-06, + "loss": 0.9017, + "mean_token_accuracy": 0.7219020128250122, + "num_tokens": 432848109.0, + "step": 17018 + }, + { + "epoch": 1.868987480781902, + "grad_norm": 1.9594272375106812, + "learning_rate": 1e-06, + "loss": 0.9448, + "mean_token_accuracy": 0.703808605670929, + "num_tokens": 432881558.0, + "step": 17019 + }, + { + "epoch": 1.8690972984845158, + "grad_norm": 2.369004249572754, + "learning_rate": 1e-06, + "loss": 0.8025, + "mean_token_accuracy": 0.7514196634292603, + "num_tokens": 432903784.0, + "step": 17020 + }, + { + "epoch": 1.8692071161871293, + "grad_norm": 2.150709867477417, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7224223017692566, + "num_tokens": 432929262.0, + "step": 17021 + }, + { + "epoch": 1.869316933889743, + "grad_norm": 2.088674545288086, + "learning_rate": 1e-06, + "loss": 0.874, + "mean_token_accuracy": 0.7263886332511902, + "num_tokens": 432954940.0, + "step": 17022 + }, + { + "epoch": 1.8694267515923566, + "grad_norm": 2.160051107406616, + "learning_rate": 1e-06, + "loss": 0.8598, + "mean_token_accuracy": 0.7294542193412781, + "num_tokens": 432981142.0, + "step": 17023 + }, + { + "epoch": 1.8695365692949704, + "grad_norm": 2.2108612060546875, + "learning_rate": 1e-06, + "loss": 0.8858, + "mean_token_accuracy": 0.718063473701477, + "num_tokens": 433007586.0, + "step": 17024 + }, + { + "epoch": 1.8696463869975841, + "grad_norm": 2.501542091369629, + "learning_rate": 1e-06, + "loss": 0.8111, + "mean_token_accuracy": 0.7394130229949951, + "num_tokens": 433030665.0, + "step": 17025 + }, + { + "epoch": 1.8697562047001977, + "grad_norm": 2.259500741958618, + "learning_rate": 1e-06, + "loss": 0.8342, + "mean_token_accuracy": 0.736053466796875, + "num_tokens": 433053626.0, + "step": 17026 + }, + { + "epoch": 1.8698660224028112, + "grad_norm": 2.305565595626831, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.699162483215332, + "num_tokens": 433078910.0, + "step": 17027 + }, + { + "epoch": 1.869975840105425, + "grad_norm": 2.377413272857666, + "learning_rate": 1e-06, + "loss": 0.8674, + "mean_token_accuracy": 0.7282223105430603, + "num_tokens": 433101105.0, + "step": 17028 + }, + { + "epoch": 1.8700856578080387, + "grad_norm": 2.2907893657684326, + "learning_rate": 1e-06, + "loss": 0.9795, + "mean_token_accuracy": 0.6987144947052002, + "num_tokens": 433126987.0, + "step": 17029 + }, + { + "epoch": 1.8701954755106525, + "grad_norm": 2.267890691757202, + "learning_rate": 1e-06, + "loss": 0.9395, + "mean_token_accuracy": 0.7121391296386719, + "num_tokens": 433153491.0, + "step": 17030 + }, + { + "epoch": 1.870305293213266, + "grad_norm": 2.9433999061584473, + "learning_rate": 1e-06, + "loss": 0.7406, + "mean_token_accuracy": 0.7508814334869385, + "num_tokens": 433168591.0, + "step": 17031 + }, + { + "epoch": 1.8704151109158795, + "grad_norm": 2.195793390274048, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7266465425491333, + "num_tokens": 433193345.0, + "step": 17032 + }, + { + "epoch": 1.8705249286184933, + "grad_norm": 2.4491498470306396, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7305613160133362, + "num_tokens": 433215633.0, + "step": 17033 + }, + { + "epoch": 1.870634746321107, + "grad_norm": 2.073917865753174, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.7214925289154053, + "num_tokens": 433244770.0, + "step": 17034 + }, + { + "epoch": 1.8707445640237206, + "grad_norm": 2.2953104972839355, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.720271646976471, + "num_tokens": 433269724.0, + "step": 17035 + }, + { + "epoch": 1.8708543817263343, + "grad_norm": 2.1689047813415527, + "learning_rate": 1e-06, + "loss": 0.9609, + "mean_token_accuracy": 0.706937849521637, + "num_tokens": 433297530.0, + "step": 17036 + }, + { + "epoch": 1.8709641994289479, + "grad_norm": 2.161684989929199, + "learning_rate": 1e-06, + "loss": 0.9906, + "mean_token_accuracy": 0.6944646835327148, + "num_tokens": 433324586.0, + "step": 17037 + }, + { + "epoch": 1.8710740171315616, + "grad_norm": 2.436039686203003, + "learning_rate": 1e-06, + "loss": 0.8175, + "mean_token_accuracy": 0.7399848103523254, + "num_tokens": 433347167.0, + "step": 17038 + }, + { + "epoch": 1.8711838348341754, + "grad_norm": 2.1466193199157715, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7257588505744934, + "num_tokens": 433372610.0, + "step": 17039 + }, + { + "epoch": 1.871293652536789, + "grad_norm": 2.2291100025177, + "learning_rate": 1e-06, + "loss": 0.9517, + "mean_token_accuracy": 0.7011620998382568, + "num_tokens": 433398749.0, + "step": 17040 + }, + { + "epoch": 1.8714034702394025, + "grad_norm": 2.41521954536438, + "learning_rate": 1e-06, + "loss": 0.8799, + "mean_token_accuracy": 0.7325555086135864, + "num_tokens": 433421024.0, + "step": 17041 + }, + { + "epoch": 1.8715132879420162, + "grad_norm": 2.24397611618042, + "learning_rate": 1e-06, + "loss": 0.8377, + "mean_token_accuracy": 0.7364890575408936, + "num_tokens": 433445938.0, + "step": 17042 + }, + { + "epoch": 1.87162310564463, + "grad_norm": 2.13340425491333, + "learning_rate": 1e-06, + "loss": 0.8214, + "mean_token_accuracy": 0.7399719953536987, + "num_tokens": 433471529.0, + "step": 17043 + }, + { + "epoch": 1.8717329233472437, + "grad_norm": 2.1418824195861816, + "learning_rate": 1e-06, + "loss": 0.9262, + "mean_token_accuracy": 0.7172457575798035, + "num_tokens": 433497455.0, + "step": 17044 + }, + { + "epoch": 1.8718427410498573, + "grad_norm": 2.3580050468444824, + "learning_rate": 1e-06, + "loss": 0.7693, + "mean_token_accuracy": 0.7548147439956665, + "num_tokens": 433520655.0, + "step": 17045 + }, + { + "epoch": 1.8719525587524708, + "grad_norm": 2.066037178039551, + "learning_rate": 1e-06, + "loss": 0.9584, + "mean_token_accuracy": 0.7037050724029541, + "num_tokens": 433554944.0, + "step": 17046 + }, + { + "epoch": 1.8720623764550846, + "grad_norm": 2.36000394821167, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7171459197998047, + "num_tokens": 433576339.0, + "step": 17047 + }, + { + "epoch": 1.8721721941576983, + "grad_norm": 2.173768997192383, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7261940240859985, + "num_tokens": 433601789.0, + "step": 17048 + }, + { + "epoch": 1.8722820118603118, + "grad_norm": 2.2576329708099365, + "learning_rate": 1e-06, + "loss": 0.9474, + "mean_token_accuracy": 0.7016414999961853, + "num_tokens": 433628079.0, + "step": 17049 + }, + { + "epoch": 1.8723918295629254, + "grad_norm": 2.2995243072509766, + "learning_rate": 1e-06, + "loss": 0.8891, + "mean_token_accuracy": 0.7226572036743164, + "num_tokens": 433654187.0, + "step": 17050 + }, + { + "epoch": 1.8725016472655391, + "grad_norm": 2.2982730865478516, + "learning_rate": 1e-06, + "loss": 0.8703, + "mean_token_accuracy": 0.7193581461906433, + "num_tokens": 433678815.0, + "step": 17051 + }, + { + "epoch": 1.872611464968153, + "grad_norm": 2.1873457431793213, + "learning_rate": 1e-06, + "loss": 0.9554, + "mean_token_accuracy": 0.7010329961776733, + "num_tokens": 433706853.0, + "step": 17052 + }, + { + "epoch": 1.8727212826707667, + "grad_norm": 2.1712818145751953, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.722678005695343, + "num_tokens": 433731908.0, + "step": 17053 + }, + { + "epoch": 1.8728311003733802, + "grad_norm": 2.1710891723632812, + "learning_rate": 1e-06, + "loss": 0.9503, + "mean_token_accuracy": 0.7042641639709473, + "num_tokens": 433759578.0, + "step": 17054 + }, + { + "epoch": 1.8729409180759937, + "grad_norm": 2.1711859703063965, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.705743134021759, + "num_tokens": 433787194.0, + "step": 17055 + }, + { + "epoch": 1.8730507357786075, + "grad_norm": 2.306027889251709, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7185332775115967, + "num_tokens": 433811025.0, + "step": 17056 + }, + { + "epoch": 1.8731605534812212, + "grad_norm": 2.1611247062683105, + "learning_rate": 1e-06, + "loss": 0.9318, + "mean_token_accuracy": 0.721287727355957, + "num_tokens": 433839005.0, + "step": 17057 + }, + { + "epoch": 1.873270371183835, + "grad_norm": 2.797853469848633, + "learning_rate": 1e-06, + "loss": 0.796, + "mean_token_accuracy": 0.7461874485015869, + "num_tokens": 433856293.0, + "step": 17058 + }, + { + "epoch": 1.8733801888864485, + "grad_norm": 2.667863368988037, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7318698167800903, + "num_tokens": 433875544.0, + "step": 17059 + }, + { + "epoch": 1.873490006589062, + "grad_norm": 2.015160322189331, + "learning_rate": 1e-06, + "loss": 0.9477, + "mean_token_accuracy": 0.7103427052497864, + "num_tokens": 433907592.0, + "step": 17060 + }, + { + "epoch": 1.8735998242916758, + "grad_norm": 2.023995876312256, + "learning_rate": 1e-06, + "loss": 1.024, + "mean_token_accuracy": 0.6881142258644104, + "num_tokens": 433942512.0, + "step": 17061 + }, + { + "epoch": 1.8737096419942896, + "grad_norm": 2.398608446121216, + "learning_rate": 1e-06, + "loss": 0.8343, + "mean_token_accuracy": 0.7350646257400513, + "num_tokens": 433964427.0, + "step": 17062 + }, + { + "epoch": 1.8738194596969031, + "grad_norm": 2.326077699661255, + "learning_rate": 1e-06, + "loss": 0.9005, + "mean_token_accuracy": 0.7139964699745178, + "num_tokens": 433989206.0, + "step": 17063 + }, + { + "epoch": 1.8739292773995166, + "grad_norm": 2.097947597503662, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7056957483291626, + "num_tokens": 434018829.0, + "step": 17064 + }, + { + "epoch": 1.8740390951021304, + "grad_norm": 2.2004709243774414, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7459501624107361, + "num_tokens": 434043520.0, + "step": 17065 + }, + { + "epoch": 1.8741489128047442, + "grad_norm": 2.5235960483551025, + "learning_rate": 1e-06, + "loss": 0.831, + "mean_token_accuracy": 0.7347272038459778, + "num_tokens": 434063841.0, + "step": 17066 + }, + { + "epoch": 1.874258730507358, + "grad_norm": 2.1751534938812256, + "learning_rate": 1e-06, + "loss": 0.9698, + "mean_token_accuracy": 0.696364164352417, + "num_tokens": 434090441.0, + "step": 17067 + }, + { + "epoch": 1.8743685482099715, + "grad_norm": 2.0962841510772705, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.7275622487068176, + "num_tokens": 434119740.0, + "step": 17068 + }, + { + "epoch": 1.874478365912585, + "grad_norm": 2.180791139602661, + "learning_rate": 1e-06, + "loss": 0.9255, + "mean_token_accuracy": 0.7273748517036438, + "num_tokens": 434145824.0, + "step": 17069 + }, + { + "epoch": 1.8745881836151987, + "grad_norm": 2.202591896057129, + "learning_rate": 1e-06, + "loss": 0.8736, + "mean_token_accuracy": 0.7306628227233887, + "num_tokens": 434171678.0, + "step": 17070 + }, + { + "epoch": 1.8746980013178125, + "grad_norm": 2.3004376888275146, + "learning_rate": 1e-06, + "loss": 0.8819, + "mean_token_accuracy": 0.7212541103363037, + "num_tokens": 434197892.0, + "step": 17071 + }, + { + "epoch": 1.8748078190204263, + "grad_norm": 2.061326503753662, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7073612213134766, + "num_tokens": 434229004.0, + "step": 17072 + }, + { + "epoch": 1.8749176367230398, + "grad_norm": 2.2373242378234863, + "learning_rate": 1e-06, + "loss": 0.9074, + "mean_token_accuracy": 0.7208189368247986, + "num_tokens": 434254004.0, + "step": 17073 + }, + { + "epoch": 1.8750274544256533, + "grad_norm": 2.7972779273986816, + "learning_rate": 1e-06, + "loss": 0.9097, + "mean_token_accuracy": 0.715490460395813, + "num_tokens": 434272608.0, + "step": 17074 + }, + { + "epoch": 1.875137272128267, + "grad_norm": 2.3354759216308594, + "learning_rate": 1e-06, + "loss": 0.8732, + "mean_token_accuracy": 0.7259222269058228, + "num_tokens": 434295663.0, + "step": 17075 + }, + { + "epoch": 1.8752470898308808, + "grad_norm": 2.253041982650757, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7152774333953857, + "num_tokens": 434320714.0, + "step": 17076 + }, + { + "epoch": 1.8753569075334944, + "grad_norm": 2.269429922103882, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7193660736083984, + "num_tokens": 434345488.0, + "step": 17077 + }, + { + "epoch": 1.875466725236108, + "grad_norm": 2.272956371307373, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.714913547039032, + "num_tokens": 434370221.0, + "step": 17078 + }, + { + "epoch": 1.8755765429387217, + "grad_norm": 2.0358550548553467, + "learning_rate": 1e-06, + "loss": 0.8723, + "mean_token_accuracy": 0.7301496267318726, + "num_tokens": 434401666.0, + "step": 17079 + }, + { + "epoch": 1.8756863606413354, + "grad_norm": 2.3950929641723633, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7413253784179688, + "num_tokens": 434426300.0, + "step": 17080 + }, + { + "epoch": 1.8757961783439492, + "grad_norm": 2.417283296585083, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7202193737030029, + "num_tokens": 434452039.0, + "step": 17081 + }, + { + "epoch": 1.8759059960465627, + "grad_norm": 2.2263681888580322, + "learning_rate": 1e-06, + "loss": 0.899, + "mean_token_accuracy": 0.7277003526687622, + "num_tokens": 434479651.0, + "step": 17082 + }, + { + "epoch": 1.8760158137491763, + "grad_norm": 2.5722410678863525, + "learning_rate": 1e-06, + "loss": 0.9024, + "mean_token_accuracy": 0.7253418564796448, + "num_tokens": 434500363.0, + "step": 17083 + }, + { + "epoch": 1.87612563145179, + "grad_norm": 2.180267095565796, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7258789539337158, + "num_tokens": 434525591.0, + "step": 17084 + }, + { + "epoch": 1.8762354491544038, + "grad_norm": 2.1806676387786865, + "learning_rate": 1e-06, + "loss": 0.8726, + "mean_token_accuracy": 0.729903519153595, + "num_tokens": 434552909.0, + "step": 17085 + }, + { + "epoch": 1.8763452668570173, + "grad_norm": 2.022425413131714, + "learning_rate": 1e-06, + "loss": 0.9488, + "mean_token_accuracy": 0.7109394073486328, + "num_tokens": 434584794.0, + "step": 17086 + }, + { + "epoch": 1.876455084559631, + "grad_norm": 2.3186306953430176, + "learning_rate": 1e-06, + "loss": 0.8594, + "mean_token_accuracy": 0.7293044924736023, + "num_tokens": 434609136.0, + "step": 17087 + }, + { + "epoch": 1.8765649022622446, + "grad_norm": 2.459510087966919, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7251014709472656, + "num_tokens": 434629838.0, + "step": 17088 + }, + { + "epoch": 1.8766747199648584, + "grad_norm": 2.1512298583984375, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7148762345314026, + "num_tokens": 434657228.0, + "step": 17089 + }, + { + "epoch": 1.876784537667472, + "grad_norm": 2.1995370388031006, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7170857191085815, + "num_tokens": 434681875.0, + "step": 17090 + }, + { + "epoch": 1.8768943553700856, + "grad_norm": 2.282524824142456, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7419365048408508, + "num_tokens": 434704792.0, + "step": 17091 + }, + { + "epoch": 1.8770041730726992, + "grad_norm": 2.134791135787964, + "learning_rate": 1e-06, + "loss": 0.8832, + "mean_token_accuracy": 0.7172783613204956, + "num_tokens": 434731946.0, + "step": 17092 + }, + { + "epoch": 1.877113990775313, + "grad_norm": 2.12229585647583, + "learning_rate": 1e-06, + "loss": 0.9889, + "mean_token_accuracy": 0.6905332803726196, + "num_tokens": 434759025.0, + "step": 17093 + }, + { + "epoch": 1.8772238084779267, + "grad_norm": 2.7074761390686035, + "learning_rate": 1e-06, + "loss": 0.9076, + "mean_token_accuracy": 0.7186552286148071, + "num_tokens": 434780215.0, + "step": 17094 + }, + { + "epoch": 1.8773336261805404, + "grad_norm": 2.0984034538269043, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7432690858840942, + "num_tokens": 434806678.0, + "step": 17095 + }, + { + "epoch": 1.877443443883154, + "grad_norm": 2.2414262294769287, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7231860756874084, + "num_tokens": 434831600.0, + "step": 17096 + }, + { + "epoch": 1.8775532615857675, + "grad_norm": 2.096060276031494, + "learning_rate": 1e-06, + "loss": 0.857, + "mean_token_accuracy": 0.7295662760734558, + "num_tokens": 434858286.0, + "step": 17097 + }, + { + "epoch": 1.8776630792883813, + "grad_norm": 2.3936188220977783, + "learning_rate": 1e-06, + "loss": 0.8438, + "mean_token_accuracy": 0.7359161972999573, + "num_tokens": 434879756.0, + "step": 17098 + }, + { + "epoch": 1.877772896990995, + "grad_norm": 2.2043142318725586, + "learning_rate": 1e-06, + "loss": 0.8561, + "mean_token_accuracy": 0.7300034165382385, + "num_tokens": 434904235.0, + "step": 17099 + }, + { + "epoch": 1.8778827146936086, + "grad_norm": 1.9177836179733276, + "learning_rate": 1e-06, + "loss": 0.9601, + "mean_token_accuracy": 0.7073039412498474, + "num_tokens": 434938758.0, + "step": 17100 + }, + { + "epoch": 1.8779925323962223, + "grad_norm": 2.379117727279663, + "learning_rate": 1e-06, + "loss": 0.9196, + "mean_token_accuracy": 0.7095118165016174, + "num_tokens": 434963591.0, + "step": 17101 + }, + { + "epoch": 1.8781023500988359, + "grad_norm": 2.272026538848877, + "learning_rate": 1e-06, + "loss": 0.9322, + "mean_token_accuracy": 0.7153055667877197, + "num_tokens": 434988749.0, + "step": 17102 + }, + { + "epoch": 1.8782121678014496, + "grad_norm": 2.3085930347442627, + "learning_rate": 1e-06, + "loss": 0.9872, + "mean_token_accuracy": 0.6953070163726807, + "num_tokens": 435014821.0, + "step": 17103 + }, + { + "epoch": 1.8783219855040634, + "grad_norm": 2.5613856315612793, + "learning_rate": 1e-06, + "loss": 0.7957, + "mean_token_accuracy": 0.744484543800354, + "num_tokens": 435034388.0, + "step": 17104 + }, + { + "epoch": 1.878431803206677, + "grad_norm": 2.1306116580963135, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7065520286560059, + "num_tokens": 435062816.0, + "step": 17105 + }, + { + "epoch": 1.8785416209092904, + "grad_norm": 1.9173181056976318, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7220859527587891, + "num_tokens": 435097117.0, + "step": 17106 + }, + { + "epoch": 1.8786514386119042, + "grad_norm": 2.4173922538757324, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7234982252120972, + "num_tokens": 435121630.0, + "step": 17107 + }, + { + "epoch": 1.878761256314518, + "grad_norm": 2.5418102741241455, + "learning_rate": 1e-06, + "loss": 0.8769, + "mean_token_accuracy": 0.7280250787734985, + "num_tokens": 435141224.0, + "step": 17108 + }, + { + "epoch": 1.8788710740171317, + "grad_norm": 2.0671298503875732, + "learning_rate": 1e-06, + "loss": 0.8999, + "mean_token_accuracy": 0.7216125726699829, + "num_tokens": 435168903.0, + "step": 17109 + }, + { + "epoch": 1.8789808917197452, + "grad_norm": 1.9046369791030884, + "learning_rate": 1e-06, + "loss": 0.9128, + "mean_token_accuracy": 0.7128499150276184, + "num_tokens": 435202549.0, + "step": 17110 + }, + { + "epoch": 1.8790907094223588, + "grad_norm": 2.0255825519561768, + "learning_rate": 1e-06, + "loss": 0.9307, + "mean_token_accuracy": 0.7056607604026794, + "num_tokens": 435231515.0, + "step": 17111 + }, + { + "epoch": 1.8792005271249725, + "grad_norm": 2.3563103675842285, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.72654789686203, + "num_tokens": 435255209.0, + "step": 17112 + }, + { + "epoch": 1.8793103448275863, + "grad_norm": 2.2607603073120117, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7263836860656738, + "num_tokens": 435279010.0, + "step": 17113 + }, + { + "epoch": 1.8794201625301998, + "grad_norm": 2.290527582168579, + "learning_rate": 1e-06, + "loss": 0.977, + "mean_token_accuracy": 0.7007381319999695, + "num_tokens": 435305442.0, + "step": 17114 + }, + { + "epoch": 1.8795299802328134, + "grad_norm": 2.5663375854492188, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7244083881378174, + "num_tokens": 435325338.0, + "step": 17115 + }, + { + "epoch": 1.8796397979354271, + "grad_norm": 2.079253911972046, + "learning_rate": 1e-06, + "loss": 0.88, + "mean_token_accuracy": 0.7201648354530334, + "num_tokens": 435353365.0, + "step": 17116 + }, + { + "epoch": 1.8797496156380409, + "grad_norm": 2.447955369949341, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.705945611000061, + "num_tokens": 435375057.0, + "step": 17117 + }, + { + "epoch": 1.8798594333406546, + "grad_norm": 2.372891426086426, + "learning_rate": 1e-06, + "loss": 0.8057, + "mean_token_accuracy": 0.7499842047691345, + "num_tokens": 435396367.0, + "step": 17118 + }, + { + "epoch": 1.8799692510432682, + "grad_norm": 2.252704620361328, + "learning_rate": 1e-06, + "loss": 0.9686, + "mean_token_accuracy": 0.70456862449646, + "num_tokens": 435423106.0, + "step": 17119 + }, + { + "epoch": 1.8800790687458817, + "grad_norm": 2.4618632793426514, + "learning_rate": 1e-06, + "loss": 0.8862, + "mean_token_accuracy": 0.724109411239624, + "num_tokens": 435445738.0, + "step": 17120 + }, + { + "epoch": 1.8801888864484955, + "grad_norm": 2.5813705921173096, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7453001737594604, + "num_tokens": 435464262.0, + "step": 17121 + }, + { + "epoch": 1.8802987041511092, + "grad_norm": 1.9819164276123047, + "learning_rate": 1e-06, + "loss": 0.8748, + "mean_token_accuracy": 0.7282134294509888, + "num_tokens": 435493699.0, + "step": 17122 + }, + { + "epoch": 1.880408521853723, + "grad_norm": 2.1863138675689697, + "learning_rate": 1e-06, + "loss": 0.9563, + "mean_token_accuracy": 0.7114214897155762, + "num_tokens": 435520432.0, + "step": 17123 + }, + { + "epoch": 1.8805183395563365, + "grad_norm": 2.1938283443450928, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.714634358882904, + "num_tokens": 435546127.0, + "step": 17124 + }, + { + "epoch": 1.88062815725895, + "grad_norm": 2.0449397563934326, + "learning_rate": 1e-06, + "loss": 0.9315, + "mean_token_accuracy": 0.7083118557929993, + "num_tokens": 435575421.0, + "step": 17125 + }, + { + "epoch": 1.8807379749615638, + "grad_norm": 1.961157202720642, + "learning_rate": 1e-06, + "loss": 0.9167, + "mean_token_accuracy": 0.7164260149002075, + "num_tokens": 435606457.0, + "step": 17126 + }, + { + "epoch": 1.8808477926641776, + "grad_norm": 2.37782621383667, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.724053144454956, + "num_tokens": 435629504.0, + "step": 17127 + }, + { + "epoch": 1.880957610366791, + "grad_norm": 2.076627731323242, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7182388305664062, + "num_tokens": 435661463.0, + "step": 17128 + }, + { + "epoch": 1.8810674280694046, + "grad_norm": 2.298518419265747, + "learning_rate": 1e-06, + "loss": 0.8683, + "mean_token_accuracy": 0.728113055229187, + "num_tokens": 435685902.0, + "step": 17129 + }, + { + "epoch": 1.8811772457720184, + "grad_norm": 2.3581912517547607, + "learning_rate": 1e-06, + "loss": 0.8082, + "mean_token_accuracy": 0.7403934597969055, + "num_tokens": 435706884.0, + "step": 17130 + }, + { + "epoch": 1.8812870634746321, + "grad_norm": 2.2978439331054688, + "learning_rate": 1e-06, + "loss": 0.8951, + "mean_token_accuracy": 0.7205885648727417, + "num_tokens": 435733178.0, + "step": 17131 + }, + { + "epoch": 1.881396881177246, + "grad_norm": 2.364227056503296, + "learning_rate": 1e-06, + "loss": 0.9216, + "mean_token_accuracy": 0.7117941379547119, + "num_tokens": 435757047.0, + "step": 17132 + }, + { + "epoch": 1.8815066988798594, + "grad_norm": 2.111035108566284, + "learning_rate": 1e-06, + "loss": 0.8066, + "mean_token_accuracy": 0.7468793392181396, + "num_tokens": 435783557.0, + "step": 17133 + }, + { + "epoch": 1.881616516582473, + "grad_norm": 2.1052544116973877, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7144907712936401, + "num_tokens": 435811389.0, + "step": 17134 + }, + { + "epoch": 1.8817263342850867, + "grad_norm": 2.305222511291504, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7362514734268188, + "num_tokens": 435835416.0, + "step": 17135 + }, + { + "epoch": 1.8818361519877005, + "grad_norm": 2.457681894302368, + "learning_rate": 1e-06, + "loss": 0.8434, + "mean_token_accuracy": 0.741082489490509, + "num_tokens": 435856578.0, + "step": 17136 + }, + { + "epoch": 1.881945969690314, + "grad_norm": 2.391918659210205, + "learning_rate": 1e-06, + "loss": 0.9171, + "mean_token_accuracy": 0.7263902425765991, + "num_tokens": 435880871.0, + "step": 17137 + }, + { + "epoch": 1.8820557873929278, + "grad_norm": 2.5691301822662354, + "learning_rate": 1e-06, + "loss": 0.7753, + "mean_token_accuracy": 0.7486196160316467, + "num_tokens": 435900904.0, + "step": 17138 + }, + { + "epoch": 1.8821656050955413, + "grad_norm": 2.3268842697143555, + "learning_rate": 1e-06, + "loss": 0.8292, + "mean_token_accuracy": 0.7419093251228333, + "num_tokens": 435923556.0, + "step": 17139 + }, + { + "epoch": 1.882275422798155, + "grad_norm": 2.3824284076690674, + "learning_rate": 1e-06, + "loss": 0.9841, + "mean_token_accuracy": 0.7013416290283203, + "num_tokens": 435949385.0, + "step": 17140 + }, + { + "epoch": 1.8823852405007688, + "grad_norm": 2.363900899887085, + "learning_rate": 1e-06, + "loss": 0.8621, + "mean_token_accuracy": 0.7272845506668091, + "num_tokens": 435974118.0, + "step": 17141 + }, + { + "epoch": 1.8824950582033824, + "grad_norm": 1.9192808866500854, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7408381700515747, + "num_tokens": 436005541.0, + "step": 17142 + }, + { + "epoch": 1.882604875905996, + "grad_norm": 2.2437832355499268, + "learning_rate": 1e-06, + "loss": 0.8686, + "mean_token_accuracy": 0.7200169563293457, + "num_tokens": 436031010.0, + "step": 17143 + }, + { + "epoch": 1.8827146936086097, + "grad_norm": 2.3743209838867188, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7266168594360352, + "num_tokens": 436055780.0, + "step": 17144 + }, + { + "epoch": 1.8828245113112234, + "grad_norm": 2.308582067489624, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7108352780342102, + "num_tokens": 436079929.0, + "step": 17145 + }, + { + "epoch": 1.8829343290138372, + "grad_norm": 2.152225971221924, + "learning_rate": 1e-06, + "loss": 0.8898, + "mean_token_accuracy": 0.7233415842056274, + "num_tokens": 436106511.0, + "step": 17146 + }, + { + "epoch": 1.8830441467164507, + "grad_norm": 2.35404372215271, + "learning_rate": 1e-06, + "loss": 1.0034, + "mean_token_accuracy": 0.6982371211051941, + "num_tokens": 436131656.0, + "step": 17147 + }, + { + "epoch": 1.8831539644190642, + "grad_norm": 2.0366339683532715, + "learning_rate": 1e-06, + "loss": 0.9384, + "mean_token_accuracy": 0.7052814960479736, + "num_tokens": 436162180.0, + "step": 17148 + }, + { + "epoch": 1.883263782121678, + "grad_norm": 2.2467658519744873, + "learning_rate": 1e-06, + "loss": 0.84, + "mean_token_accuracy": 0.7387039661407471, + "num_tokens": 436185124.0, + "step": 17149 + }, + { + "epoch": 1.8833735998242918, + "grad_norm": 2.1365725994110107, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7162296772003174, + "num_tokens": 436212805.0, + "step": 17150 + }, + { + "epoch": 1.8834834175269053, + "grad_norm": 2.4513659477233887, + "learning_rate": 1e-06, + "loss": 0.8466, + "mean_token_accuracy": 0.7291600108146667, + "num_tokens": 436236828.0, + "step": 17151 + }, + { + "epoch": 1.883593235229519, + "grad_norm": 2.2253758907318115, + "learning_rate": 1e-06, + "loss": 0.8207, + "mean_token_accuracy": 0.7406399250030518, + "num_tokens": 436260894.0, + "step": 17152 + }, + { + "epoch": 1.8837030529321326, + "grad_norm": 2.536519765853882, + "learning_rate": 1e-06, + "loss": 0.8702, + "mean_token_accuracy": 0.7228155136108398, + "num_tokens": 436281516.0, + "step": 17153 + }, + { + "epoch": 1.8838128706347463, + "grad_norm": 2.2918412685394287, + "learning_rate": 1e-06, + "loss": 0.8417, + "mean_token_accuracy": 0.7274223566055298, + "num_tokens": 436304908.0, + "step": 17154 + }, + { + "epoch": 1.88392268833736, + "grad_norm": 2.2371022701263428, + "learning_rate": 1e-06, + "loss": 0.9184, + "mean_token_accuracy": 0.7234170436859131, + "num_tokens": 436329370.0, + "step": 17155 + }, + { + "epoch": 1.8840325060399736, + "grad_norm": 2.2786686420440674, + "learning_rate": 1e-06, + "loss": 0.809, + "mean_token_accuracy": 0.73427814245224, + "num_tokens": 436351626.0, + "step": 17156 + }, + { + "epoch": 1.8841423237425872, + "grad_norm": 2.35062313079834, + "learning_rate": 1e-06, + "loss": 0.9295, + "mean_token_accuracy": 0.7082945108413696, + "num_tokens": 436375777.0, + "step": 17157 + }, + { + "epoch": 1.884252141445201, + "grad_norm": 2.2191505432128906, + "learning_rate": 1e-06, + "loss": 0.8601, + "mean_token_accuracy": 0.7260220050811768, + "num_tokens": 436401103.0, + "step": 17158 + }, + { + "epoch": 1.8843619591478147, + "grad_norm": 2.691643476486206, + "learning_rate": 1e-06, + "loss": 0.8481, + "mean_token_accuracy": 0.7326551675796509, + "num_tokens": 436420380.0, + "step": 17159 + }, + { + "epoch": 1.8844717768504284, + "grad_norm": 2.218181848526001, + "learning_rate": 1e-06, + "loss": 0.8652, + "mean_token_accuracy": 0.732860803604126, + "num_tokens": 436446290.0, + "step": 17160 + }, + { + "epoch": 1.884581594553042, + "grad_norm": 2.4669580459594727, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7199912071228027, + "num_tokens": 436467339.0, + "step": 17161 + }, + { + "epoch": 1.8846914122556555, + "grad_norm": 2.319636821746826, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7326919436454773, + "num_tokens": 436489662.0, + "step": 17162 + }, + { + "epoch": 1.8848012299582693, + "grad_norm": 2.1732518672943115, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7378749251365662, + "num_tokens": 436515945.0, + "step": 17163 + }, + { + "epoch": 1.884911047660883, + "grad_norm": 2.1209521293640137, + "learning_rate": 1e-06, + "loss": 0.8836, + "mean_token_accuracy": 0.7236571311950684, + "num_tokens": 436543884.0, + "step": 17164 + }, + { + "epoch": 1.8850208653634966, + "grad_norm": 2.0325138568878174, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7077591419219971, + "num_tokens": 436575695.0, + "step": 17165 + }, + { + "epoch": 1.88513068306611, + "grad_norm": 2.1821773052215576, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7161521315574646, + "num_tokens": 436602396.0, + "step": 17166 + }, + { + "epoch": 1.8852405007687238, + "grad_norm": 2.320223808288574, + "learning_rate": 1e-06, + "loss": 0.8997, + "mean_token_accuracy": 0.7246333956718445, + "num_tokens": 436626791.0, + "step": 17167 + }, + { + "epoch": 1.8853503184713376, + "grad_norm": 2.2447028160095215, + "learning_rate": 1e-06, + "loss": 0.8246, + "mean_token_accuracy": 0.7368754744529724, + "num_tokens": 436651204.0, + "step": 17168 + }, + { + "epoch": 1.8854601361739514, + "grad_norm": 2.235292673110962, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7274670600891113, + "num_tokens": 436677121.0, + "step": 17169 + }, + { + "epoch": 1.885569953876565, + "grad_norm": 2.0257279872894287, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7107290029525757, + "num_tokens": 436705564.0, + "step": 17170 + }, + { + "epoch": 1.8856797715791784, + "grad_norm": 2.0517642498016357, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7105754613876343, + "num_tokens": 436735908.0, + "step": 17171 + }, + { + "epoch": 1.8857895892817922, + "grad_norm": 2.3600685596466064, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7338210344314575, + "num_tokens": 436758281.0, + "step": 17172 + }, + { + "epoch": 1.885899406984406, + "grad_norm": 2.1577961444854736, + "learning_rate": 1e-06, + "loss": 0.9426, + "mean_token_accuracy": 0.7024234533309937, + "num_tokens": 436784271.0, + "step": 17173 + }, + { + "epoch": 1.8860092246870197, + "grad_norm": 2.6761271953582764, + "learning_rate": 1e-06, + "loss": 0.892, + "mean_token_accuracy": 0.723800778388977, + "num_tokens": 436805033.0, + "step": 17174 + }, + { + "epoch": 1.8861190423896332, + "grad_norm": 2.0533571243286133, + "learning_rate": 1e-06, + "loss": 1.0003, + "mean_token_accuracy": 0.6896881461143494, + "num_tokens": 436835537.0, + "step": 17175 + }, + { + "epoch": 1.8862288600922468, + "grad_norm": 2.221893310546875, + "learning_rate": 1e-06, + "loss": 0.9612, + "mean_token_accuracy": 0.705348789691925, + "num_tokens": 436861649.0, + "step": 17176 + }, + { + "epoch": 1.8863386777948605, + "grad_norm": 2.140453577041626, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.713509738445282, + "num_tokens": 436888308.0, + "step": 17177 + }, + { + "epoch": 1.8864484954974743, + "grad_norm": 2.2049548625946045, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7132009267807007, + "num_tokens": 436914723.0, + "step": 17178 + }, + { + "epoch": 1.8865583132000878, + "grad_norm": 2.155494451522827, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7238131761550903, + "num_tokens": 436941148.0, + "step": 17179 + }, + { + "epoch": 1.8866681309027014, + "grad_norm": 1.9041298627853394, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.7251922488212585, + "num_tokens": 436973004.0, + "step": 17180 + }, + { + "epoch": 1.8867779486053151, + "grad_norm": 2.0840468406677246, + "learning_rate": 1e-06, + "loss": 0.9987, + "mean_token_accuracy": 0.690723180770874, + "num_tokens": 437005436.0, + "step": 17181 + }, + { + "epoch": 1.8868877663079289, + "grad_norm": 1.9738987684249878, + "learning_rate": 1e-06, + "loss": 0.9668, + "mean_token_accuracy": 0.7009323835372925, + "num_tokens": 437038348.0, + "step": 17182 + }, + { + "epoch": 1.8869975840105426, + "grad_norm": 2.588027238845825, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7314740419387817, + "num_tokens": 437058490.0, + "step": 17183 + }, + { + "epoch": 1.8871074017131562, + "grad_norm": 2.160733699798584, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7410211563110352, + "num_tokens": 437085084.0, + "step": 17184 + }, + { + "epoch": 1.8872172194157697, + "grad_norm": 2.2518131732940674, + "learning_rate": 1e-06, + "loss": 1.0022, + "mean_token_accuracy": 0.6960660219192505, + "num_tokens": 437112007.0, + "step": 17185 + }, + { + "epoch": 1.8873270371183835, + "grad_norm": 2.6063477993011475, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7256556749343872, + "num_tokens": 437132056.0, + "step": 17186 + }, + { + "epoch": 1.8874368548209972, + "grad_norm": 2.2378463745117188, + "learning_rate": 1e-06, + "loss": 0.876, + "mean_token_accuracy": 0.7193378210067749, + "num_tokens": 437157693.0, + "step": 17187 + }, + { + "epoch": 1.887546672523611, + "grad_norm": 2.1043992042541504, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.7299994230270386, + "num_tokens": 437184700.0, + "step": 17188 + }, + { + "epoch": 1.8876564902262245, + "grad_norm": 2.474318265914917, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7368102669715881, + "num_tokens": 437206269.0, + "step": 17189 + }, + { + "epoch": 1.887766307928838, + "grad_norm": 2.3861846923828125, + "learning_rate": 1e-06, + "loss": 0.9938, + "mean_token_accuracy": 0.6882554888725281, + "num_tokens": 437230768.0, + "step": 17190 + }, + { + "epoch": 1.8878761256314518, + "grad_norm": 2.2241177558898926, + "learning_rate": 1e-06, + "loss": 0.8306, + "mean_token_accuracy": 0.730435311794281, + "num_tokens": 437257442.0, + "step": 17191 + }, + { + "epoch": 1.8879859433340656, + "grad_norm": 1.9543870687484741, + "learning_rate": 1e-06, + "loss": 0.907, + "mean_token_accuracy": 0.7158803939819336, + "num_tokens": 437289906.0, + "step": 17192 + }, + { + "epoch": 1.888095761036679, + "grad_norm": 2.2928385734558105, + "learning_rate": 1e-06, + "loss": 0.8981, + "mean_token_accuracy": 0.7333744764328003, + "num_tokens": 437314812.0, + "step": 17193 + }, + { + "epoch": 1.8882055787392926, + "grad_norm": 2.1249728202819824, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7196921706199646, + "num_tokens": 437342486.0, + "step": 17194 + }, + { + "epoch": 1.8883153964419064, + "grad_norm": 2.2622463703155518, + "learning_rate": 1e-06, + "loss": 0.9997, + "mean_token_accuracy": 0.711334228515625, + "num_tokens": 437367133.0, + "step": 17195 + }, + { + "epoch": 1.8884252141445201, + "grad_norm": 2.1189029216766357, + "learning_rate": 1e-06, + "loss": 0.8534, + "mean_token_accuracy": 0.7281606793403625, + "num_tokens": 437393483.0, + "step": 17196 + }, + { + "epoch": 1.888535031847134, + "grad_norm": 2.4483985900878906, + "learning_rate": 1e-06, + "loss": 0.8448, + "mean_token_accuracy": 0.7333995699882507, + "num_tokens": 437413566.0, + "step": 17197 + }, + { + "epoch": 1.8886448495497474, + "grad_norm": 2.3074965476989746, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7103466987609863, + "num_tokens": 437438516.0, + "step": 17198 + }, + { + "epoch": 1.888754667252361, + "grad_norm": 2.291922092437744, + "learning_rate": 1e-06, + "loss": 0.9994, + "mean_token_accuracy": 0.6903669238090515, + "num_tokens": 437467968.0, + "step": 17199 + }, + { + "epoch": 1.8888644849549747, + "grad_norm": 2.0970640182495117, + "learning_rate": 1e-06, + "loss": 0.935, + "mean_token_accuracy": 0.7186750173568726, + "num_tokens": 437496289.0, + "step": 17200 + }, + { + "epoch": 1.8889743026575885, + "grad_norm": 2.3997747898101807, + "learning_rate": 1e-06, + "loss": 0.9238, + "mean_token_accuracy": 0.7104975581169128, + "num_tokens": 437520001.0, + "step": 17201 + }, + { + "epoch": 1.889084120360202, + "grad_norm": 2.420757293701172, + "learning_rate": 1e-06, + "loss": 0.8315, + "mean_token_accuracy": 0.7411549091339111, + "num_tokens": 437540946.0, + "step": 17202 + }, + { + "epoch": 1.8891939380628158, + "grad_norm": 2.1312172412872314, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6908245086669922, + "num_tokens": 437570225.0, + "step": 17203 + }, + { + "epoch": 1.8893037557654293, + "grad_norm": 2.3281407356262207, + "learning_rate": 1e-06, + "loss": 0.8295, + "mean_token_accuracy": 0.7397341728210449, + "num_tokens": 437593255.0, + "step": 17204 + }, + { + "epoch": 1.889413573468043, + "grad_norm": 2.0804667472839355, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7158524394035339, + "num_tokens": 437624303.0, + "step": 17205 + }, + { + "epoch": 1.8895233911706568, + "grad_norm": 2.538053035736084, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7105297446250916, + "num_tokens": 437645615.0, + "step": 17206 + }, + { + "epoch": 1.8896332088732704, + "grad_norm": 2.118870496749878, + "learning_rate": 1e-06, + "loss": 0.8107, + "mean_token_accuracy": 0.7418080568313599, + "num_tokens": 437671102.0, + "step": 17207 + }, + { + "epoch": 1.8897430265758839, + "grad_norm": 2.251244306564331, + "learning_rate": 1e-06, + "loss": 0.8628, + "mean_token_accuracy": 0.7288833260536194, + "num_tokens": 437696719.0, + "step": 17208 + }, + { + "epoch": 1.8898528442784976, + "grad_norm": 2.375452756881714, + "learning_rate": 1e-06, + "loss": 0.8139, + "mean_token_accuracy": 0.7420674562454224, + "num_tokens": 437721814.0, + "step": 17209 + }, + { + "epoch": 1.8899626619811114, + "grad_norm": 1.9362376928329468, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7078461050987244, + "num_tokens": 437753844.0, + "step": 17210 + }, + { + "epoch": 1.8900724796837252, + "grad_norm": 2.379077672958374, + "learning_rate": 1e-06, + "loss": 0.9034, + "mean_token_accuracy": 0.7260031700134277, + "num_tokens": 437778744.0, + "step": 17211 + }, + { + "epoch": 1.8901822973863387, + "grad_norm": 2.15423583984375, + "learning_rate": 1e-06, + "loss": 0.8884, + "mean_token_accuracy": 0.7340834140777588, + "num_tokens": 437804487.0, + "step": 17212 + }, + { + "epoch": 1.8902921150889522, + "grad_norm": 2.4296181201934814, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7400678396224976, + "num_tokens": 437825484.0, + "step": 17213 + }, + { + "epoch": 1.890401932791566, + "grad_norm": 2.5438718795776367, + "learning_rate": 1e-06, + "loss": 0.851, + "mean_token_accuracy": 0.7318617105484009, + "num_tokens": 437846935.0, + "step": 17214 + }, + { + "epoch": 1.8905117504941797, + "grad_norm": 2.352145195007324, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7181082963943481, + "num_tokens": 437869755.0, + "step": 17215 + }, + { + "epoch": 1.8906215681967933, + "grad_norm": 2.3752999305725098, + "learning_rate": 1e-06, + "loss": 0.8143, + "mean_token_accuracy": 0.7380601763725281, + "num_tokens": 437892156.0, + "step": 17216 + }, + { + "epoch": 1.890731385899407, + "grad_norm": 2.3434269428253174, + "learning_rate": 1e-06, + "loss": 0.8809, + "mean_token_accuracy": 0.721036434173584, + "num_tokens": 437917462.0, + "step": 17217 + }, + { + "epoch": 1.8908412036020206, + "grad_norm": 2.3722188472747803, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7298097610473633, + "num_tokens": 437938929.0, + "step": 17218 + }, + { + "epoch": 1.8909510213046343, + "grad_norm": 2.248793363571167, + "learning_rate": 1e-06, + "loss": 0.8861, + "mean_token_accuracy": 0.7230619788169861, + "num_tokens": 437962546.0, + "step": 17219 + }, + { + "epoch": 1.891060839007248, + "grad_norm": 2.225245714187622, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7235250473022461, + "num_tokens": 437987447.0, + "step": 17220 + }, + { + "epoch": 1.8911706567098616, + "grad_norm": 2.1138417720794678, + "learning_rate": 1e-06, + "loss": 0.9115, + "mean_token_accuracy": 0.7210697531700134, + "num_tokens": 438014780.0, + "step": 17221 + }, + { + "epoch": 1.8912804744124752, + "grad_norm": 2.2166504859924316, + "learning_rate": 1e-06, + "loss": 0.8679, + "mean_token_accuracy": 0.7279698848724365, + "num_tokens": 438038968.0, + "step": 17222 + }, + { + "epoch": 1.891390292115089, + "grad_norm": 2.294403076171875, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7234602570533752, + "num_tokens": 438063972.0, + "step": 17223 + }, + { + "epoch": 1.8915001098177027, + "grad_norm": 2.043760299682617, + "learning_rate": 1e-06, + "loss": 0.9629, + "mean_token_accuracy": 0.6988970041275024, + "num_tokens": 438092498.0, + "step": 17224 + }, + { + "epoch": 1.8916099275203164, + "grad_norm": 2.2712481021881104, + "learning_rate": 1e-06, + "loss": 0.9547, + "mean_token_accuracy": 0.7086953520774841, + "num_tokens": 438120923.0, + "step": 17225 + }, + { + "epoch": 1.89171974522293, + "grad_norm": 2.406442165374756, + "learning_rate": 1e-06, + "loss": 0.9232, + "mean_token_accuracy": 0.7168552875518799, + "num_tokens": 438143451.0, + "step": 17226 + }, + { + "epoch": 1.8918295629255435, + "grad_norm": 2.1492559909820557, + "learning_rate": 1e-06, + "loss": 0.9445, + "mean_token_accuracy": 0.7075605392456055, + "num_tokens": 438170863.0, + "step": 17227 + }, + { + "epoch": 1.8919393806281573, + "grad_norm": 2.2895870208740234, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7136862277984619, + "num_tokens": 438194024.0, + "step": 17228 + }, + { + "epoch": 1.892049198330771, + "grad_norm": 2.1896004676818848, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7128100395202637, + "num_tokens": 438219372.0, + "step": 17229 + }, + { + "epoch": 1.8921590160333845, + "grad_norm": 2.4273738861083984, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7343369722366333, + "num_tokens": 438241245.0, + "step": 17230 + }, + { + "epoch": 1.892268833735998, + "grad_norm": 2.5343551635742188, + "learning_rate": 1e-06, + "loss": 0.8637, + "mean_token_accuracy": 0.7264692783355713, + "num_tokens": 438262316.0, + "step": 17231 + }, + { + "epoch": 1.8923786514386118, + "grad_norm": 2.049720287322998, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7240474820137024, + "num_tokens": 438291732.0, + "step": 17232 + }, + { + "epoch": 1.8924884691412256, + "grad_norm": 2.2114994525909424, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7409438490867615, + "num_tokens": 438315554.0, + "step": 17233 + }, + { + "epoch": 1.8925982868438394, + "grad_norm": 2.0627899169921875, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7473734617233276, + "num_tokens": 438342121.0, + "step": 17234 + }, + { + "epoch": 1.8927081045464529, + "grad_norm": 2.165982484817505, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7219932675361633, + "num_tokens": 438371593.0, + "step": 17235 + }, + { + "epoch": 1.8928179222490664, + "grad_norm": 2.186711072921753, + "learning_rate": 1e-06, + "loss": 0.8074, + "mean_token_accuracy": 0.7404068112373352, + "num_tokens": 438395984.0, + "step": 17236 + }, + { + "epoch": 1.8929277399516802, + "grad_norm": 2.1558406352996826, + "learning_rate": 1e-06, + "loss": 0.8225, + "mean_token_accuracy": 0.7389386892318726, + "num_tokens": 438424288.0, + "step": 17237 + }, + { + "epoch": 1.893037557654294, + "grad_norm": 2.246358633041382, + "learning_rate": 1e-06, + "loss": 0.9358, + "mean_token_accuracy": 0.7138690948486328, + "num_tokens": 438450026.0, + "step": 17238 + }, + { + "epoch": 1.8931473753569077, + "grad_norm": 2.318054437637329, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7260500192642212, + "num_tokens": 438474576.0, + "step": 17239 + }, + { + "epoch": 1.8932571930595212, + "grad_norm": 2.0050716400146484, + "learning_rate": 1e-06, + "loss": 0.9622, + "mean_token_accuracy": 0.7013283371925354, + "num_tokens": 438506391.0, + "step": 17240 + }, + { + "epoch": 1.8933670107621348, + "grad_norm": 2.250251293182373, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7418920993804932, + "num_tokens": 438529228.0, + "step": 17241 + }, + { + "epoch": 1.8934768284647485, + "grad_norm": 2.215123176574707, + "learning_rate": 1e-06, + "loss": 0.9618, + "mean_token_accuracy": 0.7069758772850037, + "num_tokens": 438557819.0, + "step": 17242 + }, + { + "epoch": 1.8935866461673623, + "grad_norm": 2.431821584701538, + "learning_rate": 1e-06, + "loss": 0.8307, + "mean_token_accuracy": 0.734563946723938, + "num_tokens": 438580546.0, + "step": 17243 + }, + { + "epoch": 1.8936964638699758, + "grad_norm": 2.2007791996002197, + "learning_rate": 1e-06, + "loss": 0.8622, + "mean_token_accuracy": 0.7268222570419312, + "num_tokens": 438604846.0, + "step": 17244 + }, + { + "epoch": 1.8938062815725893, + "grad_norm": 2.377962827682495, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7156611680984497, + "num_tokens": 438628579.0, + "step": 17245 + }, + { + "epoch": 1.893916099275203, + "grad_norm": 2.4066717624664307, + "learning_rate": 1e-06, + "loss": 0.8518, + "mean_token_accuracy": 0.7323179244995117, + "num_tokens": 438651414.0, + "step": 17246 + }, + { + "epoch": 1.8940259169778169, + "grad_norm": 2.1454780101776123, + "learning_rate": 1e-06, + "loss": 1.0055, + "mean_token_accuracy": 0.7010614275932312, + "num_tokens": 438680521.0, + "step": 17247 + }, + { + "epoch": 1.8941357346804306, + "grad_norm": 2.2221908569335938, + "learning_rate": 1e-06, + "loss": 0.923, + "mean_token_accuracy": 0.7129577398300171, + "num_tokens": 438708211.0, + "step": 17248 + }, + { + "epoch": 1.8942455523830442, + "grad_norm": 2.297637939453125, + "learning_rate": 1e-06, + "loss": 0.8177, + "mean_token_accuracy": 0.7378008365631104, + "num_tokens": 438732318.0, + "step": 17249 + }, + { + "epoch": 1.8943553700856577, + "grad_norm": 2.1147539615631104, + "learning_rate": 1e-06, + "loss": 0.9055, + "mean_token_accuracy": 0.7196990251541138, + "num_tokens": 438757946.0, + "step": 17250 + }, + { + "epoch": 1.8944651877882714, + "grad_norm": 2.0583972930908203, + "learning_rate": 1e-06, + "loss": 0.9086, + "mean_token_accuracy": 0.7199821472167969, + "num_tokens": 438787805.0, + "step": 17251 + }, + { + "epoch": 1.8945750054908852, + "grad_norm": 2.102144718170166, + "learning_rate": 1e-06, + "loss": 0.9501, + "mean_token_accuracy": 0.7020972371101379, + "num_tokens": 438817159.0, + "step": 17252 + }, + { + "epoch": 1.894684823193499, + "grad_norm": 2.0535852909088135, + "learning_rate": 1e-06, + "loss": 0.8973, + "mean_token_accuracy": 0.721351146697998, + "num_tokens": 438847008.0, + "step": 17253 + }, + { + "epoch": 1.8947946408961125, + "grad_norm": 2.421391010284424, + "learning_rate": 1e-06, + "loss": 0.7355, + "mean_token_accuracy": 0.7669409513473511, + "num_tokens": 438867761.0, + "step": 17254 + }, + { + "epoch": 1.894904458598726, + "grad_norm": 2.4208147525787354, + "learning_rate": 1e-06, + "loss": 0.9174, + "mean_token_accuracy": 0.7189825177192688, + "num_tokens": 438889677.0, + "step": 17255 + }, + { + "epoch": 1.8950142763013398, + "grad_norm": 2.058359146118164, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7245859503746033, + "num_tokens": 438919254.0, + "step": 17256 + }, + { + "epoch": 1.8951240940039535, + "grad_norm": 2.342716693878174, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7169749736785889, + "num_tokens": 438942492.0, + "step": 17257 + }, + { + "epoch": 1.895233911706567, + "grad_norm": 2.157825231552124, + "learning_rate": 1e-06, + "loss": 0.8817, + "mean_token_accuracy": 0.7232779860496521, + "num_tokens": 438968472.0, + "step": 17258 + }, + { + "epoch": 1.8953437294091806, + "grad_norm": 2.389356851577759, + "learning_rate": 1e-06, + "loss": 0.8762, + "mean_token_accuracy": 0.7226974964141846, + "num_tokens": 438991554.0, + "step": 17259 + }, + { + "epoch": 1.8954535471117944, + "grad_norm": 2.242371082305908, + "learning_rate": 1e-06, + "loss": 0.849, + "mean_token_accuracy": 0.7288423776626587, + "num_tokens": 439016779.0, + "step": 17260 + }, + { + "epoch": 1.8955633648144081, + "grad_norm": 2.1729111671447754, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7087362408638, + "num_tokens": 439044634.0, + "step": 17261 + }, + { + "epoch": 1.8956731825170219, + "grad_norm": 2.3434300422668457, + "learning_rate": 1e-06, + "loss": 0.8171, + "mean_token_accuracy": 0.7413844466209412, + "num_tokens": 439065635.0, + "step": 17262 + }, + { + "epoch": 1.8957830002196354, + "grad_norm": 2.2556333541870117, + "learning_rate": 1e-06, + "loss": 1.0137, + "mean_token_accuracy": 0.6963433027267456, + "num_tokens": 439090058.0, + "step": 17263 + }, + { + "epoch": 1.895892817922249, + "grad_norm": 2.371544361114502, + "learning_rate": 1e-06, + "loss": 0.9126, + "mean_token_accuracy": 0.7137382626533508, + "num_tokens": 439114368.0, + "step": 17264 + }, + { + "epoch": 1.8960026356248627, + "grad_norm": 2.11407732963562, + "learning_rate": 1e-06, + "loss": 0.9489, + "mean_token_accuracy": 0.7204881906509399, + "num_tokens": 439142095.0, + "step": 17265 + }, + { + "epoch": 1.8961124533274765, + "grad_norm": 2.029970169067383, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.7111159563064575, + "num_tokens": 439171003.0, + "step": 17266 + }, + { + "epoch": 1.89622227103009, + "grad_norm": 2.362534523010254, + "learning_rate": 1e-06, + "loss": 0.891, + "mean_token_accuracy": 0.7181777954101562, + "num_tokens": 439193548.0, + "step": 17267 + }, + { + "epoch": 1.8963320887327038, + "grad_norm": 2.1997127532958984, + "learning_rate": 1e-06, + "loss": 0.879, + "mean_token_accuracy": 0.7218472361564636, + "num_tokens": 439219091.0, + "step": 17268 + }, + { + "epoch": 1.8964419064353173, + "grad_norm": 2.425513982772827, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.7068244814872742, + "num_tokens": 439243730.0, + "step": 17269 + }, + { + "epoch": 1.896551724137931, + "grad_norm": 2.047501564025879, + "learning_rate": 1e-06, + "loss": 0.8216, + "mean_token_accuracy": 0.738471508026123, + "num_tokens": 439272418.0, + "step": 17270 + }, + { + "epoch": 1.8966615418405448, + "grad_norm": 2.3591713905334473, + "learning_rate": 1e-06, + "loss": 0.9675, + "mean_token_accuracy": 0.700407862663269, + "num_tokens": 439296686.0, + "step": 17271 + }, + { + "epoch": 1.8967713595431583, + "grad_norm": 2.298205852508545, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.728498101234436, + "num_tokens": 439320698.0, + "step": 17272 + }, + { + "epoch": 1.8968811772457719, + "grad_norm": 2.1590540409088135, + "learning_rate": 1e-06, + "loss": 0.9065, + "mean_token_accuracy": 0.7270252108573914, + "num_tokens": 439347213.0, + "step": 17273 + }, + { + "epoch": 1.8969909949483856, + "grad_norm": 2.3019604682922363, + "learning_rate": 1e-06, + "loss": 0.8634, + "mean_token_accuracy": 0.729962944984436, + "num_tokens": 439371259.0, + "step": 17274 + }, + { + "epoch": 1.8971008126509994, + "grad_norm": 2.5657002925872803, + "learning_rate": 1e-06, + "loss": 0.9136, + "mean_token_accuracy": 0.7212401628494263, + "num_tokens": 439393266.0, + "step": 17275 + }, + { + "epoch": 1.8972106303536131, + "grad_norm": 2.487323045730591, + "learning_rate": 1e-06, + "loss": 0.9165, + "mean_token_accuracy": 0.7147775888442993, + "num_tokens": 439415693.0, + "step": 17276 + }, + { + "epoch": 1.8973204480562267, + "grad_norm": 2.5888352394104004, + "learning_rate": 1e-06, + "loss": 0.9281, + "mean_token_accuracy": 0.7144003510475159, + "num_tokens": 439436950.0, + "step": 17277 + }, + { + "epoch": 1.8974302657588402, + "grad_norm": 1.98691725730896, + "learning_rate": 1e-06, + "loss": 0.9632, + "mean_token_accuracy": 0.699865996837616, + "num_tokens": 439467401.0, + "step": 17278 + }, + { + "epoch": 1.897540083461454, + "grad_norm": 2.2272934913635254, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7338371872901917, + "num_tokens": 439493059.0, + "step": 17279 + }, + { + "epoch": 1.8976499011640677, + "grad_norm": 2.0803699493408203, + "learning_rate": 1e-06, + "loss": 0.8046, + "mean_token_accuracy": 0.744236946105957, + "num_tokens": 439519483.0, + "step": 17280 + }, + { + "epoch": 1.8977597188666813, + "grad_norm": 2.5571165084838867, + "learning_rate": 1e-06, + "loss": 0.8402, + "mean_token_accuracy": 0.7341822385787964, + "num_tokens": 439538987.0, + "step": 17281 + }, + { + "epoch": 1.897869536569295, + "grad_norm": 2.2643802165985107, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.718837559223175, + "num_tokens": 439564757.0, + "step": 17282 + }, + { + "epoch": 1.8979793542719086, + "grad_norm": 2.2916793823242188, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7159367799758911, + "num_tokens": 439589343.0, + "step": 17283 + }, + { + "epoch": 1.8980891719745223, + "grad_norm": 2.2238872051239014, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.7104187607765198, + "num_tokens": 439617604.0, + "step": 17284 + }, + { + "epoch": 1.898198989677136, + "grad_norm": 2.264601230621338, + "learning_rate": 1e-06, + "loss": 0.8933, + "mean_token_accuracy": 0.7146902680397034, + "num_tokens": 439642641.0, + "step": 17285 + }, + { + "epoch": 1.8983088073797496, + "grad_norm": 2.1254074573516846, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7270397543907166, + "num_tokens": 439670515.0, + "step": 17286 + }, + { + "epoch": 1.8984186250823631, + "grad_norm": 2.070711851119995, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7339446544647217, + "num_tokens": 439699017.0, + "step": 17287 + }, + { + "epoch": 1.898528442784977, + "grad_norm": 2.3253133296966553, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7220438122749329, + "num_tokens": 439724751.0, + "step": 17288 + }, + { + "epoch": 1.8986382604875907, + "grad_norm": 2.051305055618286, + "learning_rate": 1e-06, + "loss": 0.9284, + "mean_token_accuracy": 0.7212883234024048, + "num_tokens": 439756013.0, + "step": 17289 + }, + { + "epoch": 1.8987480781902044, + "grad_norm": 2.1302080154418945, + "learning_rate": 1e-06, + "loss": 0.9389, + "mean_token_accuracy": 0.7068273425102234, + "num_tokens": 439786352.0, + "step": 17290 + }, + { + "epoch": 1.898857895892818, + "grad_norm": 2.436708927154541, + "learning_rate": 1e-06, + "loss": 0.8183, + "mean_token_accuracy": 0.7417185306549072, + "num_tokens": 439808728.0, + "step": 17291 + }, + { + "epoch": 1.8989677135954315, + "grad_norm": 2.4886624813079834, + "learning_rate": 1e-06, + "loss": 0.8211, + "mean_token_accuracy": 0.7361112833023071, + "num_tokens": 439829259.0, + "step": 17292 + }, + { + "epoch": 1.8990775312980452, + "grad_norm": 2.1849758625030518, + "learning_rate": 1e-06, + "loss": 0.9357, + "mean_token_accuracy": 0.7115532159805298, + "num_tokens": 439856937.0, + "step": 17293 + }, + { + "epoch": 1.899187349000659, + "grad_norm": 2.809915065765381, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7323142290115356, + "num_tokens": 439875275.0, + "step": 17294 + }, + { + "epoch": 1.8992971667032725, + "grad_norm": 2.069856643676758, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7162476778030396, + "num_tokens": 439906448.0, + "step": 17295 + }, + { + "epoch": 1.899406984405886, + "grad_norm": 2.4884378910064697, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7233822345733643, + "num_tokens": 439928447.0, + "step": 17296 + }, + { + "epoch": 1.8995168021084998, + "grad_norm": 2.6718955039978027, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7420392632484436, + "num_tokens": 439946701.0, + "step": 17297 + }, + { + "epoch": 1.8996266198111136, + "grad_norm": 2.594571113586426, + "learning_rate": 1e-06, + "loss": 0.837, + "mean_token_accuracy": 0.7339186072349548, + "num_tokens": 439965447.0, + "step": 17298 + }, + { + "epoch": 1.8997364375137273, + "grad_norm": 2.8225321769714355, + "learning_rate": 1e-06, + "loss": 0.8886, + "mean_token_accuracy": 0.7231912016868591, + "num_tokens": 439983393.0, + "step": 17299 + }, + { + "epoch": 1.8998462552163409, + "grad_norm": 2.094672679901123, + "learning_rate": 1e-06, + "loss": 0.9782, + "mean_token_accuracy": 0.6999791860580444, + "num_tokens": 440013430.0, + "step": 17300 + }, + { + "epoch": 1.8999560729189544, + "grad_norm": 2.300232172012329, + "learning_rate": 1e-06, + "loss": 0.9063, + "mean_token_accuracy": 0.7176696062088013, + "num_tokens": 440037364.0, + "step": 17301 + }, + { + "epoch": 1.9000658906215682, + "grad_norm": 2.317838430404663, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7097647190093994, + "num_tokens": 440061580.0, + "step": 17302 + }, + { + "epoch": 1.900175708324182, + "grad_norm": 2.0820391178131104, + "learning_rate": 1e-06, + "loss": 0.8714, + "mean_token_accuracy": 0.7221357226371765, + "num_tokens": 440090709.0, + "step": 17303 + }, + { + "epoch": 1.9002855260267957, + "grad_norm": 2.1289918422698975, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7148884534835815, + "num_tokens": 440118291.0, + "step": 17304 + }, + { + "epoch": 1.9003953437294092, + "grad_norm": 2.205183506011963, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7247928380966187, + "num_tokens": 440143864.0, + "step": 17305 + }, + { + "epoch": 1.9005051614320227, + "grad_norm": 2.368482828140259, + "learning_rate": 1e-06, + "loss": 0.915, + "mean_token_accuracy": 0.7102293968200684, + "num_tokens": 440167745.0, + "step": 17306 + }, + { + "epoch": 1.9006149791346365, + "grad_norm": 2.1994218826293945, + "learning_rate": 1e-06, + "loss": 0.8024, + "mean_token_accuracy": 0.7486166954040527, + "num_tokens": 440191035.0, + "step": 17307 + }, + { + "epoch": 1.9007247968372503, + "grad_norm": 2.1890830993652344, + "learning_rate": 1e-06, + "loss": 0.8051, + "mean_token_accuracy": 0.743629515171051, + "num_tokens": 440216004.0, + "step": 17308 + }, + { + "epoch": 1.9008346145398638, + "grad_norm": 2.219557046890259, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7069317698478699, + "num_tokens": 440243243.0, + "step": 17309 + }, + { + "epoch": 1.9009444322424773, + "grad_norm": 2.1230111122131348, + "learning_rate": 1e-06, + "loss": 0.896, + "mean_token_accuracy": 0.7285618185997009, + "num_tokens": 440271239.0, + "step": 17310 + }, + { + "epoch": 1.901054249945091, + "grad_norm": 2.125755786895752, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7075752019882202, + "num_tokens": 440299843.0, + "step": 17311 + }, + { + "epoch": 1.9011640676477048, + "grad_norm": 2.213503837585449, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7390384078025818, + "num_tokens": 440325034.0, + "step": 17312 + }, + { + "epoch": 1.9012738853503186, + "grad_norm": 2.3716626167297363, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7177583575248718, + "num_tokens": 440347854.0, + "step": 17313 + }, + { + "epoch": 1.9013837030529321, + "grad_norm": 2.174457550048828, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.7158830165863037, + "num_tokens": 440376376.0, + "step": 17314 + }, + { + "epoch": 1.9014935207555457, + "grad_norm": 2.5884788036346436, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.733938992023468, + "num_tokens": 440398398.0, + "step": 17315 + }, + { + "epoch": 1.9016033384581594, + "grad_norm": 2.592390537261963, + "learning_rate": 1e-06, + "loss": 0.7931, + "mean_token_accuracy": 0.7448521852493286, + "num_tokens": 440417534.0, + "step": 17316 + }, + { + "epoch": 1.9017131561607732, + "grad_norm": 2.132323980331421, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7178112268447876, + "num_tokens": 440445575.0, + "step": 17317 + }, + { + "epoch": 1.9018229738633867, + "grad_norm": 2.336947202682495, + "learning_rate": 1e-06, + "loss": 0.9416, + "mean_token_accuracy": 0.7141415476799011, + "num_tokens": 440468881.0, + "step": 17318 + }, + { + "epoch": 1.9019327915660005, + "grad_norm": 2.446197271347046, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7299634218215942, + "num_tokens": 440490441.0, + "step": 17319 + }, + { + "epoch": 1.902042609268614, + "grad_norm": 2.0458035469055176, + "learning_rate": 1e-06, + "loss": 0.9029, + "mean_token_accuracy": 0.7236658334732056, + "num_tokens": 440519102.0, + "step": 17320 + }, + { + "epoch": 1.9021524269712278, + "grad_norm": 2.3555877208709717, + "learning_rate": 1e-06, + "loss": 0.9016, + "mean_token_accuracy": 0.7271710634231567, + "num_tokens": 440542004.0, + "step": 17321 + }, + { + "epoch": 1.9022622446738415, + "grad_norm": 2.417637825012207, + "learning_rate": 1e-06, + "loss": 0.9328, + "mean_token_accuracy": 0.7207615375518799, + "num_tokens": 440565130.0, + "step": 17322 + }, + { + "epoch": 1.902372062376455, + "grad_norm": 2.0337390899658203, + "learning_rate": 1e-06, + "loss": 0.9871, + "mean_token_accuracy": 0.7010011076927185, + "num_tokens": 440596143.0, + "step": 17323 + }, + { + "epoch": 1.9024818800790686, + "grad_norm": 2.27638578414917, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7350260019302368, + "num_tokens": 440620949.0, + "step": 17324 + }, + { + "epoch": 1.9025916977816824, + "grad_norm": 2.2356526851654053, + "learning_rate": 1e-06, + "loss": 0.8507, + "mean_token_accuracy": 0.7425536513328552, + "num_tokens": 440644648.0, + "step": 17325 + }, + { + "epoch": 1.9027015154842961, + "grad_norm": 2.4201443195343018, + "learning_rate": 1e-06, + "loss": 0.9146, + "mean_token_accuracy": 0.7181358337402344, + "num_tokens": 440666990.0, + "step": 17326 + }, + { + "epoch": 1.9028113331869099, + "grad_norm": 2.058342933654785, + "learning_rate": 1e-06, + "loss": 0.9774, + "mean_token_accuracy": 0.7131816744804382, + "num_tokens": 440696320.0, + "step": 17327 + }, + { + "epoch": 1.9029211508895234, + "grad_norm": 2.159135341644287, + "learning_rate": 1e-06, + "loss": 0.9028, + "mean_token_accuracy": 0.7187501192092896, + "num_tokens": 440723079.0, + "step": 17328 + }, + { + "epoch": 1.903030968592137, + "grad_norm": 2.6591992378234863, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7032199501991272, + "num_tokens": 440745457.0, + "step": 17329 + }, + { + "epoch": 1.9031407862947507, + "grad_norm": 2.104391098022461, + "learning_rate": 1e-06, + "loss": 0.8735, + "mean_token_accuracy": 0.7329793572425842, + "num_tokens": 440771288.0, + "step": 17330 + }, + { + "epoch": 1.9032506039973645, + "grad_norm": 2.0044405460357666, + "learning_rate": 1e-06, + "loss": 0.9379, + "mean_token_accuracy": 0.7122797966003418, + "num_tokens": 440803837.0, + "step": 17331 + }, + { + "epoch": 1.903360421699978, + "grad_norm": 2.2476251125335693, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7191733717918396, + "num_tokens": 440828867.0, + "step": 17332 + }, + { + "epoch": 1.9034702394025917, + "grad_norm": 2.0921146869659424, + "learning_rate": 1e-06, + "loss": 0.9806, + "mean_token_accuracy": 0.6954020857810974, + "num_tokens": 440858167.0, + "step": 17333 + }, + { + "epoch": 1.9035800571052053, + "grad_norm": 2.141922950744629, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7259608507156372, + "num_tokens": 440887204.0, + "step": 17334 + }, + { + "epoch": 1.903689874807819, + "grad_norm": 2.2062015533447266, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7203535437583923, + "num_tokens": 440911922.0, + "step": 17335 + }, + { + "epoch": 1.9037996925104328, + "grad_norm": 2.1905617713928223, + "learning_rate": 1e-06, + "loss": 0.94, + "mean_token_accuracy": 0.703693151473999, + "num_tokens": 440937099.0, + "step": 17336 + }, + { + "epoch": 1.9039095102130463, + "grad_norm": 2.194079637527466, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7219321727752686, + "num_tokens": 440964172.0, + "step": 17337 + }, + { + "epoch": 1.9040193279156599, + "grad_norm": 2.4478671550750732, + "learning_rate": 1e-06, + "loss": 0.8206, + "mean_token_accuracy": 0.7440477609634399, + "num_tokens": 440986132.0, + "step": 17338 + }, + { + "epoch": 1.9041291456182736, + "grad_norm": 1.9252077341079712, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.7073239088058472, + "num_tokens": 441022043.0, + "step": 17339 + }, + { + "epoch": 1.9042389633208874, + "grad_norm": 2.3725364208221436, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.723770022392273, + "num_tokens": 441045390.0, + "step": 17340 + }, + { + "epoch": 1.9043487810235011, + "grad_norm": 2.1220173835754395, + "learning_rate": 1e-06, + "loss": 0.9236, + "mean_token_accuracy": 0.7091500759124756, + "num_tokens": 441072965.0, + "step": 17341 + }, + { + "epoch": 1.9044585987261147, + "grad_norm": 1.866014003753662, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7362678647041321, + "num_tokens": 441108551.0, + "step": 17342 + }, + { + "epoch": 1.9045684164287282, + "grad_norm": 2.2338523864746094, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.7241469025611877, + "num_tokens": 441132999.0, + "step": 17343 + }, + { + "epoch": 1.904678234131342, + "grad_norm": 2.0852272510528564, + "learning_rate": 1e-06, + "loss": 0.8589, + "mean_token_accuracy": 0.7273896336555481, + "num_tokens": 441158852.0, + "step": 17344 + }, + { + "epoch": 1.9047880518339557, + "grad_norm": 2.186583995819092, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.73405921459198, + "num_tokens": 441184396.0, + "step": 17345 + }, + { + "epoch": 1.9048978695365693, + "grad_norm": 2.1513969898223877, + "learning_rate": 1e-06, + "loss": 0.9355, + "mean_token_accuracy": 0.719028115272522, + "num_tokens": 441211604.0, + "step": 17346 + }, + { + "epoch": 1.9050076872391828, + "grad_norm": 2.418073892593384, + "learning_rate": 1e-06, + "loss": 0.8264, + "mean_token_accuracy": 0.736004114151001, + "num_tokens": 441232667.0, + "step": 17347 + }, + { + "epoch": 1.9051175049417965, + "grad_norm": 2.652693748474121, + "learning_rate": 1e-06, + "loss": 0.9156, + "mean_token_accuracy": 0.7188310623168945, + "num_tokens": 441253163.0, + "step": 17348 + }, + { + "epoch": 1.9052273226444103, + "grad_norm": 2.2663843631744385, + "learning_rate": 1e-06, + "loss": 0.8994, + "mean_token_accuracy": 0.7177824974060059, + "num_tokens": 441278076.0, + "step": 17349 + }, + { + "epoch": 1.905337140347024, + "grad_norm": 2.2464182376861572, + "learning_rate": 1e-06, + "loss": 0.8938, + "mean_token_accuracy": 0.7190909385681152, + "num_tokens": 441303189.0, + "step": 17350 + }, + { + "epoch": 1.9054469580496376, + "grad_norm": 2.483051061630249, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.7304829359054565, + "num_tokens": 441323729.0, + "step": 17351 + }, + { + "epoch": 1.9055567757522511, + "grad_norm": 2.058274745941162, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.725175678730011, + "num_tokens": 441354097.0, + "step": 17352 + }, + { + "epoch": 1.9056665934548649, + "grad_norm": 2.275254726409912, + "learning_rate": 1e-06, + "loss": 0.9006, + "mean_token_accuracy": 0.7207244038581848, + "num_tokens": 441380138.0, + "step": 17353 + }, + { + "epoch": 1.9057764111574786, + "grad_norm": 2.371965169906616, + "learning_rate": 1e-06, + "loss": 0.8718, + "mean_token_accuracy": 0.7269785404205322, + "num_tokens": 441402905.0, + "step": 17354 + }, + { + "epoch": 1.9058862288600924, + "grad_norm": 1.993319034576416, + "learning_rate": 1e-06, + "loss": 1.0208, + "mean_token_accuracy": 0.6858320236206055, + "num_tokens": 441434984.0, + "step": 17355 + }, + { + "epoch": 1.905996046562706, + "grad_norm": 2.5373504161834717, + "learning_rate": 1e-06, + "loss": 0.7859, + "mean_token_accuracy": 0.7480468153953552, + "num_tokens": 441455180.0, + "step": 17356 + }, + { + "epoch": 1.9061058642653195, + "grad_norm": 2.3914170265197754, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7401562929153442, + "num_tokens": 441476191.0, + "step": 17357 + }, + { + "epoch": 1.9062156819679332, + "grad_norm": 2.228327989578247, + "learning_rate": 1e-06, + "loss": 0.9977, + "mean_token_accuracy": 0.7007311582565308, + "num_tokens": 441503560.0, + "step": 17358 + }, + { + "epoch": 1.906325499670547, + "grad_norm": 2.2229440212249756, + "learning_rate": 1e-06, + "loss": 0.8397, + "mean_token_accuracy": 0.7388007640838623, + "num_tokens": 441527109.0, + "step": 17359 + }, + { + "epoch": 1.9064353173731605, + "grad_norm": 2.1601622104644775, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7273080348968506, + "num_tokens": 441553766.0, + "step": 17360 + }, + { + "epoch": 1.906545135075774, + "grad_norm": 2.1983840465545654, + "learning_rate": 1e-06, + "loss": 0.812, + "mean_token_accuracy": 0.7411666512489319, + "num_tokens": 441579636.0, + "step": 17361 + }, + { + "epoch": 1.9066549527783878, + "grad_norm": 2.347935914993286, + "learning_rate": 1e-06, + "loss": 0.946, + "mean_token_accuracy": 0.721381664276123, + "num_tokens": 441605121.0, + "step": 17362 + }, + { + "epoch": 1.9067647704810016, + "grad_norm": 2.5123932361602783, + "learning_rate": 1e-06, + "loss": 0.9203, + "mean_token_accuracy": 0.7259074449539185, + "num_tokens": 441627015.0, + "step": 17363 + }, + { + "epoch": 1.9068745881836153, + "grad_norm": 2.38608455657959, + "learning_rate": 1e-06, + "loss": 0.8346, + "mean_token_accuracy": 0.7330390810966492, + "num_tokens": 441648880.0, + "step": 17364 + }, + { + "epoch": 1.9069844058862289, + "grad_norm": 2.296675205230713, + "learning_rate": 1e-06, + "loss": 0.9287, + "mean_token_accuracy": 0.7090743184089661, + "num_tokens": 441674042.0, + "step": 17365 + }, + { + "epoch": 1.9070942235888424, + "grad_norm": 2.1700785160064697, + "learning_rate": 1e-06, + "loss": 0.8472, + "mean_token_accuracy": 0.7478775382041931, + "num_tokens": 441698274.0, + "step": 17366 + }, + { + "epoch": 1.9072040412914562, + "grad_norm": 2.1387553215026855, + "learning_rate": 1e-06, + "loss": 0.9466, + "mean_token_accuracy": 0.7025454044342041, + "num_tokens": 441727732.0, + "step": 17367 + }, + { + "epoch": 1.90731385899407, + "grad_norm": 2.1904237270355225, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.7216395139694214, + "num_tokens": 441755068.0, + "step": 17368 + }, + { + "epoch": 1.9074236766966837, + "grad_norm": 2.2139225006103516, + "learning_rate": 1e-06, + "loss": 0.8089, + "mean_token_accuracy": 0.7390746474266052, + "num_tokens": 441779299.0, + "step": 17369 + }, + { + "epoch": 1.9075334943992972, + "grad_norm": 2.126845121383667, + "learning_rate": 1e-06, + "loss": 0.9319, + "mean_token_accuracy": 0.7059522867202759, + "num_tokens": 441807164.0, + "step": 17370 + }, + { + "epoch": 1.9076433121019107, + "grad_norm": 1.918338656425476, + "learning_rate": 1e-06, + "loss": 0.9253, + "mean_token_accuracy": 0.7219862341880798, + "num_tokens": 441842654.0, + "step": 17371 + }, + { + "epoch": 1.9077531298045245, + "grad_norm": 2.0869972705841064, + "learning_rate": 1e-06, + "loss": 1.0445, + "mean_token_accuracy": 0.6865274310112, + "num_tokens": 441873226.0, + "step": 17372 + }, + { + "epoch": 1.9078629475071383, + "grad_norm": 2.134188413619995, + "learning_rate": 1e-06, + "loss": 0.9649, + "mean_token_accuracy": 0.7011851072311401, + "num_tokens": 441900208.0, + "step": 17373 + }, + { + "epoch": 1.9079727652097518, + "grad_norm": 2.210055351257324, + "learning_rate": 1e-06, + "loss": 0.8453, + "mean_token_accuracy": 0.7335985898971558, + "num_tokens": 441924237.0, + "step": 17374 + }, + { + "epoch": 1.9080825829123653, + "grad_norm": 2.3860690593719482, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7193797826766968, + "num_tokens": 441947211.0, + "step": 17375 + }, + { + "epoch": 1.908192400614979, + "grad_norm": 2.1489460468292236, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7309894561767578, + "num_tokens": 441972405.0, + "step": 17376 + }, + { + "epoch": 1.9083022183175928, + "grad_norm": 2.208780527114868, + "learning_rate": 1e-06, + "loss": 0.9191, + "mean_token_accuracy": 0.7170233726501465, + "num_tokens": 441998771.0, + "step": 17377 + }, + { + "epoch": 1.9084120360202066, + "grad_norm": 2.32175874710083, + "learning_rate": 1e-06, + "loss": 0.8798, + "mean_token_accuracy": 0.736646294593811, + "num_tokens": 442023079.0, + "step": 17378 + }, + { + "epoch": 1.9085218537228201, + "grad_norm": 2.7114787101745605, + "learning_rate": 1e-06, + "loss": 0.8899, + "mean_token_accuracy": 0.7201029658317566, + "num_tokens": 442041837.0, + "step": 17379 + }, + { + "epoch": 1.9086316714254337, + "grad_norm": 2.179943323135376, + "learning_rate": 1e-06, + "loss": 0.9513, + "mean_token_accuracy": 0.7159072756767273, + "num_tokens": 442068469.0, + "step": 17380 + }, + { + "epoch": 1.9087414891280474, + "grad_norm": 2.3144800662994385, + "learning_rate": 1e-06, + "loss": 0.883, + "mean_token_accuracy": 0.7257770299911499, + "num_tokens": 442091949.0, + "step": 17381 + }, + { + "epoch": 1.9088513068306612, + "grad_norm": 2.4772653579711914, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7216494679450989, + "num_tokens": 442113822.0, + "step": 17382 + }, + { + "epoch": 1.9089611245332747, + "grad_norm": 2.2799220085144043, + "learning_rate": 1e-06, + "loss": 0.8671, + "mean_token_accuracy": 0.7352633476257324, + "num_tokens": 442138379.0, + "step": 17383 + }, + { + "epoch": 1.9090709422358885, + "grad_norm": 2.6332919597625732, + "learning_rate": 1e-06, + "loss": 0.7986, + "mean_token_accuracy": 0.7436619997024536, + "num_tokens": 442156193.0, + "step": 17384 + }, + { + "epoch": 1.909180759938502, + "grad_norm": 2.254171848297119, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7158671617507935, + "num_tokens": 442179668.0, + "step": 17385 + }, + { + "epoch": 1.9092905776411158, + "grad_norm": 2.092698574066162, + "learning_rate": 1e-06, + "loss": 0.9656, + "mean_token_accuracy": 0.7014471888542175, + "num_tokens": 442210066.0, + "step": 17386 + }, + { + "epoch": 1.9094003953437295, + "grad_norm": 2.323021650314331, + "learning_rate": 1e-06, + "loss": 0.8793, + "mean_token_accuracy": 0.7217041254043579, + "num_tokens": 442235134.0, + "step": 17387 + }, + { + "epoch": 1.909510213046343, + "grad_norm": 2.3789563179016113, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7163422107696533, + "num_tokens": 442259096.0, + "step": 17388 + }, + { + "epoch": 1.9096200307489566, + "grad_norm": 2.220984935760498, + "learning_rate": 1e-06, + "loss": 0.9123, + "mean_token_accuracy": 0.7161234021186829, + "num_tokens": 442285556.0, + "step": 17389 + }, + { + "epoch": 1.9097298484515703, + "grad_norm": 2.4912350177764893, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7386308908462524, + "num_tokens": 442307039.0, + "step": 17390 + }, + { + "epoch": 1.909839666154184, + "grad_norm": 2.397059440612793, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7182272672653198, + "num_tokens": 442330640.0, + "step": 17391 + }, + { + "epoch": 1.9099494838567979, + "grad_norm": 2.0142812728881836, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7112267017364502, + "num_tokens": 442361004.0, + "step": 17392 + }, + { + "epoch": 1.9100593015594114, + "grad_norm": 2.6214771270751953, + "learning_rate": 1e-06, + "loss": 0.8889, + "mean_token_accuracy": 0.7152292132377625, + "num_tokens": 442380083.0, + "step": 17393 + }, + { + "epoch": 1.910169119262025, + "grad_norm": 2.262845993041992, + "learning_rate": 1e-06, + "loss": 0.9381, + "mean_token_accuracy": 0.7118587493896484, + "num_tokens": 442406885.0, + "step": 17394 + }, + { + "epoch": 1.9102789369646387, + "grad_norm": 2.187877655029297, + "learning_rate": 1e-06, + "loss": 0.8529, + "mean_token_accuracy": 0.7307750582695007, + "num_tokens": 442431857.0, + "step": 17395 + }, + { + "epoch": 1.9103887546672524, + "grad_norm": 2.044443368911743, + "learning_rate": 1e-06, + "loss": 0.9441, + "mean_token_accuracy": 0.7138819098472595, + "num_tokens": 442461725.0, + "step": 17396 + }, + { + "epoch": 1.910498572369866, + "grad_norm": 2.3789124488830566, + "learning_rate": 1e-06, + "loss": 0.7602, + "mean_token_accuracy": 0.7570508718490601, + "num_tokens": 442482942.0, + "step": 17397 + }, + { + "epoch": 1.9106083900724797, + "grad_norm": 2.058967351913452, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7115926146507263, + "num_tokens": 442513642.0, + "step": 17398 + }, + { + "epoch": 1.9107182077750933, + "grad_norm": 2.344348669052124, + "learning_rate": 1e-06, + "loss": 0.8272, + "mean_token_accuracy": 0.737295389175415, + "num_tokens": 442538060.0, + "step": 17399 + }, + { + "epoch": 1.910828025477707, + "grad_norm": 2.1741223335266113, + "learning_rate": 1e-06, + "loss": 0.8545, + "mean_token_accuracy": 0.7324307560920715, + "num_tokens": 442563163.0, + "step": 17400 + }, + { + "epoch": 1.9109378431803208, + "grad_norm": 2.1865105628967285, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7216626405715942, + "num_tokens": 442589570.0, + "step": 17401 + }, + { + "epoch": 1.9110476608829343, + "grad_norm": 2.243504047393799, + "learning_rate": 1e-06, + "loss": 0.934, + "mean_token_accuracy": 0.7190660834312439, + "num_tokens": 442614044.0, + "step": 17402 + }, + { + "epoch": 1.9111574785855479, + "grad_norm": 2.7183666229248047, + "learning_rate": 1e-06, + "loss": 0.8763, + "mean_token_accuracy": 0.7240787744522095, + "num_tokens": 442632639.0, + "step": 17403 + }, + { + "epoch": 1.9112672962881616, + "grad_norm": 2.0880565643310547, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.7193894386291504, + "num_tokens": 442661930.0, + "step": 17404 + }, + { + "epoch": 1.9113771139907754, + "grad_norm": 2.5560457706451416, + "learning_rate": 1e-06, + "loss": 0.845, + "mean_token_accuracy": 0.7298034429550171, + "num_tokens": 442682173.0, + "step": 17405 + }, + { + "epoch": 1.9114869316933891, + "grad_norm": 2.47414231300354, + "learning_rate": 1e-06, + "loss": 0.8421, + "mean_token_accuracy": 0.7347781658172607, + "num_tokens": 442705010.0, + "step": 17406 + }, + { + "epoch": 1.9115967493960027, + "grad_norm": 2.4429142475128174, + "learning_rate": 1e-06, + "loss": 0.8345, + "mean_token_accuracy": 0.7349019050598145, + "num_tokens": 442724706.0, + "step": 17407 + }, + { + "epoch": 1.9117065670986162, + "grad_norm": 2.3474504947662354, + "learning_rate": 1e-06, + "loss": 0.8727, + "mean_token_accuracy": 0.7268702387809753, + "num_tokens": 442747500.0, + "step": 17408 + }, + { + "epoch": 1.91181638480123, + "grad_norm": 2.1941211223602295, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7211645245552063, + "num_tokens": 442776073.0, + "step": 17409 + }, + { + "epoch": 1.9119262025038437, + "grad_norm": 2.1336114406585693, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7142475843429565, + "num_tokens": 442802984.0, + "step": 17410 + }, + { + "epoch": 1.9120360202064572, + "grad_norm": 2.382550001144409, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7105538845062256, + "num_tokens": 442825356.0, + "step": 17411 + }, + { + "epoch": 1.9121458379090708, + "grad_norm": 2.2054286003112793, + "learning_rate": 1e-06, + "loss": 0.8585, + "mean_token_accuracy": 0.7334479093551636, + "num_tokens": 442851731.0, + "step": 17412 + }, + { + "epoch": 1.9122556556116845, + "grad_norm": 2.0804646015167236, + "learning_rate": 1e-06, + "loss": 0.8242, + "mean_token_accuracy": 0.744890570640564, + "num_tokens": 442877945.0, + "step": 17413 + }, + { + "epoch": 1.9123654733142983, + "grad_norm": 2.3583171367645264, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7318970561027527, + "num_tokens": 442901696.0, + "step": 17414 + }, + { + "epoch": 1.912475291016912, + "grad_norm": 2.2256553173065186, + "learning_rate": 1e-06, + "loss": 0.9836, + "mean_token_accuracy": 0.6993652582168579, + "num_tokens": 442928225.0, + "step": 17415 + }, + { + "epoch": 1.9125851087195256, + "grad_norm": 2.3331458568573, + "learning_rate": 1e-06, + "loss": 0.9023, + "mean_token_accuracy": 0.723314642906189, + "num_tokens": 442952444.0, + "step": 17416 + }, + { + "epoch": 1.9126949264221391, + "grad_norm": 2.143284559249878, + "learning_rate": 1e-06, + "loss": 0.813, + "mean_token_accuracy": 0.7397628426551819, + "num_tokens": 442978827.0, + "step": 17417 + }, + { + "epoch": 1.9128047441247529, + "grad_norm": 2.3254823684692383, + "learning_rate": 1e-06, + "loss": 0.8468, + "mean_token_accuracy": 0.7356197834014893, + "num_tokens": 443002549.0, + "step": 17418 + }, + { + "epoch": 1.9129145618273666, + "grad_norm": 2.155047655105591, + "learning_rate": 1e-06, + "loss": 0.8407, + "mean_token_accuracy": 0.7426550984382629, + "num_tokens": 443027999.0, + "step": 17419 + }, + { + "epoch": 1.9130243795299804, + "grad_norm": 2.526090621948242, + "learning_rate": 1e-06, + "loss": 0.8408, + "mean_token_accuracy": 0.7310643792152405, + "num_tokens": 443049814.0, + "step": 17420 + }, + { + "epoch": 1.913134197232594, + "grad_norm": 2.44826078414917, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.7370800971984863, + "num_tokens": 443071433.0, + "step": 17421 + }, + { + "epoch": 1.9132440149352075, + "grad_norm": 2.268709897994995, + "learning_rate": 1e-06, + "loss": 0.9188, + "mean_token_accuracy": 0.7162367105484009, + "num_tokens": 443098684.0, + "step": 17422 + }, + { + "epoch": 1.9133538326378212, + "grad_norm": 2.4666500091552734, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7097901105880737, + "num_tokens": 443123637.0, + "step": 17423 + }, + { + "epoch": 1.913463650340435, + "grad_norm": 2.440263032913208, + "learning_rate": 1e-06, + "loss": 0.8977, + "mean_token_accuracy": 0.721574068069458, + "num_tokens": 443147827.0, + "step": 17424 + }, + { + "epoch": 1.9135734680430485, + "grad_norm": 2.1993746757507324, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.719825029373169, + "num_tokens": 443173872.0, + "step": 17425 + }, + { + "epoch": 1.913683285745662, + "grad_norm": 2.2729625701904297, + "learning_rate": 1e-06, + "loss": 0.8767, + "mean_token_accuracy": 0.7282062768936157, + "num_tokens": 443198110.0, + "step": 17426 + }, + { + "epoch": 1.9137931034482758, + "grad_norm": 2.3487982749938965, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7200674414634705, + "num_tokens": 443221657.0, + "step": 17427 + }, + { + "epoch": 1.9139029211508896, + "grad_norm": 1.8565446138381958, + "learning_rate": 1e-06, + "loss": 0.9478, + "mean_token_accuracy": 0.7144209146499634, + "num_tokens": 443256792.0, + "step": 17428 + }, + { + "epoch": 1.9140127388535033, + "grad_norm": 2.546532392501831, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7230721116065979, + "num_tokens": 443277984.0, + "step": 17429 + }, + { + "epoch": 1.9141225565561168, + "grad_norm": 2.2233705520629883, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7100514769554138, + "num_tokens": 443305120.0, + "step": 17430 + }, + { + "epoch": 1.9142323742587304, + "grad_norm": 2.4296419620513916, + "learning_rate": 1e-06, + "loss": 0.9265, + "mean_token_accuracy": 0.7116507291793823, + "num_tokens": 443327555.0, + "step": 17431 + }, + { + "epoch": 1.9143421919613441, + "grad_norm": 2.1976070404052734, + "learning_rate": 1e-06, + "loss": 0.9039, + "mean_token_accuracy": 0.7166774272918701, + "num_tokens": 443354077.0, + "step": 17432 + }, + { + "epoch": 1.914452009663958, + "grad_norm": 2.1188855171203613, + "learning_rate": 1e-06, + "loss": 0.8048, + "mean_token_accuracy": 0.7483252286911011, + "num_tokens": 443381760.0, + "step": 17433 + }, + { + "epoch": 1.9145618273665717, + "grad_norm": 2.405578851699829, + "learning_rate": 1e-06, + "loss": 0.847, + "mean_token_accuracy": 0.736099898815155, + "num_tokens": 443403599.0, + "step": 17434 + }, + { + "epoch": 1.9146716450691852, + "grad_norm": 2.458650827407837, + "learning_rate": 1e-06, + "loss": 0.833, + "mean_token_accuracy": 0.7340362668037415, + "num_tokens": 443424747.0, + "step": 17435 + }, + { + "epoch": 1.9147814627717987, + "grad_norm": 2.3040902614593506, + "learning_rate": 1e-06, + "loss": 0.9536, + "mean_token_accuracy": 0.7017807960510254, + "num_tokens": 443449818.0, + "step": 17436 + }, + { + "epoch": 1.9148912804744125, + "grad_norm": 2.376375675201416, + "learning_rate": 1e-06, + "loss": 0.9821, + "mean_token_accuracy": 0.6988064050674438, + "num_tokens": 443475723.0, + "step": 17437 + }, + { + "epoch": 1.9150010981770262, + "grad_norm": 2.310375928878784, + "learning_rate": 1e-06, + "loss": 0.9353, + "mean_token_accuracy": 0.7051835656166077, + "num_tokens": 443501987.0, + "step": 17438 + }, + { + "epoch": 1.9151109158796398, + "grad_norm": 2.2034342288970947, + "learning_rate": 1e-06, + "loss": 0.9749, + "mean_token_accuracy": 0.7051937580108643, + "num_tokens": 443530728.0, + "step": 17439 + }, + { + "epoch": 1.9152207335822533, + "grad_norm": 2.4827537536621094, + "learning_rate": 1e-06, + "loss": 0.866, + "mean_token_accuracy": 0.7313588857650757, + "num_tokens": 443553047.0, + "step": 17440 + }, + { + "epoch": 1.915330551284867, + "grad_norm": 2.0690057277679443, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.729898989200592, + "num_tokens": 443581478.0, + "step": 17441 + }, + { + "epoch": 1.9154403689874808, + "grad_norm": 2.2619173526763916, + "learning_rate": 1e-06, + "loss": 0.8035, + "mean_token_accuracy": 0.7467923164367676, + "num_tokens": 443605663.0, + "step": 17442 + }, + { + "epoch": 1.9155501866900946, + "grad_norm": 2.023391008377075, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7062618732452393, + "num_tokens": 443637506.0, + "step": 17443 + }, + { + "epoch": 1.9156600043927081, + "grad_norm": 2.3391146659851074, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7255963087081909, + "num_tokens": 443661863.0, + "step": 17444 + }, + { + "epoch": 1.9157698220953217, + "grad_norm": 2.2242653369903564, + "learning_rate": 1e-06, + "loss": 0.91, + "mean_token_accuracy": 0.7171052694320679, + "num_tokens": 443687990.0, + "step": 17445 + }, + { + "epoch": 1.9158796397979354, + "grad_norm": 2.411550283432007, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7379382252693176, + "num_tokens": 443710012.0, + "step": 17446 + }, + { + "epoch": 1.9159894575005492, + "grad_norm": 2.220027446746826, + "learning_rate": 1e-06, + "loss": 0.9431, + "mean_token_accuracy": 0.702750027179718, + "num_tokens": 443736939.0, + "step": 17447 + }, + { + "epoch": 1.9160992752031627, + "grad_norm": 2.058915376663208, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7041769027709961, + "num_tokens": 443766592.0, + "step": 17448 + }, + { + "epoch": 1.9162090929057765, + "grad_norm": 2.0706028938293457, + "learning_rate": 1e-06, + "loss": 0.865, + "mean_token_accuracy": 0.729748010635376, + "num_tokens": 443795865.0, + "step": 17449 + }, + { + "epoch": 1.91631891060839, + "grad_norm": 2.054865598678589, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7110409736633301, + "num_tokens": 443825156.0, + "step": 17450 + }, + { + "epoch": 1.9164287283110037, + "grad_norm": 2.080941915512085, + "learning_rate": 1e-06, + "loss": 0.8968, + "mean_token_accuracy": 0.724838137626648, + "num_tokens": 443853725.0, + "step": 17451 + }, + { + "epoch": 1.9165385460136175, + "grad_norm": 2.3053832054138184, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.698188841342926, + "num_tokens": 443879776.0, + "step": 17452 + }, + { + "epoch": 1.916648363716231, + "grad_norm": 2.2288200855255127, + "learning_rate": 1e-06, + "loss": 0.9206, + "mean_token_accuracy": 0.7157824039459229, + "num_tokens": 443904416.0, + "step": 17453 + }, + { + "epoch": 1.9167581814188446, + "grad_norm": 2.2385966777801514, + "learning_rate": 1e-06, + "loss": 0.8821, + "mean_token_accuracy": 0.7248520851135254, + "num_tokens": 443930031.0, + "step": 17454 + }, + { + "epoch": 1.9168679991214583, + "grad_norm": 2.1830368041992188, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7197409868240356, + "num_tokens": 443956998.0, + "step": 17455 + }, + { + "epoch": 1.916977816824072, + "grad_norm": 2.196136951446533, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7119365930557251, + "num_tokens": 443982475.0, + "step": 17456 + }, + { + "epoch": 1.9170876345266858, + "grad_norm": 2.374462127685547, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7368846535682678, + "num_tokens": 444004449.0, + "step": 17457 + }, + { + "epoch": 1.9171974522292994, + "grad_norm": 2.455153703689575, + "learning_rate": 1e-06, + "loss": 0.9631, + "mean_token_accuracy": 0.7040761709213257, + "num_tokens": 444029811.0, + "step": 17458 + }, + { + "epoch": 1.917307269931913, + "grad_norm": 2.1459779739379883, + "learning_rate": 1e-06, + "loss": 0.9178, + "mean_token_accuracy": 0.7159512042999268, + "num_tokens": 444057486.0, + "step": 17459 + }, + { + "epoch": 1.9174170876345267, + "grad_norm": 1.9031081199645996, + "learning_rate": 1e-06, + "loss": 0.9539, + "mean_token_accuracy": 0.7108820676803589, + "num_tokens": 444090963.0, + "step": 17460 + }, + { + "epoch": 1.9175269053371404, + "grad_norm": 2.368461847305298, + "learning_rate": 1e-06, + "loss": 0.8694, + "mean_token_accuracy": 0.7369377613067627, + "num_tokens": 444115202.0, + "step": 17461 + }, + { + "epoch": 1.917636723039754, + "grad_norm": 2.136692523956299, + "learning_rate": 1e-06, + "loss": 0.8787, + "mean_token_accuracy": 0.7222603559494019, + "num_tokens": 444142587.0, + "step": 17462 + }, + { + "epoch": 1.9177465407423677, + "grad_norm": 2.1817126274108887, + "learning_rate": 1e-06, + "loss": 0.8957, + "mean_token_accuracy": 0.7314577698707581, + "num_tokens": 444167412.0, + "step": 17463 + }, + { + "epoch": 1.9178563584449813, + "grad_norm": 2.4305121898651123, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7074384093284607, + "num_tokens": 444189705.0, + "step": 17464 + }, + { + "epoch": 1.917966176147595, + "grad_norm": 2.1666040420532227, + "learning_rate": 1e-06, + "loss": 0.8684, + "mean_token_accuracy": 0.7386674880981445, + "num_tokens": 444214855.0, + "step": 17465 + }, + { + "epoch": 1.9180759938502088, + "grad_norm": 2.0463333129882812, + "learning_rate": 1e-06, + "loss": 0.9794, + "mean_token_accuracy": 0.6976107954978943, + "num_tokens": 444246060.0, + "step": 17466 + }, + { + "epoch": 1.9181858115528223, + "grad_norm": 2.143728494644165, + "learning_rate": 1e-06, + "loss": 0.8845, + "mean_token_accuracy": 0.7267388105392456, + "num_tokens": 444273092.0, + "step": 17467 + }, + { + "epoch": 1.9182956292554358, + "grad_norm": 2.3296735286712646, + "learning_rate": 1e-06, + "loss": 0.9277, + "mean_token_accuracy": 0.7094998359680176, + "num_tokens": 444298471.0, + "step": 17468 + }, + { + "epoch": 1.9184054469580496, + "grad_norm": 2.345079183578491, + "learning_rate": 1e-06, + "loss": 0.8719, + "mean_token_accuracy": 0.7267123460769653, + "num_tokens": 444321225.0, + "step": 17469 + }, + { + "epoch": 1.9185152646606634, + "grad_norm": 2.2502455711364746, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7128254175186157, + "num_tokens": 444345323.0, + "step": 17470 + }, + { + "epoch": 1.9186250823632771, + "grad_norm": 2.190495729446411, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.6963980197906494, + "num_tokens": 444371833.0, + "step": 17471 + }, + { + "epoch": 1.9187349000658906, + "grad_norm": 2.5851924419403076, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.7075064182281494, + "num_tokens": 444393692.0, + "step": 17472 + }, + { + "epoch": 1.9188447177685042, + "grad_norm": 2.281682014465332, + "learning_rate": 1e-06, + "loss": 0.7443, + "mean_token_accuracy": 0.7618696689605713, + "num_tokens": 444415901.0, + "step": 17473 + }, + { + "epoch": 1.918954535471118, + "grad_norm": 2.1888515949249268, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7191895842552185, + "num_tokens": 444442592.0, + "step": 17474 + }, + { + "epoch": 1.9190643531737317, + "grad_norm": 2.0704245567321777, + "learning_rate": 1e-06, + "loss": 0.8304, + "mean_token_accuracy": 0.7359561324119568, + "num_tokens": 444470204.0, + "step": 17475 + }, + { + "epoch": 1.9191741708763452, + "grad_norm": 1.8929877281188965, + "learning_rate": 1e-06, + "loss": 0.9014, + "mean_token_accuracy": 0.7161503434181213, + "num_tokens": 444502559.0, + "step": 17476 + }, + { + "epoch": 1.9192839885789588, + "grad_norm": 2.3823635578155518, + "learning_rate": 1e-06, + "loss": 0.8281, + "mean_token_accuracy": 0.7373891472816467, + "num_tokens": 444524219.0, + "step": 17477 + }, + { + "epoch": 1.9193938062815725, + "grad_norm": 2.0086727142333984, + "learning_rate": 1e-06, + "loss": 0.9865, + "mean_token_accuracy": 0.6996273994445801, + "num_tokens": 444554903.0, + "step": 17478 + }, + { + "epoch": 1.9195036239841863, + "grad_norm": 1.9182502031326294, + "learning_rate": 1e-06, + "loss": 0.9137, + "mean_token_accuracy": 0.7217229604721069, + "num_tokens": 444585520.0, + "step": 17479 + }, + { + "epoch": 1.9196134416868, + "grad_norm": 2.5128118991851807, + "learning_rate": 1e-06, + "loss": 0.8293, + "mean_token_accuracy": 0.7308422327041626, + "num_tokens": 444604998.0, + "step": 17480 + }, + { + "epoch": 1.9197232593894136, + "grad_norm": 2.668604850769043, + "learning_rate": 1e-06, + "loss": 0.7794, + "mean_token_accuracy": 0.7430775165557861, + "num_tokens": 444623788.0, + "step": 17481 + }, + { + "epoch": 1.919833077092027, + "grad_norm": 2.0675368309020996, + "learning_rate": 1e-06, + "loss": 0.9266, + "mean_token_accuracy": 0.7115433216094971, + "num_tokens": 444653073.0, + "step": 17482 + }, + { + "epoch": 1.9199428947946409, + "grad_norm": 2.3396997451782227, + "learning_rate": 1e-06, + "loss": 0.8205, + "mean_token_accuracy": 0.7425109148025513, + "num_tokens": 444674793.0, + "step": 17483 + }, + { + "epoch": 1.9200527124972546, + "grad_norm": 2.1192808151245117, + "learning_rate": 1e-06, + "loss": 0.9589, + "mean_token_accuracy": 0.7037542462348938, + "num_tokens": 444703328.0, + "step": 17484 + }, + { + "epoch": 1.9201625301998684, + "grad_norm": 2.0449461936950684, + "learning_rate": 1e-06, + "loss": 0.9565, + "mean_token_accuracy": 0.7036712169647217, + "num_tokens": 444732305.0, + "step": 17485 + }, + { + "epoch": 1.920272347902482, + "grad_norm": 2.1613457202911377, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7051030397415161, + "num_tokens": 444761070.0, + "step": 17486 + }, + { + "epoch": 1.9203821656050954, + "grad_norm": 2.476781129837036, + "learning_rate": 1e-06, + "loss": 0.7545, + "mean_token_accuracy": 0.7572047710418701, + "num_tokens": 444779646.0, + "step": 17487 + }, + { + "epoch": 1.9204919833077092, + "grad_norm": 2.331047534942627, + "learning_rate": 1e-06, + "loss": 0.9524, + "mean_token_accuracy": 0.7029038667678833, + "num_tokens": 444802722.0, + "step": 17488 + }, + { + "epoch": 1.920601801010323, + "grad_norm": 2.0110385417938232, + "learning_rate": 1e-06, + "loss": 0.8812, + "mean_token_accuracy": 0.7227860689163208, + "num_tokens": 444835210.0, + "step": 17489 + }, + { + "epoch": 1.9207116187129365, + "grad_norm": 2.5172853469848633, + "learning_rate": 1e-06, + "loss": 0.9305, + "mean_token_accuracy": 0.715493381023407, + "num_tokens": 444859482.0, + "step": 17490 + }, + { + "epoch": 1.92082143641555, + "grad_norm": 2.4330759048461914, + "learning_rate": 1e-06, + "loss": 0.9437, + "mean_token_accuracy": 0.7085809707641602, + "num_tokens": 444884527.0, + "step": 17491 + }, + { + "epoch": 1.9209312541181638, + "grad_norm": 2.5429129600524902, + "learning_rate": 1e-06, + "loss": 0.9025, + "mean_token_accuracy": 0.7198584079742432, + "num_tokens": 444903850.0, + "step": 17492 + }, + { + "epoch": 1.9210410718207775, + "grad_norm": 2.1446309089660645, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7174152135848999, + "num_tokens": 444933805.0, + "step": 17493 + }, + { + "epoch": 1.9211508895233913, + "grad_norm": 2.6337318420410156, + "learning_rate": 1e-06, + "loss": 0.8021, + "mean_token_accuracy": 0.7441578507423401, + "num_tokens": 444954400.0, + "step": 17494 + }, + { + "epoch": 1.9212607072260048, + "grad_norm": 2.1955835819244385, + "learning_rate": 1e-06, + "loss": 0.8724, + "mean_token_accuracy": 0.7300823926925659, + "num_tokens": 444979698.0, + "step": 17495 + }, + { + "epoch": 1.9213705249286184, + "grad_norm": 2.0987987518310547, + "learning_rate": 1e-06, + "loss": 0.9035, + "mean_token_accuracy": 0.7226803302764893, + "num_tokens": 445011248.0, + "step": 17496 + }, + { + "epoch": 1.9214803426312321, + "grad_norm": 2.0669028759002686, + "learning_rate": 1e-06, + "loss": 0.7877, + "mean_token_accuracy": 0.7509075403213501, + "num_tokens": 445039133.0, + "step": 17497 + }, + { + "epoch": 1.9215901603338459, + "grad_norm": 2.0589077472686768, + "learning_rate": 1e-06, + "loss": 0.9047, + "mean_token_accuracy": 0.7208627462387085, + "num_tokens": 445068376.0, + "step": 17498 + }, + { + "epoch": 1.9216999780364594, + "grad_norm": 2.125431776046753, + "learning_rate": 1e-06, + "loss": 0.9439, + "mean_token_accuracy": 0.7079111933708191, + "num_tokens": 445097493.0, + "step": 17499 + }, + { + "epoch": 1.9218097957390732, + "grad_norm": 2.2176895141601562, + "learning_rate": 1e-06, + "loss": 0.962, + "mean_token_accuracy": 0.6967045068740845, + "num_tokens": 445124675.0, + "step": 17500 + }, + { + "epoch": 1.9219196134416867, + "grad_norm": 2.2959184646606445, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7024389505386353, + "num_tokens": 445149973.0, + "step": 17501 + }, + { + "epoch": 1.9220294311443005, + "grad_norm": 2.6077067852020264, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7112358808517456, + "num_tokens": 445169844.0, + "step": 17502 + }, + { + "epoch": 1.9221392488469142, + "grad_norm": 2.4477930068969727, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7360169887542725, + "num_tokens": 445192497.0, + "step": 17503 + }, + { + "epoch": 1.9222490665495278, + "grad_norm": 2.189892292022705, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.7081260085105896, + "num_tokens": 445219782.0, + "step": 17504 + }, + { + "epoch": 1.9223588842521413, + "grad_norm": 2.3804261684417725, + "learning_rate": 1e-06, + "loss": 0.8566, + "mean_token_accuracy": 0.7310453653335571, + "num_tokens": 445241837.0, + "step": 17505 + }, + { + "epoch": 1.922468701954755, + "grad_norm": 2.0727145671844482, + "learning_rate": 1e-06, + "loss": 0.9981, + "mean_token_accuracy": 0.6906462907791138, + "num_tokens": 445272353.0, + "step": 17506 + }, + { + "epoch": 1.9225785196573688, + "grad_norm": 2.39674973487854, + "learning_rate": 1e-06, + "loss": 0.8593, + "mean_token_accuracy": 0.726553201675415, + "num_tokens": 445293600.0, + "step": 17507 + }, + { + "epoch": 1.9226883373599826, + "grad_norm": 2.5779740810394287, + "learning_rate": 1e-06, + "loss": 0.779, + "mean_token_accuracy": 0.758133053779602, + "num_tokens": 445312201.0, + "step": 17508 + }, + { + "epoch": 1.922798155062596, + "grad_norm": 2.587514877319336, + "learning_rate": 1e-06, + "loss": 0.848, + "mean_token_accuracy": 0.7244306206703186, + "num_tokens": 445332364.0, + "step": 17509 + }, + { + "epoch": 1.9229079727652096, + "grad_norm": 2.1552071571350098, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.7227529287338257, + "num_tokens": 445359811.0, + "step": 17510 + }, + { + "epoch": 1.9230177904678234, + "grad_norm": 2.5629732608795166, + "learning_rate": 1e-06, + "loss": 0.839, + "mean_token_accuracy": 0.7433261275291443, + "num_tokens": 445380988.0, + "step": 17511 + }, + { + "epoch": 1.9231276081704372, + "grad_norm": 1.9516338109970093, + "learning_rate": 1e-06, + "loss": 0.9779, + "mean_token_accuracy": 0.7046690583229065, + "num_tokens": 445414481.0, + "step": 17512 + }, + { + "epoch": 1.9232374258730507, + "grad_norm": 2.0584380626678467, + "learning_rate": 1e-06, + "loss": 0.9096, + "mean_token_accuracy": 0.7126266956329346, + "num_tokens": 445443714.0, + "step": 17513 + }, + { + "epoch": 1.9233472435756644, + "grad_norm": 2.388594150543213, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7140763998031616, + "num_tokens": 445467068.0, + "step": 17514 + }, + { + "epoch": 1.923457061278278, + "grad_norm": 2.2636852264404297, + "learning_rate": 1e-06, + "loss": 1.0258, + "mean_token_accuracy": 0.6874221563339233, + "num_tokens": 445493824.0, + "step": 17515 + }, + { + "epoch": 1.9235668789808917, + "grad_norm": 2.253818988800049, + "learning_rate": 1e-06, + "loss": 0.8229, + "mean_token_accuracy": 0.7343906164169312, + "num_tokens": 445517818.0, + "step": 17516 + }, + { + "epoch": 1.9236766966835055, + "grad_norm": 2.022857904434204, + "learning_rate": 1e-06, + "loss": 0.8688, + "mean_token_accuracy": 0.7287799119949341, + "num_tokens": 445547931.0, + "step": 17517 + }, + { + "epoch": 1.923786514386119, + "grad_norm": 2.7221479415893555, + "learning_rate": 1e-06, + "loss": 0.8644, + "mean_token_accuracy": 0.7220366597175598, + "num_tokens": 445565738.0, + "step": 17518 + }, + { + "epoch": 1.9238963320887326, + "grad_norm": 2.2125439643859863, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7082692384719849, + "num_tokens": 445591370.0, + "step": 17519 + }, + { + "epoch": 1.9240061497913463, + "grad_norm": 2.528076171875, + "learning_rate": 1e-06, + "loss": 0.8366, + "mean_token_accuracy": 0.730754554271698, + "num_tokens": 445611913.0, + "step": 17520 + }, + { + "epoch": 1.92411596749396, + "grad_norm": 2.206087112426758, + "learning_rate": 1e-06, + "loss": 0.9154, + "mean_token_accuracy": 0.7098110914230347, + "num_tokens": 445638347.0, + "step": 17521 + }, + { + "epoch": 1.9242257851965738, + "grad_norm": 2.2460384368896484, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7045755982398987, + "num_tokens": 445666375.0, + "step": 17522 + }, + { + "epoch": 1.9243356028991874, + "grad_norm": 2.294548511505127, + "learning_rate": 1e-06, + "loss": 0.9143, + "mean_token_accuracy": 0.7197883129119873, + "num_tokens": 445691414.0, + "step": 17523 + }, + { + "epoch": 1.924445420601801, + "grad_norm": 2.2697811126708984, + "learning_rate": 1e-06, + "loss": 0.9424, + "mean_token_accuracy": 0.708000898361206, + "num_tokens": 445717121.0, + "step": 17524 + }, + { + "epoch": 1.9245552383044147, + "grad_norm": 2.22758150100708, + "learning_rate": 1e-06, + "loss": 0.8641, + "mean_token_accuracy": 0.7229865789413452, + "num_tokens": 445741956.0, + "step": 17525 + }, + { + "epoch": 1.9246650560070284, + "grad_norm": 2.3607630729675293, + "learning_rate": 1e-06, + "loss": 0.8822, + "mean_token_accuracy": 0.7348689436912537, + "num_tokens": 445765447.0, + "step": 17526 + }, + { + "epoch": 1.924774873709642, + "grad_norm": 2.641341209411621, + "learning_rate": 1e-06, + "loss": 0.9289, + "mean_token_accuracy": 0.7186429500579834, + "num_tokens": 445784894.0, + "step": 17527 + }, + { + "epoch": 1.9248846914122557, + "grad_norm": 2.3908112049102783, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7280346751213074, + "num_tokens": 445808557.0, + "step": 17528 + }, + { + "epoch": 1.9249945091148692, + "grad_norm": 2.279334545135498, + "learning_rate": 1e-06, + "loss": 0.8741, + "mean_token_accuracy": 0.725597620010376, + "num_tokens": 445834707.0, + "step": 17529 + }, + { + "epoch": 1.925104326817483, + "grad_norm": 2.613748788833618, + "learning_rate": 1e-06, + "loss": 0.9292, + "mean_token_accuracy": 0.7141956686973572, + "num_tokens": 445854884.0, + "step": 17530 + }, + { + "epoch": 1.9252141445200968, + "grad_norm": 2.3581528663635254, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7214937210083008, + "num_tokens": 445879370.0, + "step": 17531 + }, + { + "epoch": 1.9253239622227103, + "grad_norm": 2.2336411476135254, + "learning_rate": 1e-06, + "loss": 0.8236, + "mean_token_accuracy": 0.7510049343109131, + "num_tokens": 445906252.0, + "step": 17532 + }, + { + "epoch": 1.9254337799253238, + "grad_norm": 1.8987908363342285, + "learning_rate": 1e-06, + "loss": 0.9549, + "mean_token_accuracy": 0.7051460146903992, + "num_tokens": 445938198.0, + "step": 17533 + }, + { + "epoch": 1.9255435976279376, + "grad_norm": 2.4355576038360596, + "learning_rate": 1e-06, + "loss": 0.8112, + "mean_token_accuracy": 0.7366066575050354, + "num_tokens": 445959142.0, + "step": 17534 + }, + { + "epoch": 1.9256534153305513, + "grad_norm": 2.2970645427703857, + "learning_rate": 1e-06, + "loss": 0.9242, + "mean_token_accuracy": 0.7071092128753662, + "num_tokens": 445985086.0, + "step": 17535 + }, + { + "epoch": 1.925763233033165, + "grad_norm": 2.182450532913208, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.704903244972229, + "num_tokens": 446011588.0, + "step": 17536 + }, + { + "epoch": 1.9258730507357786, + "grad_norm": 2.188323497772217, + "learning_rate": 1e-06, + "loss": 0.9008, + "mean_token_accuracy": 0.7222923636436462, + "num_tokens": 446039399.0, + "step": 17537 + }, + { + "epoch": 1.9259828684383922, + "grad_norm": 2.089219570159912, + "learning_rate": 1e-06, + "loss": 0.9658, + "mean_token_accuracy": 0.7004479765892029, + "num_tokens": 446069156.0, + "step": 17538 + }, + { + "epoch": 1.926092686141006, + "grad_norm": 2.154261350631714, + "learning_rate": 1e-06, + "loss": 0.8282, + "mean_token_accuracy": 0.7363288402557373, + "num_tokens": 446095193.0, + "step": 17539 + }, + { + "epoch": 1.9262025038436197, + "grad_norm": 2.249683380126953, + "learning_rate": 1e-06, + "loss": 0.8627, + "mean_token_accuracy": 0.7282201051712036, + "num_tokens": 446121688.0, + "step": 17540 + }, + { + "epoch": 1.9263123215462332, + "grad_norm": 2.274888038635254, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7203616499900818, + "num_tokens": 446145663.0, + "step": 17541 + }, + { + "epoch": 1.9264221392488468, + "grad_norm": 2.3606669902801514, + "learning_rate": 1e-06, + "loss": 0.8165, + "mean_token_accuracy": 0.74044269323349, + "num_tokens": 446171164.0, + "step": 17542 + }, + { + "epoch": 1.9265319569514605, + "grad_norm": 2.1893913745880127, + "learning_rate": 1e-06, + "loss": 0.967, + "mean_token_accuracy": 0.700320839881897, + "num_tokens": 446201742.0, + "step": 17543 + }, + { + "epoch": 1.9266417746540743, + "grad_norm": 2.456355571746826, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7144153118133545, + "num_tokens": 446223791.0, + "step": 17544 + }, + { + "epoch": 1.926751592356688, + "grad_norm": 2.459545135498047, + "learning_rate": 1e-06, + "loss": 0.8223, + "mean_token_accuracy": 0.7358208894729614, + "num_tokens": 446245195.0, + "step": 17545 + }, + { + "epoch": 1.9268614100593016, + "grad_norm": 2.067408561706543, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7136691212654114, + "num_tokens": 446275062.0, + "step": 17546 + }, + { + "epoch": 1.926971227761915, + "grad_norm": 2.429948329925537, + "learning_rate": 1e-06, + "loss": 0.8604, + "mean_token_accuracy": 0.7370957732200623, + "num_tokens": 446297751.0, + "step": 17547 + }, + { + "epoch": 1.9270810454645289, + "grad_norm": 2.039937734603882, + "learning_rate": 1e-06, + "loss": 1.0135, + "mean_token_accuracy": 0.6916265487670898, + "num_tokens": 446328381.0, + "step": 17548 + }, + { + "epoch": 1.9271908631671426, + "grad_norm": 2.2424752712249756, + "learning_rate": 1e-06, + "loss": 0.9511, + "mean_token_accuracy": 0.701500415802002, + "num_tokens": 446354827.0, + "step": 17549 + }, + { + "epoch": 1.9273006808697564, + "grad_norm": 2.262221097946167, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7196453213691711, + "num_tokens": 446380317.0, + "step": 17550 + }, + { + "epoch": 1.92741049857237, + "grad_norm": 2.0209994316101074, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7231578826904297, + "num_tokens": 446411646.0, + "step": 17551 + }, + { + "epoch": 1.9275203162749834, + "grad_norm": 2.1908810138702393, + "learning_rate": 1e-06, + "loss": 0.9754, + "mean_token_accuracy": 0.6950534582138062, + "num_tokens": 446438828.0, + "step": 17552 + }, + { + "epoch": 1.9276301339775972, + "grad_norm": 2.2627053260803223, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7186741828918457, + "num_tokens": 446464707.0, + "step": 17553 + }, + { + "epoch": 1.927739951680211, + "grad_norm": 2.3706743717193604, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.728113055229187, + "num_tokens": 446486373.0, + "step": 17554 + }, + { + "epoch": 1.9278497693828245, + "grad_norm": 2.1004278659820557, + "learning_rate": 1e-06, + "loss": 0.9971, + "mean_token_accuracy": 0.6969113349914551, + "num_tokens": 446513556.0, + "step": 17555 + }, + { + "epoch": 1.927959587085438, + "grad_norm": 2.3665213584899902, + "learning_rate": 1e-06, + "loss": 0.8918, + "mean_token_accuracy": 0.7182176113128662, + "num_tokens": 446538072.0, + "step": 17556 + }, + { + "epoch": 1.9280694047880518, + "grad_norm": 2.15303897857666, + "learning_rate": 1e-06, + "loss": 0.9607, + "mean_token_accuracy": 0.7082297205924988, + "num_tokens": 446565088.0, + "step": 17557 + }, + { + "epoch": 1.9281792224906655, + "grad_norm": 2.2602927684783936, + "learning_rate": 1e-06, + "loss": 0.8607, + "mean_token_accuracy": 0.7298797369003296, + "num_tokens": 446591320.0, + "step": 17558 + }, + { + "epoch": 1.9282890401932793, + "grad_norm": 2.1996774673461914, + "learning_rate": 1e-06, + "loss": 0.9414, + "mean_token_accuracy": 0.7092286348342896, + "num_tokens": 446618279.0, + "step": 17559 + }, + { + "epoch": 1.9283988578958928, + "grad_norm": 2.5814898014068604, + "learning_rate": 1e-06, + "loss": 0.8429, + "mean_token_accuracy": 0.7382652759552002, + "num_tokens": 446637717.0, + "step": 17560 + }, + { + "epoch": 1.9285086755985064, + "grad_norm": 2.063753604888916, + "learning_rate": 1e-06, + "loss": 0.9953, + "mean_token_accuracy": 0.6991074085235596, + "num_tokens": 446669725.0, + "step": 17561 + }, + { + "epoch": 1.9286184933011201, + "grad_norm": 2.028369188308716, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.6888653039932251, + "num_tokens": 446699509.0, + "step": 17562 + }, + { + "epoch": 1.9287283110037339, + "grad_norm": 2.2617604732513428, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7113626003265381, + "num_tokens": 446726911.0, + "step": 17563 + }, + { + "epoch": 1.9288381287063474, + "grad_norm": 2.483313798904419, + "learning_rate": 1e-06, + "loss": 0.9442, + "mean_token_accuracy": 0.7219699025154114, + "num_tokens": 446750982.0, + "step": 17564 + }, + { + "epoch": 1.9289479464089612, + "grad_norm": 2.141939878463745, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7332414388656616, + "num_tokens": 446778283.0, + "step": 17565 + }, + { + "epoch": 1.9290577641115747, + "grad_norm": 2.237278938293457, + "learning_rate": 1e-06, + "loss": 0.8807, + "mean_token_accuracy": 0.7152901887893677, + "num_tokens": 446801630.0, + "step": 17566 + }, + { + "epoch": 1.9291675818141885, + "grad_norm": 2.201732873916626, + "learning_rate": 1e-06, + "loss": 0.9687, + "mean_token_accuracy": 0.6988175511360168, + "num_tokens": 446828314.0, + "step": 17567 + }, + { + "epoch": 1.9292773995168022, + "grad_norm": 2.3995559215545654, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7175796031951904, + "num_tokens": 446849505.0, + "step": 17568 + }, + { + "epoch": 1.9293872172194158, + "grad_norm": 2.3731226921081543, + "learning_rate": 1e-06, + "loss": 0.9022, + "mean_token_accuracy": 0.718001663684845, + "num_tokens": 446871060.0, + "step": 17569 + }, + { + "epoch": 1.9294970349220293, + "grad_norm": 2.3639841079711914, + "learning_rate": 1e-06, + "loss": 0.9803, + "mean_token_accuracy": 0.7049050331115723, + "num_tokens": 446896313.0, + "step": 17570 + }, + { + "epoch": 1.929606852624643, + "grad_norm": 2.3630588054656982, + "learning_rate": 1e-06, + "loss": 0.8823, + "mean_token_accuracy": 0.7242129445075989, + "num_tokens": 446920675.0, + "step": 17571 + }, + { + "epoch": 1.9297166703272568, + "grad_norm": 2.3160617351531982, + "learning_rate": 1e-06, + "loss": 0.9788, + "mean_token_accuracy": 0.7007683515548706, + "num_tokens": 446946652.0, + "step": 17572 + }, + { + "epoch": 1.9298264880298706, + "grad_norm": 2.409984588623047, + "learning_rate": 1e-06, + "loss": 0.8668, + "mean_token_accuracy": 0.7299213409423828, + "num_tokens": 446968746.0, + "step": 17573 + }, + { + "epoch": 1.929936305732484, + "grad_norm": 2.612144708633423, + "learning_rate": 1e-06, + "loss": 0.8153, + "mean_token_accuracy": 0.7432084679603577, + "num_tokens": 446988650.0, + "step": 17574 + }, + { + "epoch": 1.9300461234350976, + "grad_norm": 2.466641664505005, + "learning_rate": 1e-06, + "loss": 0.9683, + "mean_token_accuracy": 0.7023895382881165, + "num_tokens": 447011010.0, + "step": 17575 + }, + { + "epoch": 1.9301559411377114, + "grad_norm": 2.4543228149414062, + "learning_rate": 1e-06, + "loss": 0.8494, + "mean_token_accuracy": 0.7282810211181641, + "num_tokens": 447031901.0, + "step": 17576 + }, + { + "epoch": 1.9302657588403251, + "grad_norm": 2.589007616043091, + "learning_rate": 1e-06, + "loss": 0.8939, + "mean_token_accuracy": 0.7248576879501343, + "num_tokens": 447052785.0, + "step": 17577 + }, + { + "epoch": 1.9303755765429387, + "grad_norm": 2.255688428878784, + "learning_rate": 1e-06, + "loss": 0.8525, + "mean_token_accuracy": 0.7315580248832703, + "num_tokens": 447078840.0, + "step": 17578 + }, + { + "epoch": 1.9304853942455524, + "grad_norm": 2.172433853149414, + "learning_rate": 1e-06, + "loss": 0.9131, + "mean_token_accuracy": 0.7135599851608276, + "num_tokens": 447104146.0, + "step": 17579 + }, + { + "epoch": 1.930595211948166, + "grad_norm": 2.4762465953826904, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7171788811683655, + "num_tokens": 447126043.0, + "step": 17580 + }, + { + "epoch": 1.9307050296507797, + "grad_norm": 2.169581174850464, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7061499357223511, + "num_tokens": 447154353.0, + "step": 17581 + }, + { + "epoch": 1.9308148473533935, + "grad_norm": 2.363673210144043, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7267322540283203, + "num_tokens": 447178521.0, + "step": 17582 + }, + { + "epoch": 1.930924665056007, + "grad_norm": 2.3535878658294678, + "learning_rate": 1e-06, + "loss": 0.9351, + "mean_token_accuracy": 0.7053202986717224, + "num_tokens": 447202371.0, + "step": 17583 + }, + { + "epoch": 1.9310344827586206, + "grad_norm": 2.604748249053955, + "learning_rate": 1e-06, + "loss": 0.8778, + "mean_token_accuracy": 0.725548505783081, + "num_tokens": 447222608.0, + "step": 17584 + }, + { + "epoch": 1.9311443004612343, + "grad_norm": 2.385164976119995, + "learning_rate": 1e-06, + "loss": 1.0006, + "mean_token_accuracy": 0.6864388585090637, + "num_tokens": 447246732.0, + "step": 17585 + }, + { + "epoch": 1.931254118163848, + "grad_norm": 2.157266139984131, + "learning_rate": 1e-06, + "loss": 0.8254, + "mean_token_accuracy": 0.7399711012840271, + "num_tokens": 447272839.0, + "step": 17586 + }, + { + "epoch": 1.9313639358664618, + "grad_norm": 2.140256643295288, + "learning_rate": 1e-06, + "loss": 0.9069, + "mean_token_accuracy": 0.722895085811615, + "num_tokens": 447297829.0, + "step": 17587 + }, + { + "epoch": 1.9314737535690754, + "grad_norm": 2.210123062133789, + "learning_rate": 1e-06, + "loss": 0.8965, + "mean_token_accuracy": 0.7134658694267273, + "num_tokens": 447325023.0, + "step": 17588 + }, + { + "epoch": 1.931583571271689, + "grad_norm": 2.547882556915283, + "learning_rate": 1e-06, + "loss": 0.8698, + "mean_token_accuracy": 0.7329381704330444, + "num_tokens": 447345608.0, + "step": 17589 + }, + { + "epoch": 1.9316933889743026, + "grad_norm": 2.3396735191345215, + "learning_rate": 1e-06, + "loss": 0.8792, + "mean_token_accuracy": 0.7264617681503296, + "num_tokens": 447370431.0, + "step": 17590 + }, + { + "epoch": 1.9318032066769164, + "grad_norm": 2.276416778564453, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7148810625076294, + "num_tokens": 447396291.0, + "step": 17591 + }, + { + "epoch": 1.93191302437953, + "grad_norm": 2.0604238510131836, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7255699634552002, + "num_tokens": 447426985.0, + "step": 17592 + }, + { + "epoch": 1.9320228420821435, + "grad_norm": 2.1511900424957275, + "learning_rate": 1e-06, + "loss": 0.8761, + "mean_token_accuracy": 0.7233942747116089, + "num_tokens": 447453318.0, + "step": 17593 + }, + { + "epoch": 1.9321326597847572, + "grad_norm": 2.595144748687744, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7229384779930115, + "num_tokens": 447473452.0, + "step": 17594 + }, + { + "epoch": 1.932242477487371, + "grad_norm": 2.1290152072906494, + "learning_rate": 1e-06, + "loss": 0.9224, + "mean_token_accuracy": 0.7174409627914429, + "num_tokens": 447499509.0, + "step": 17595 + }, + { + "epoch": 1.9323522951899847, + "grad_norm": 2.010030508041382, + "learning_rate": 1e-06, + "loss": 0.7984, + "mean_token_accuracy": 0.7455000877380371, + "num_tokens": 447528900.0, + "step": 17596 + }, + { + "epoch": 1.9324621128925983, + "grad_norm": 2.2128634452819824, + "learning_rate": 1e-06, + "loss": 0.8647, + "mean_token_accuracy": 0.735531210899353, + "num_tokens": 447555758.0, + "step": 17597 + }, + { + "epoch": 1.9325719305952118, + "grad_norm": 2.329678773880005, + "learning_rate": 1e-06, + "loss": 0.7987, + "mean_token_accuracy": 0.7491366863250732, + "num_tokens": 447577679.0, + "step": 17598 + }, + { + "epoch": 1.9326817482978256, + "grad_norm": 1.9719117879867554, + "learning_rate": 1e-06, + "loss": 0.9331, + "mean_token_accuracy": 0.7145851850509644, + "num_tokens": 447610228.0, + "step": 17599 + }, + { + "epoch": 1.9327915660004393, + "grad_norm": 2.0229263305664062, + "learning_rate": 1e-06, + "loss": 0.8806, + "mean_token_accuracy": 0.7179633378982544, + "num_tokens": 447639140.0, + "step": 17600 + }, + { + "epoch": 1.932901383703053, + "grad_norm": 2.517411708831787, + "learning_rate": 1e-06, + "loss": 0.9002, + "mean_token_accuracy": 0.7275975346565247, + "num_tokens": 447660549.0, + "step": 17601 + }, + { + "epoch": 1.9330112014056666, + "grad_norm": 2.2482333183288574, + "learning_rate": 1e-06, + "loss": 0.8835, + "mean_token_accuracy": 0.7279149293899536, + "num_tokens": 447685971.0, + "step": 17602 + }, + { + "epoch": 1.9331210191082802, + "grad_norm": 2.1694953441619873, + "learning_rate": 1e-06, + "loss": 0.8381, + "mean_token_accuracy": 0.7336435317993164, + "num_tokens": 447714448.0, + "step": 17603 + }, + { + "epoch": 1.933230836810894, + "grad_norm": 2.396230697631836, + "learning_rate": 1e-06, + "loss": 0.9475, + "mean_token_accuracy": 0.7125802040100098, + "num_tokens": 447738482.0, + "step": 17604 + }, + { + "epoch": 1.9333406545135077, + "grad_norm": 1.971277117729187, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7059569954872131, + "num_tokens": 447771260.0, + "step": 17605 + }, + { + "epoch": 1.9334504722161212, + "grad_norm": 2.3847827911376953, + "learning_rate": 1e-06, + "loss": 0.9433, + "mean_token_accuracy": 0.7049073576927185, + "num_tokens": 447795458.0, + "step": 17606 + }, + { + "epoch": 1.9335602899187347, + "grad_norm": 2.237945795059204, + "learning_rate": 1e-06, + "loss": 0.9613, + "mean_token_accuracy": 0.7011312246322632, + "num_tokens": 447823704.0, + "step": 17607 + }, + { + "epoch": 1.9336701076213485, + "grad_norm": 2.570396900177002, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7331132292747498, + "num_tokens": 447844688.0, + "step": 17608 + }, + { + "epoch": 1.9337799253239623, + "grad_norm": 2.1825432777404785, + "learning_rate": 1e-06, + "loss": 0.9283, + "mean_token_accuracy": 0.7154197692871094, + "num_tokens": 447871911.0, + "step": 17609 + }, + { + "epoch": 1.933889743026576, + "grad_norm": 2.136859655380249, + "learning_rate": 1e-06, + "loss": 0.9263, + "mean_token_accuracy": 0.7094295620918274, + "num_tokens": 447901814.0, + "step": 17610 + }, + { + "epoch": 1.9339995607291895, + "grad_norm": 2.2614216804504395, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7098351120948792, + "num_tokens": 447928946.0, + "step": 17611 + }, + { + "epoch": 1.934109378431803, + "grad_norm": 2.3954882621765137, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7249447107315063, + "num_tokens": 447950589.0, + "step": 17612 + }, + { + "epoch": 1.9342191961344168, + "grad_norm": 2.373126745223999, + "learning_rate": 1e-06, + "loss": 0.8506, + "mean_token_accuracy": 0.7281980514526367, + "num_tokens": 447972077.0, + "step": 17613 + }, + { + "epoch": 1.9343290138370306, + "grad_norm": 2.34018874168396, + "learning_rate": 1e-06, + "loss": 0.8279, + "mean_token_accuracy": 0.7379513382911682, + "num_tokens": 447995233.0, + "step": 17614 + }, + { + "epoch": 1.9344388315396444, + "grad_norm": 2.1596500873565674, + "learning_rate": 1e-06, + "loss": 0.7965, + "mean_token_accuracy": 0.7492808103561401, + "num_tokens": 448020299.0, + "step": 17615 + }, + { + "epoch": 1.934548649242258, + "grad_norm": 2.4275007247924805, + "learning_rate": 1e-06, + "loss": 0.8323, + "mean_token_accuracy": 0.740443766117096, + "num_tokens": 448042279.0, + "step": 17616 + }, + { + "epoch": 1.9346584669448714, + "grad_norm": 2.08382248878479, + "learning_rate": 1e-06, + "loss": 0.9413, + "mean_token_accuracy": 0.7115843296051025, + "num_tokens": 448070332.0, + "step": 17617 + }, + { + "epoch": 1.9347682846474852, + "grad_norm": 2.2549233436584473, + "learning_rate": 1e-06, + "loss": 0.9312, + "mean_token_accuracy": 0.712374746799469, + "num_tokens": 448097250.0, + "step": 17618 + }, + { + "epoch": 1.934878102350099, + "grad_norm": 2.2111551761627197, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7174903750419617, + "num_tokens": 448124001.0, + "step": 17619 + }, + { + "epoch": 1.9349879200527125, + "grad_norm": 2.458136796951294, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7326517701148987, + "num_tokens": 448144994.0, + "step": 17620 + }, + { + "epoch": 1.935097737755326, + "grad_norm": 2.2849488258361816, + "learning_rate": 1e-06, + "loss": 0.9422, + "mean_token_accuracy": 0.7106244564056396, + "num_tokens": 448171567.0, + "step": 17621 + }, + { + "epoch": 1.9352075554579398, + "grad_norm": 2.1397640705108643, + "learning_rate": 1e-06, + "loss": 0.9963, + "mean_token_accuracy": 0.7022483348846436, + "num_tokens": 448199978.0, + "step": 17622 + }, + { + "epoch": 1.9353173731605535, + "grad_norm": 2.4502477645874023, + "learning_rate": 1e-06, + "loss": 0.8665, + "mean_token_accuracy": 0.7222772836685181, + "num_tokens": 448220635.0, + "step": 17623 + }, + { + "epoch": 1.9354271908631673, + "grad_norm": 2.585059642791748, + "learning_rate": 1e-06, + "loss": 0.8392, + "mean_token_accuracy": 0.7334203720092773, + "num_tokens": 448240418.0, + "step": 17624 + }, + { + "epoch": 1.9355370085657808, + "grad_norm": 2.139122247695923, + "learning_rate": 1e-06, + "loss": 0.922, + "mean_token_accuracy": 0.7160906791687012, + "num_tokens": 448267859.0, + "step": 17625 + }, + { + "epoch": 1.9356468262683943, + "grad_norm": 2.272735834121704, + "learning_rate": 1e-06, + "loss": 0.8513, + "mean_token_accuracy": 0.7352838516235352, + "num_tokens": 448292034.0, + "step": 17626 + }, + { + "epoch": 1.935756643971008, + "grad_norm": 2.241366147994995, + "learning_rate": 1e-06, + "loss": 0.966, + "mean_token_accuracy": 0.6969648599624634, + "num_tokens": 448318038.0, + "step": 17627 + }, + { + "epoch": 1.9358664616736219, + "grad_norm": 2.17024564743042, + "learning_rate": 1e-06, + "loss": 0.9205, + "mean_token_accuracy": 0.7123169302940369, + "num_tokens": 448344581.0, + "step": 17628 + }, + { + "epoch": 1.9359762793762354, + "grad_norm": 2.160095691680908, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.728683590888977, + "num_tokens": 448370229.0, + "step": 17629 + }, + { + "epoch": 1.9360860970788492, + "grad_norm": 2.315631866455078, + "learning_rate": 1e-06, + "loss": 0.8533, + "mean_token_accuracy": 0.7336771488189697, + "num_tokens": 448394972.0, + "step": 17630 + }, + { + "epoch": 1.9361959147814627, + "grad_norm": 2.0175492763519287, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.6995723247528076, + "num_tokens": 448426872.0, + "step": 17631 + }, + { + "epoch": 1.9363057324840764, + "grad_norm": 2.508002281188965, + "learning_rate": 1e-06, + "loss": 0.9383, + "mean_token_accuracy": 0.7087655663490295, + "num_tokens": 448449733.0, + "step": 17632 + }, + { + "epoch": 1.9364155501866902, + "grad_norm": 2.2222559452056885, + "learning_rate": 1e-06, + "loss": 0.7757, + "mean_token_accuracy": 0.7535042762756348, + "num_tokens": 448472458.0, + "step": 17633 + }, + { + "epoch": 1.9365253678893037, + "grad_norm": 1.999157428741455, + "learning_rate": 1e-06, + "loss": 0.8785, + "mean_token_accuracy": 0.7241368293762207, + "num_tokens": 448505259.0, + "step": 17634 + }, + { + "epoch": 1.9366351855919173, + "grad_norm": 1.867337703704834, + "learning_rate": 1e-06, + "loss": 0.9015, + "mean_token_accuracy": 0.7129005789756775, + "num_tokens": 448537000.0, + "step": 17635 + }, + { + "epoch": 1.936745003294531, + "grad_norm": 2.099461317062378, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7172055244445801, + "num_tokens": 448563399.0, + "step": 17636 + }, + { + "epoch": 1.9368548209971448, + "grad_norm": 2.336674451828003, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7176458239555359, + "num_tokens": 448585713.0, + "step": 17637 + }, + { + "epoch": 1.9369646386997585, + "grad_norm": 2.560279369354248, + "learning_rate": 1e-06, + "loss": 0.7958, + "mean_token_accuracy": 0.7476385831832886, + "num_tokens": 448604766.0, + "step": 17638 + }, + { + "epoch": 1.937074456402372, + "grad_norm": 2.0450072288513184, + "learning_rate": 1e-06, + "loss": 0.8362, + "mean_token_accuracy": 0.7380569577217102, + "num_tokens": 448630757.0, + "step": 17639 + }, + { + "epoch": 1.9371842741049856, + "grad_norm": 2.357247829437256, + "learning_rate": 1e-06, + "loss": 0.9705, + "mean_token_accuracy": 0.7026103138923645, + "num_tokens": 448656328.0, + "step": 17640 + }, + { + "epoch": 1.9372940918075994, + "grad_norm": 2.248840570449829, + "learning_rate": 1e-06, + "loss": 1.0014, + "mean_token_accuracy": 0.6977760195732117, + "num_tokens": 448684395.0, + "step": 17641 + }, + { + "epoch": 1.9374039095102131, + "grad_norm": 2.1541855335235596, + "learning_rate": 1e-06, + "loss": 0.969, + "mean_token_accuracy": 0.7082067131996155, + "num_tokens": 448710385.0, + "step": 17642 + }, + { + "epoch": 1.9375137272128267, + "grad_norm": 2.4062023162841797, + "learning_rate": 1e-06, + "loss": 0.9762, + "mean_token_accuracy": 0.6980417966842651, + "num_tokens": 448733673.0, + "step": 17643 + }, + { + "epoch": 1.9376235449154404, + "grad_norm": 2.147998571395874, + "learning_rate": 1e-06, + "loss": 0.9177, + "mean_token_accuracy": 0.7141985297203064, + "num_tokens": 448761080.0, + "step": 17644 + }, + { + "epoch": 1.937733362618054, + "grad_norm": 2.2226076126098633, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7301527261734009, + "num_tokens": 448785686.0, + "step": 17645 + }, + { + "epoch": 1.9378431803206677, + "grad_norm": 1.873779058456421, + "learning_rate": 1e-06, + "loss": 0.9071, + "mean_token_accuracy": 0.7130941152572632, + "num_tokens": 448818792.0, + "step": 17646 + }, + { + "epoch": 1.9379529980232815, + "grad_norm": 2.158785343170166, + "learning_rate": 1e-06, + "loss": 0.8737, + "mean_token_accuracy": 0.7306736707687378, + "num_tokens": 448845642.0, + "step": 17647 + }, + { + "epoch": 1.938062815725895, + "grad_norm": 2.258981227874756, + "learning_rate": 1e-06, + "loss": 0.8869, + "mean_token_accuracy": 0.7193537950515747, + "num_tokens": 448871213.0, + "step": 17648 + }, + { + "epoch": 1.9381726334285085, + "grad_norm": 2.292100667953491, + "learning_rate": 1e-06, + "loss": 0.8964, + "mean_token_accuracy": 0.7213006019592285, + "num_tokens": 448894519.0, + "step": 17649 + }, + { + "epoch": 1.9382824511311223, + "grad_norm": 2.2657296657562256, + "learning_rate": 1e-06, + "loss": 0.8456, + "mean_token_accuracy": 0.7349204421043396, + "num_tokens": 448918566.0, + "step": 17650 + }, + { + "epoch": 1.938392268833736, + "grad_norm": 2.3589112758636475, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7203731536865234, + "num_tokens": 448939729.0, + "step": 17651 + }, + { + "epoch": 1.9385020865363498, + "grad_norm": 2.4414007663726807, + "learning_rate": 1e-06, + "loss": 0.9152, + "mean_token_accuracy": 0.7241230010986328, + "num_tokens": 448963016.0, + "step": 17652 + }, + { + "epoch": 1.9386119042389633, + "grad_norm": 2.396951675415039, + "learning_rate": 1e-06, + "loss": 0.9325, + "mean_token_accuracy": 0.7161095142364502, + "num_tokens": 448986161.0, + "step": 17653 + }, + { + "epoch": 1.9387217219415769, + "grad_norm": 2.2134666442871094, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7335812449455261, + "num_tokens": 449011901.0, + "step": 17654 + }, + { + "epoch": 1.9388315396441906, + "grad_norm": 2.1644043922424316, + "learning_rate": 1e-06, + "loss": 0.9753, + "mean_token_accuracy": 0.7038161754608154, + "num_tokens": 449040208.0, + "step": 17655 + }, + { + "epoch": 1.9389413573468044, + "grad_norm": 2.614232301712036, + "learning_rate": 1e-06, + "loss": 0.8265, + "mean_token_accuracy": 0.7381401062011719, + "num_tokens": 449060337.0, + "step": 17656 + }, + { + "epoch": 1.939051175049418, + "grad_norm": 2.3171045780181885, + "learning_rate": 1e-06, + "loss": 0.9548, + "mean_token_accuracy": 0.70237135887146, + "num_tokens": 449084985.0, + "step": 17657 + }, + { + "epoch": 1.9391609927520315, + "grad_norm": 2.5162858963012695, + "learning_rate": 1e-06, + "loss": 0.9627, + "mean_token_accuracy": 0.7110438346862793, + "num_tokens": 449106329.0, + "step": 17658 + }, + { + "epoch": 1.9392708104546452, + "grad_norm": 1.8450731039047241, + "learning_rate": 1e-06, + "loss": 0.9215, + "mean_token_accuracy": 0.722541093826294, + "num_tokens": 449138292.0, + "step": 17659 + }, + { + "epoch": 1.939380628157259, + "grad_norm": 2.1171586513519287, + "learning_rate": 1e-06, + "loss": 1.0233, + "mean_token_accuracy": 0.697324812412262, + "num_tokens": 449167144.0, + "step": 17660 + }, + { + "epoch": 1.9394904458598727, + "grad_norm": 2.283787488937378, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7426518201828003, + "num_tokens": 449189879.0, + "step": 17661 + }, + { + "epoch": 1.9396002635624863, + "grad_norm": 2.439704418182373, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7237308621406555, + "num_tokens": 449211218.0, + "step": 17662 + }, + { + "epoch": 1.9397100812650998, + "grad_norm": 2.4933366775512695, + "learning_rate": 1e-06, + "loss": 0.7115, + "mean_token_accuracy": 0.768338680267334, + "num_tokens": 449229997.0, + "step": 17663 + }, + { + "epoch": 1.9398198989677136, + "grad_norm": 2.4332802295684814, + "learning_rate": 1e-06, + "loss": 0.7344, + "mean_token_accuracy": 0.7562952041625977, + "num_tokens": 449249356.0, + "step": 17664 + }, + { + "epoch": 1.9399297166703273, + "grad_norm": 2.328906297683716, + "learning_rate": 1e-06, + "loss": 0.9487, + "mean_token_accuracy": 0.7043658494949341, + "num_tokens": 449273366.0, + "step": 17665 + }, + { + "epoch": 1.940039534372941, + "grad_norm": 2.0743653774261475, + "learning_rate": 1e-06, + "loss": 0.8776, + "mean_token_accuracy": 0.7352256774902344, + "num_tokens": 449301183.0, + "step": 17666 + }, + { + "epoch": 1.9401493520755546, + "grad_norm": 2.167588710784912, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7123147249221802, + "num_tokens": 449328735.0, + "step": 17667 + }, + { + "epoch": 1.9402591697781681, + "grad_norm": 2.4588472843170166, + "learning_rate": 1e-06, + "loss": 0.9119, + "mean_token_accuracy": 0.7232222557067871, + "num_tokens": 449351179.0, + "step": 17668 + }, + { + "epoch": 1.940368987480782, + "grad_norm": 2.461303949356079, + "learning_rate": 1e-06, + "loss": 0.8308, + "mean_token_accuracy": 0.7437865734100342, + "num_tokens": 449373358.0, + "step": 17669 + }, + { + "epoch": 1.9404788051833957, + "grad_norm": 2.6611886024475098, + "learning_rate": 1e-06, + "loss": 0.873, + "mean_token_accuracy": 0.7241250276565552, + "num_tokens": 449393089.0, + "step": 17670 + }, + { + "epoch": 1.9405886228860092, + "grad_norm": 2.1535844802856445, + "learning_rate": 1e-06, + "loss": 0.9432, + "mean_token_accuracy": 0.7058815956115723, + "num_tokens": 449419914.0, + "step": 17671 + }, + { + "epoch": 1.9406984405886227, + "grad_norm": 2.226827621459961, + "learning_rate": 1e-06, + "loss": 0.8765, + "mean_token_accuracy": 0.7198687791824341, + "num_tokens": 449443910.0, + "step": 17672 + }, + { + "epoch": 1.9408082582912365, + "grad_norm": 2.1774001121520996, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.7037758827209473, + "num_tokens": 449470055.0, + "step": 17673 + }, + { + "epoch": 1.9409180759938502, + "grad_norm": 1.9793322086334229, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7050499320030212, + "num_tokens": 449501753.0, + "step": 17674 + }, + { + "epoch": 1.941027893696464, + "grad_norm": 2.1378180980682373, + "learning_rate": 1e-06, + "loss": 0.9498, + "mean_token_accuracy": 0.7043859958648682, + "num_tokens": 449529944.0, + "step": 17675 + }, + { + "epoch": 1.9411377113990775, + "grad_norm": 2.3797309398651123, + "learning_rate": 1e-06, + "loss": 0.8682, + "mean_token_accuracy": 0.7276645302772522, + "num_tokens": 449551580.0, + "step": 17676 + }, + { + "epoch": 1.941247529101691, + "grad_norm": 2.2363784313201904, + "learning_rate": 1e-06, + "loss": 0.8814, + "mean_token_accuracy": 0.7250640392303467, + "num_tokens": 449577975.0, + "step": 17677 + }, + { + "epoch": 1.9413573468043048, + "grad_norm": 2.1439685821533203, + "learning_rate": 1e-06, + "loss": 0.9887, + "mean_token_accuracy": 0.6927579045295715, + "num_tokens": 449607456.0, + "step": 17678 + }, + { + "epoch": 1.9414671645069186, + "grad_norm": 2.4823684692382812, + "learning_rate": 1e-06, + "loss": 0.7613, + "mean_token_accuracy": 0.7567689418792725, + "num_tokens": 449626572.0, + "step": 17679 + }, + { + "epoch": 1.9415769822095323, + "grad_norm": 2.1180214881896973, + "learning_rate": 1e-06, + "loss": 0.9083, + "mean_token_accuracy": 0.726004958152771, + "num_tokens": 449653141.0, + "step": 17680 + }, + { + "epoch": 1.9416867999121459, + "grad_norm": 2.46799373626709, + "learning_rate": 1e-06, + "loss": 0.8648, + "mean_token_accuracy": 0.7277622222900391, + "num_tokens": 449673998.0, + "step": 17681 + }, + { + "epoch": 1.9417966176147594, + "grad_norm": 2.4128470420837402, + "learning_rate": 1e-06, + "loss": 0.8941, + "mean_token_accuracy": 0.7330440282821655, + "num_tokens": 449694593.0, + "step": 17682 + }, + { + "epoch": 1.9419064353173732, + "grad_norm": 1.9865474700927734, + "learning_rate": 1e-06, + "loss": 0.904, + "mean_token_accuracy": 0.7165554165840149, + "num_tokens": 449724347.0, + "step": 17683 + }, + { + "epoch": 1.942016253019987, + "grad_norm": 2.3742754459381104, + "learning_rate": 1e-06, + "loss": 0.8707, + "mean_token_accuracy": 0.7317323684692383, + "num_tokens": 449746858.0, + "step": 17684 + }, + { + "epoch": 1.9421260707226005, + "grad_norm": 2.2020516395568848, + "learning_rate": 1e-06, + "loss": 0.8067, + "mean_token_accuracy": 0.7502138018608093, + "num_tokens": 449771870.0, + "step": 17685 + }, + { + "epoch": 1.942235888425214, + "grad_norm": 2.2124407291412354, + "learning_rate": 1e-06, + "loss": 0.8583, + "mean_token_accuracy": 0.731346070766449, + "num_tokens": 449800129.0, + "step": 17686 + }, + { + "epoch": 1.9423457061278278, + "grad_norm": 2.431607961654663, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7183043956756592, + "num_tokens": 449823488.0, + "step": 17687 + }, + { + "epoch": 1.9424555238304415, + "grad_norm": 2.046875238418579, + "learning_rate": 1e-06, + "loss": 0.9285, + "mean_token_accuracy": 0.7103789448738098, + "num_tokens": 449853057.0, + "step": 17688 + }, + { + "epoch": 1.9425653415330553, + "grad_norm": 2.6618406772613525, + "learning_rate": 1e-06, + "loss": 0.8768, + "mean_token_accuracy": 0.7293258905410767, + "num_tokens": 449872835.0, + "step": 17689 + }, + { + "epoch": 1.9426751592356688, + "grad_norm": 2.1356139183044434, + "learning_rate": 1e-06, + "loss": 0.9064, + "mean_token_accuracy": 0.7231460809707642, + "num_tokens": 449902834.0, + "step": 17690 + }, + { + "epoch": 1.9427849769382823, + "grad_norm": 2.3855340480804443, + "learning_rate": 1e-06, + "loss": 0.8353, + "mean_token_accuracy": 0.7332524061203003, + "num_tokens": 449925418.0, + "step": 17691 + }, + { + "epoch": 1.942894794640896, + "grad_norm": 2.0225272178649902, + "learning_rate": 1e-06, + "loss": 0.8874, + "mean_token_accuracy": 0.7216079235076904, + "num_tokens": 449956794.0, + "step": 17692 + }, + { + "epoch": 1.9430046123435099, + "grad_norm": 2.0926907062530518, + "learning_rate": 1e-06, + "loss": 0.9469, + "mean_token_accuracy": 0.7099112272262573, + "num_tokens": 449986078.0, + "step": 17693 + }, + { + "epoch": 1.9431144300461234, + "grad_norm": 1.9179662466049194, + "learning_rate": 1e-06, + "loss": 0.8631, + "mean_token_accuracy": 0.7276007533073425, + "num_tokens": 450017977.0, + "step": 17694 + }, + { + "epoch": 1.9432242477487371, + "grad_norm": 2.380338668823242, + "learning_rate": 1e-06, + "loss": 0.8458, + "mean_token_accuracy": 0.7324744462966919, + "num_tokens": 450038962.0, + "step": 17695 + }, + { + "epoch": 1.9433340654513507, + "grad_norm": 2.741605520248413, + "learning_rate": 1e-06, + "loss": 0.9112, + "mean_token_accuracy": 0.7189165949821472, + "num_tokens": 450057122.0, + "step": 17696 + }, + { + "epoch": 1.9434438831539644, + "grad_norm": 2.3381459712982178, + "learning_rate": 1e-06, + "loss": 0.8865, + "mean_token_accuracy": 0.7225481271743774, + "num_tokens": 450081499.0, + "step": 17697 + }, + { + "epoch": 1.9435537008565782, + "grad_norm": 2.3191676139831543, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7134968042373657, + "num_tokens": 450107528.0, + "step": 17698 + }, + { + "epoch": 1.9436635185591917, + "grad_norm": 2.1742186546325684, + "learning_rate": 1e-06, + "loss": 0.8439, + "mean_token_accuracy": 0.7425864934921265, + "num_tokens": 450131335.0, + "step": 17699 + }, + { + "epoch": 1.9437733362618053, + "grad_norm": 2.0690011978149414, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.703999400138855, + "num_tokens": 450161564.0, + "step": 17700 + }, + { + "epoch": 1.943883153964419, + "grad_norm": 2.3379483222961426, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.7206591367721558, + "num_tokens": 450184241.0, + "step": 17701 + }, + { + "epoch": 1.9439929716670328, + "grad_norm": 2.0689971446990967, + "learning_rate": 1e-06, + "loss": 0.7745, + "mean_token_accuracy": 0.7500638961791992, + "num_tokens": 450210632.0, + "step": 17702 + }, + { + "epoch": 1.9441027893696465, + "grad_norm": 2.1126317977905273, + "learning_rate": 1e-06, + "loss": 0.9343, + "mean_token_accuracy": 0.7097266912460327, + "num_tokens": 450239102.0, + "step": 17703 + }, + { + "epoch": 1.94421260707226, + "grad_norm": 2.285752773284912, + "learning_rate": 1e-06, + "loss": 0.9458, + "mean_token_accuracy": 0.7087405920028687, + "num_tokens": 450265788.0, + "step": 17704 + }, + { + "epoch": 1.9443224247748736, + "grad_norm": 2.1781206130981445, + "learning_rate": 1e-06, + "loss": 0.8905, + "mean_token_accuracy": 0.7240498661994934, + "num_tokens": 450291461.0, + "step": 17705 + }, + { + "epoch": 1.9444322424774874, + "grad_norm": 2.6069064140319824, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.733491063117981, + "num_tokens": 450310277.0, + "step": 17706 + }, + { + "epoch": 1.9445420601801011, + "grad_norm": 2.5938608646392822, + "learning_rate": 1e-06, + "loss": 0.8433, + "mean_token_accuracy": 0.731418788433075, + "num_tokens": 450330576.0, + "step": 17707 + }, + { + "epoch": 1.9446518778827147, + "grad_norm": 2.151458978652954, + "learning_rate": 1e-06, + "loss": 0.9863, + "mean_token_accuracy": 0.701677680015564, + "num_tokens": 450360802.0, + "step": 17708 + }, + { + "epoch": 1.9447616955853284, + "grad_norm": 2.4053359031677246, + "learning_rate": 1e-06, + "loss": 0.8351, + "mean_token_accuracy": 0.7361607551574707, + "num_tokens": 450382159.0, + "step": 17709 + }, + { + "epoch": 1.944871513287942, + "grad_norm": 2.19755220413208, + "learning_rate": 1e-06, + "loss": 0.9497, + "mean_token_accuracy": 0.7117406129837036, + "num_tokens": 450408463.0, + "step": 17710 + }, + { + "epoch": 1.9449813309905557, + "grad_norm": 2.2695605754852295, + "learning_rate": 1e-06, + "loss": 0.9521, + "mean_token_accuracy": 0.7199931144714355, + "num_tokens": 450433125.0, + "step": 17711 + }, + { + "epoch": 1.9450911486931695, + "grad_norm": 2.31024169921875, + "learning_rate": 1e-06, + "loss": 0.9153, + "mean_token_accuracy": 0.7194734811782837, + "num_tokens": 450457910.0, + "step": 17712 + }, + { + "epoch": 1.945200966395783, + "grad_norm": 2.272233009338379, + "learning_rate": 1e-06, + "loss": 0.9758, + "mean_token_accuracy": 0.7052866220474243, + "num_tokens": 450483294.0, + "step": 17713 + }, + { + "epoch": 1.9453107840983965, + "grad_norm": 1.9375133514404297, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7156797647476196, + "num_tokens": 450514863.0, + "step": 17714 + }, + { + "epoch": 1.9454206018010103, + "grad_norm": 2.284527063369751, + "learning_rate": 1e-06, + "loss": 0.9405, + "mean_token_accuracy": 0.7150370478630066, + "num_tokens": 450538342.0, + "step": 17715 + }, + { + "epoch": 1.945530419503624, + "grad_norm": 2.445615291595459, + "learning_rate": 1e-06, + "loss": 0.9038, + "mean_token_accuracy": 0.7226385474205017, + "num_tokens": 450560890.0, + "step": 17716 + }, + { + "epoch": 1.9456402372062378, + "grad_norm": 2.0293643474578857, + "learning_rate": 1e-06, + "loss": 0.9429, + "mean_token_accuracy": 0.7064244747161865, + "num_tokens": 450591821.0, + "step": 17717 + }, + { + "epoch": 1.9457500549088513, + "grad_norm": 2.738180637359619, + "learning_rate": 1e-06, + "loss": 0.8488, + "mean_token_accuracy": 0.7286025881767273, + "num_tokens": 450610391.0, + "step": 17718 + }, + { + "epoch": 1.9458598726114649, + "grad_norm": 2.0729668140411377, + "learning_rate": 1e-06, + "loss": 0.9298, + "mean_token_accuracy": 0.7108407020568848, + "num_tokens": 450639534.0, + "step": 17719 + }, + { + "epoch": 1.9459696903140786, + "grad_norm": 2.5690126419067383, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7259669303894043, + "num_tokens": 450659176.0, + "step": 17720 + }, + { + "epoch": 1.9460795080166924, + "grad_norm": 2.271972179412842, + "learning_rate": 1e-06, + "loss": 0.9302, + "mean_token_accuracy": 0.7085566520690918, + "num_tokens": 450685467.0, + "step": 17721 + }, + { + "epoch": 1.946189325719306, + "grad_norm": 1.954036831855774, + "learning_rate": 1e-06, + "loss": 0.8936, + "mean_token_accuracy": 0.720468282699585, + "num_tokens": 450715973.0, + "step": 17722 + }, + { + "epoch": 1.9462991434219195, + "grad_norm": 2.7058701515197754, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7339397668838501, + "num_tokens": 450734135.0, + "step": 17723 + }, + { + "epoch": 1.9464089611245332, + "grad_norm": 2.051888942718506, + "learning_rate": 1e-06, + "loss": 0.8186, + "mean_token_accuracy": 0.7429523468017578, + "num_tokens": 450760456.0, + "step": 17724 + }, + { + "epoch": 1.946518778827147, + "grad_norm": 2.4870641231536865, + "learning_rate": 1e-06, + "loss": 0.8329, + "mean_token_accuracy": 0.7375299334526062, + "num_tokens": 450781637.0, + "step": 17725 + }, + { + "epoch": 1.9466285965297607, + "grad_norm": 2.38476824760437, + "learning_rate": 1e-06, + "loss": 0.878, + "mean_token_accuracy": 0.7236872315406799, + "num_tokens": 450804306.0, + "step": 17726 + }, + { + "epoch": 1.9467384142323743, + "grad_norm": 2.132420539855957, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7422991991043091, + "num_tokens": 450829176.0, + "step": 17727 + }, + { + "epoch": 1.9468482319349878, + "grad_norm": 2.2655293941497803, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7196301817893982, + "num_tokens": 450854607.0, + "step": 17728 + }, + { + "epoch": 1.9469580496376016, + "grad_norm": 2.1909940242767334, + "learning_rate": 1e-06, + "loss": 0.9036, + "mean_token_accuracy": 0.7340956926345825, + "num_tokens": 450880213.0, + "step": 17729 + }, + { + "epoch": 1.9470678673402153, + "grad_norm": 2.9198713302612305, + "learning_rate": 1e-06, + "loss": 0.8369, + "mean_token_accuracy": 0.7319941520690918, + "num_tokens": 450896315.0, + "step": 17730 + }, + { + "epoch": 1.947177685042829, + "grad_norm": 2.4107978343963623, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7039793133735657, + "num_tokens": 450920854.0, + "step": 17731 + }, + { + "epoch": 1.9472875027454426, + "grad_norm": 2.4990572929382324, + "learning_rate": 1e-06, + "loss": 0.8514, + "mean_token_accuracy": 0.7282443046569824, + "num_tokens": 450942521.0, + "step": 17732 + }, + { + "epoch": 1.9473973204480561, + "grad_norm": 2.3001856803894043, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7248072624206543, + "num_tokens": 450967900.0, + "step": 17733 + }, + { + "epoch": 1.94750713815067, + "grad_norm": 2.4110567569732666, + "learning_rate": 1e-06, + "loss": 0.7249, + "mean_token_accuracy": 0.7685394287109375, + "num_tokens": 450988759.0, + "step": 17734 + }, + { + "epoch": 1.9476169558532836, + "grad_norm": 2.589369535446167, + "learning_rate": 1e-06, + "loss": 0.7322, + "mean_token_accuracy": 0.7615867257118225, + "num_tokens": 451006724.0, + "step": 17735 + }, + { + "epoch": 1.9477267735558972, + "grad_norm": 2.2481257915496826, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7134596109390259, + "num_tokens": 451031528.0, + "step": 17736 + }, + { + "epoch": 1.9478365912585107, + "grad_norm": 2.5423383712768555, + "learning_rate": 1e-06, + "loss": 0.859, + "mean_token_accuracy": 0.7234776616096497, + "num_tokens": 451052936.0, + "step": 17737 + }, + { + "epoch": 1.9479464089611245, + "grad_norm": 1.9790843725204468, + "learning_rate": 1e-06, + "loss": 0.9763, + "mean_token_accuracy": 0.7047497630119324, + "num_tokens": 451083050.0, + "step": 17738 + }, + { + "epoch": 1.9480562266637382, + "grad_norm": 2.298696517944336, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7233907580375671, + "num_tokens": 451108026.0, + "step": 17739 + }, + { + "epoch": 1.948166044366352, + "grad_norm": 2.4572041034698486, + "learning_rate": 1e-06, + "loss": 0.9114, + "mean_token_accuracy": 0.7114232778549194, + "num_tokens": 451130088.0, + "step": 17740 + }, + { + "epoch": 1.9482758620689655, + "grad_norm": 2.153090238571167, + "learning_rate": 1e-06, + "loss": 0.8469, + "mean_token_accuracy": 0.7344802021980286, + "num_tokens": 451154877.0, + "step": 17741 + }, + { + "epoch": 1.948385679771579, + "grad_norm": 2.1482627391815186, + "learning_rate": 1e-06, + "loss": 0.9148, + "mean_token_accuracy": 0.7180037498474121, + "num_tokens": 451180765.0, + "step": 17742 + }, + { + "epoch": 1.9484954974741928, + "grad_norm": 2.4322335720062256, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7413932085037231, + "num_tokens": 451202950.0, + "step": 17743 + }, + { + "epoch": 1.9486053151768066, + "grad_norm": 2.186436414718628, + "learning_rate": 1e-06, + "loss": 0.8531, + "mean_token_accuracy": 0.7372003793716431, + "num_tokens": 451230265.0, + "step": 17744 + }, + { + "epoch": 1.94871513287942, + "grad_norm": 2.1911213397979736, + "learning_rate": 1e-06, + "loss": 0.8309, + "mean_token_accuracy": 0.7377661466598511, + "num_tokens": 451254369.0, + "step": 17745 + }, + { + "epoch": 1.9488249505820339, + "grad_norm": 2.154114007949829, + "learning_rate": 1e-06, + "loss": 0.8885, + "mean_token_accuracy": 0.7279046773910522, + "num_tokens": 451281658.0, + "step": 17746 + }, + { + "epoch": 1.9489347682846474, + "grad_norm": 2.0918941497802734, + "learning_rate": 1e-06, + "loss": 0.8734, + "mean_token_accuracy": 0.7301148176193237, + "num_tokens": 451309622.0, + "step": 17747 + }, + { + "epoch": 1.9490445859872612, + "grad_norm": 2.289665937423706, + "learning_rate": 1e-06, + "loss": 0.9046, + "mean_token_accuracy": 0.723660945892334, + "num_tokens": 451333445.0, + "step": 17748 + }, + { + "epoch": 1.949154403689875, + "grad_norm": 2.4477555751800537, + "learning_rate": 1e-06, + "loss": 0.8498, + "mean_token_accuracy": 0.7288331985473633, + "num_tokens": 451355181.0, + "step": 17749 + }, + { + "epoch": 1.9492642213924884, + "grad_norm": 1.9115829467773438, + "learning_rate": 1e-06, + "loss": 0.8376, + "mean_token_accuracy": 0.7369577884674072, + "num_tokens": 451387719.0, + "step": 17750 + }, + { + "epoch": 1.949374039095102, + "grad_norm": 2.217320442199707, + "learning_rate": 1e-06, + "loss": 0.8693, + "mean_token_accuracy": 0.7236192226409912, + "num_tokens": 451415118.0, + "step": 17751 + }, + { + "epoch": 1.9494838567977157, + "grad_norm": 2.574139356613159, + "learning_rate": 1e-06, + "loss": 0.7933, + "mean_token_accuracy": 0.7444631457328796, + "num_tokens": 451434563.0, + "step": 17752 + }, + { + "epoch": 1.9495936745003295, + "grad_norm": 2.1063690185546875, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7072626352310181, + "num_tokens": 451465179.0, + "step": 17753 + }, + { + "epoch": 1.9497034922029433, + "grad_norm": 2.448014259338379, + "learning_rate": 1e-06, + "loss": 0.8157, + "mean_token_accuracy": 0.7411246299743652, + "num_tokens": 451485297.0, + "step": 17754 + }, + { + "epoch": 1.9498133099055568, + "grad_norm": 2.1699702739715576, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.7153834700584412, + "num_tokens": 451510334.0, + "step": 17755 + }, + { + "epoch": 1.9499231276081703, + "grad_norm": 1.8814224004745483, + "learning_rate": 1e-06, + "loss": 0.937, + "mean_token_accuracy": 0.7074976563453674, + "num_tokens": 451544303.0, + "step": 17756 + }, + { + "epoch": 1.950032945310784, + "grad_norm": 2.56248140335083, + "learning_rate": 1e-06, + "loss": 0.808, + "mean_token_accuracy": 0.7452960014343262, + "num_tokens": 451563211.0, + "step": 17757 + }, + { + "epoch": 1.9501427630133978, + "grad_norm": 2.039555788040161, + "learning_rate": 1e-06, + "loss": 0.8828, + "mean_token_accuracy": 0.7254654765129089, + "num_tokens": 451593060.0, + "step": 17758 + }, + { + "epoch": 1.9502525807160114, + "grad_norm": 2.3358302116394043, + "learning_rate": 1e-06, + "loss": 0.9342, + "mean_token_accuracy": 0.7074071168899536, + "num_tokens": 451615895.0, + "step": 17759 + }, + { + "epoch": 1.9503623984186251, + "grad_norm": 2.2768967151641846, + "learning_rate": 1e-06, + "loss": 0.9982, + "mean_token_accuracy": 0.6898115873336792, + "num_tokens": 451642046.0, + "step": 17760 + }, + { + "epoch": 1.9504722161212387, + "grad_norm": 2.4690160751342773, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7364678382873535, + "num_tokens": 451663514.0, + "step": 17761 + }, + { + "epoch": 1.9505820338238524, + "grad_norm": 2.234678030014038, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.7056249976158142, + "num_tokens": 451688396.0, + "step": 17762 + }, + { + "epoch": 1.9506918515264662, + "grad_norm": 1.935636281967163, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7074063420295715, + "num_tokens": 451720859.0, + "step": 17763 + }, + { + "epoch": 1.9508016692290797, + "grad_norm": 2.154719114303589, + "learning_rate": 1e-06, + "loss": 0.8937, + "mean_token_accuracy": 0.7196125388145447, + "num_tokens": 451748579.0, + "step": 17764 + }, + { + "epoch": 1.9509114869316933, + "grad_norm": 2.077115535736084, + "learning_rate": 1e-06, + "loss": 0.9299, + "mean_token_accuracy": 0.708621084690094, + "num_tokens": 451776580.0, + "step": 17765 + }, + { + "epoch": 1.951021304634307, + "grad_norm": 2.2481143474578857, + "learning_rate": 1e-06, + "loss": 0.8176, + "mean_token_accuracy": 0.7393842935562134, + "num_tokens": 451801526.0, + "step": 17766 + }, + { + "epoch": 1.9511311223369208, + "grad_norm": 2.456191301345825, + "learning_rate": 1e-06, + "loss": 0.7871, + "mean_token_accuracy": 0.741095781326294, + "num_tokens": 451824500.0, + "step": 17767 + }, + { + "epoch": 1.9512409400395345, + "grad_norm": 2.608259916305542, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7172483205795288, + "num_tokens": 451845116.0, + "step": 17768 + }, + { + "epoch": 1.951350757742148, + "grad_norm": 2.39569354057312, + "learning_rate": 1e-06, + "loss": 0.8174, + "mean_token_accuracy": 0.7354476451873779, + "num_tokens": 451868000.0, + "step": 17769 + }, + { + "epoch": 1.9514605754447616, + "grad_norm": 2.1840503215789795, + "learning_rate": 1e-06, + "loss": 0.8949, + "mean_token_accuracy": 0.7193843126296997, + "num_tokens": 451895678.0, + "step": 17770 + }, + { + "epoch": 1.9515703931473753, + "grad_norm": 2.3197073936462402, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7220929861068726, + "num_tokens": 451919846.0, + "step": 17771 + }, + { + "epoch": 1.951680210849989, + "grad_norm": 1.9998664855957031, + "learning_rate": 1e-06, + "loss": 0.9085, + "mean_token_accuracy": 0.7237236499786377, + "num_tokens": 451952246.0, + "step": 17772 + }, + { + "epoch": 1.9517900285526026, + "grad_norm": 2.3940980434417725, + "learning_rate": 1e-06, + "loss": 0.7446, + "mean_token_accuracy": 0.7571018934249878, + "num_tokens": 451972161.0, + "step": 17773 + }, + { + "epoch": 1.9518998462552162, + "grad_norm": 2.0126564502716064, + "learning_rate": 1e-06, + "loss": 0.843, + "mean_token_accuracy": 0.7438841462135315, + "num_tokens": 452000532.0, + "step": 17774 + }, + { + "epoch": 1.95200966395783, + "grad_norm": 2.2406601905822754, + "learning_rate": 1e-06, + "loss": 0.9118, + "mean_token_accuracy": 0.7290998697280884, + "num_tokens": 452025746.0, + "step": 17775 + }, + { + "epoch": 1.9521194816604437, + "grad_norm": 1.973326563835144, + "learning_rate": 1e-06, + "loss": 1.0215, + "mean_token_accuracy": 0.6891757249832153, + "num_tokens": 452057991.0, + "step": 17776 + }, + { + "epoch": 1.9522292993630574, + "grad_norm": 2.855414867401123, + "learning_rate": 1e-06, + "loss": 0.8077, + "mean_token_accuracy": 0.7399318814277649, + "num_tokens": 452074030.0, + "step": 17777 + }, + { + "epoch": 1.952339117065671, + "grad_norm": 2.0319859981536865, + "learning_rate": 1e-06, + "loss": 0.8191, + "mean_token_accuracy": 0.7402752041816711, + "num_tokens": 452103814.0, + "step": 17778 + }, + { + "epoch": 1.9524489347682845, + "grad_norm": 2.085371971130371, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7225691080093384, + "num_tokens": 452131383.0, + "step": 17779 + }, + { + "epoch": 1.9525587524708983, + "grad_norm": 2.209562063217163, + "learning_rate": 1e-06, + "loss": 0.8773, + "mean_token_accuracy": 0.7247483730316162, + "num_tokens": 452157297.0, + "step": 17780 + }, + { + "epoch": 1.952668570173512, + "grad_norm": 2.550567865371704, + "learning_rate": 1e-06, + "loss": 0.863, + "mean_token_accuracy": 0.7250851392745972, + "num_tokens": 452178466.0, + "step": 17781 + }, + { + "epoch": 1.9527783878761258, + "grad_norm": 2.2334704399108887, + "learning_rate": 1e-06, + "loss": 0.8404, + "mean_token_accuracy": 0.7383227348327637, + "num_tokens": 452204977.0, + "step": 17782 + }, + { + "epoch": 1.9528882055787393, + "grad_norm": 2.620305299758911, + "learning_rate": 1e-06, + "loss": 0.8262, + "mean_token_accuracy": 0.7425675392150879, + "num_tokens": 452225542.0, + "step": 17783 + }, + { + "epoch": 1.9529980232813529, + "grad_norm": 2.0591371059417725, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7146905064582825, + "num_tokens": 452256759.0, + "step": 17784 + }, + { + "epoch": 1.9531078409839666, + "grad_norm": 2.5348753929138184, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7543380260467529, + "num_tokens": 452277116.0, + "step": 17785 + }, + { + "epoch": 1.9532176586865804, + "grad_norm": 2.340857982635498, + "learning_rate": 1e-06, + "loss": 0.9459, + "mean_token_accuracy": 0.707668125629425, + "num_tokens": 452303322.0, + "step": 17786 + }, + { + "epoch": 1.953327476389194, + "grad_norm": 2.2233715057373047, + "learning_rate": 1e-06, + "loss": 0.8519, + "mean_token_accuracy": 0.7402331829071045, + "num_tokens": 452329212.0, + "step": 17787 + }, + { + "epoch": 1.9534372940918074, + "grad_norm": 2.062394142150879, + "learning_rate": 1e-06, + "loss": 0.8813, + "mean_token_accuracy": 0.718749463558197, + "num_tokens": 452359022.0, + "step": 17788 + }, + { + "epoch": 1.9535471117944212, + "grad_norm": 1.9943872690200806, + "learning_rate": 1e-06, + "loss": 0.9226, + "mean_token_accuracy": 0.71206134557724, + "num_tokens": 452389228.0, + "step": 17789 + }, + { + "epoch": 1.953656929497035, + "grad_norm": 2.0437538623809814, + "learning_rate": 1e-06, + "loss": 0.9654, + "mean_token_accuracy": 0.7055374383926392, + "num_tokens": 452418663.0, + "step": 17790 + }, + { + "epoch": 1.9537667471996487, + "grad_norm": 2.4310214519500732, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.7390002012252808, + "num_tokens": 452441236.0, + "step": 17791 + }, + { + "epoch": 1.9538765649022622, + "grad_norm": 2.192803382873535, + "learning_rate": 1e-06, + "loss": 0.9505, + "mean_token_accuracy": 0.6993924379348755, + "num_tokens": 452469821.0, + "step": 17792 + }, + { + "epoch": 1.9539863826048758, + "grad_norm": 2.1656155586242676, + "learning_rate": 1e-06, + "loss": 0.8863, + "mean_token_accuracy": 0.720760703086853, + "num_tokens": 452497438.0, + "step": 17793 + }, + { + "epoch": 1.9540962003074895, + "grad_norm": 2.262059450149536, + "learning_rate": 1e-06, + "loss": 0.8391, + "mean_token_accuracy": 0.7393978238105774, + "num_tokens": 452519296.0, + "step": 17794 + }, + { + "epoch": 1.9542060180101033, + "grad_norm": 2.2921249866485596, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7407945394515991, + "num_tokens": 452543572.0, + "step": 17795 + }, + { + "epoch": 1.954315835712717, + "grad_norm": 2.0059924125671387, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7226277589797974, + "num_tokens": 452575198.0, + "step": 17796 + }, + { + "epoch": 1.9544256534153306, + "grad_norm": 2.171201705932617, + "learning_rate": 1e-06, + "loss": 0.9962, + "mean_token_accuracy": 0.7026057243347168, + "num_tokens": 452603559.0, + "step": 17797 + }, + { + "epoch": 1.9545354711179441, + "grad_norm": 2.459226608276367, + "learning_rate": 1e-06, + "loss": 0.8359, + "mean_token_accuracy": 0.7385106086730957, + "num_tokens": 452625941.0, + "step": 17798 + }, + { + "epoch": 1.9546452888205579, + "grad_norm": 2.3068854808807373, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.7228198647499084, + "num_tokens": 452650080.0, + "step": 17799 + }, + { + "epoch": 1.9547551065231716, + "grad_norm": 2.088716983795166, + "learning_rate": 1e-06, + "loss": 0.9988, + "mean_token_accuracy": 0.6886286735534668, + "num_tokens": 452679760.0, + "step": 17800 + }, + { + "epoch": 1.9548649242257852, + "grad_norm": 2.260999917984009, + "learning_rate": 1e-06, + "loss": 0.9164, + "mean_token_accuracy": 0.7118768692016602, + "num_tokens": 452706749.0, + "step": 17801 + }, + { + "epoch": 1.9549747419283987, + "grad_norm": 2.305417060852051, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7005264163017273, + "num_tokens": 452733628.0, + "step": 17802 + }, + { + "epoch": 1.9550845596310125, + "grad_norm": 2.470776081085205, + "learning_rate": 1e-06, + "loss": 0.814, + "mean_token_accuracy": 0.7387341260910034, + "num_tokens": 452757260.0, + "step": 17803 + }, + { + "epoch": 1.9551943773336262, + "grad_norm": 2.2324655055999756, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7088012099266052, + "num_tokens": 452783492.0, + "step": 17804 + }, + { + "epoch": 1.95530419503624, + "grad_norm": 2.275036334991455, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7154431343078613, + "num_tokens": 452810662.0, + "step": 17805 + }, + { + "epoch": 1.9554140127388535, + "grad_norm": 2.2683074474334717, + "learning_rate": 1e-06, + "loss": 0.9528, + "mean_token_accuracy": 0.7007577419281006, + "num_tokens": 452837471.0, + "step": 17806 + }, + { + "epoch": 1.955523830441467, + "grad_norm": 2.111384868621826, + "learning_rate": 1e-06, + "loss": 0.9574, + "mean_token_accuracy": 0.698001503944397, + "num_tokens": 452867748.0, + "step": 17807 + }, + { + "epoch": 1.9556336481440808, + "grad_norm": 2.4419004917144775, + "learning_rate": 1e-06, + "loss": 0.8747, + "mean_token_accuracy": 0.7251806855201721, + "num_tokens": 452889996.0, + "step": 17808 + }, + { + "epoch": 1.9557434658466946, + "grad_norm": 2.218801736831665, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7325165271759033, + "num_tokens": 452915343.0, + "step": 17809 + }, + { + "epoch": 1.955853283549308, + "grad_norm": 2.2146875858306885, + "learning_rate": 1e-06, + "loss": 0.9463, + "mean_token_accuracy": 0.7067452073097229, + "num_tokens": 452942882.0, + "step": 17810 + }, + { + "epoch": 1.9559631012519219, + "grad_norm": 2.6372454166412354, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7221693992614746, + "num_tokens": 452961689.0, + "step": 17811 + }, + { + "epoch": 1.9560729189545354, + "grad_norm": 2.097241163253784, + "learning_rate": 1e-06, + "loss": 0.9258, + "mean_token_accuracy": 0.7098076939582825, + "num_tokens": 452989995.0, + "step": 17812 + }, + { + "epoch": 1.9561827366571491, + "grad_norm": 2.2689905166625977, + "learning_rate": 1e-06, + "loss": 0.9856, + "mean_token_accuracy": 0.6967915892601013, + "num_tokens": 453018055.0, + "step": 17813 + }, + { + "epoch": 1.956292554359763, + "grad_norm": 2.1920390129089355, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.710877537727356, + "num_tokens": 453043227.0, + "step": 17814 + }, + { + "epoch": 1.9564023720623764, + "grad_norm": 2.372756004333496, + "learning_rate": 1e-06, + "loss": 0.9274, + "mean_token_accuracy": 0.7231281399726868, + "num_tokens": 453065954.0, + "step": 17815 + }, + { + "epoch": 1.95651218976499, + "grad_norm": 2.267862558364868, + "learning_rate": 1e-06, + "loss": 0.9766, + "mean_token_accuracy": 0.69865882396698, + "num_tokens": 453091736.0, + "step": 17816 + }, + { + "epoch": 1.9566220074676037, + "grad_norm": 2.2029829025268555, + "learning_rate": 1e-06, + "loss": 0.8998, + "mean_token_accuracy": 0.7218403220176697, + "num_tokens": 453117063.0, + "step": 17817 + }, + { + "epoch": 1.9567318251702175, + "grad_norm": 2.352804660797119, + "learning_rate": 1e-06, + "loss": 0.8536, + "mean_token_accuracy": 0.7303645014762878, + "num_tokens": 453140778.0, + "step": 17818 + }, + { + "epoch": 1.9568416428728312, + "grad_norm": 2.4375412464141846, + "learning_rate": 1e-06, + "loss": 0.8834, + "mean_token_accuracy": 0.7239395380020142, + "num_tokens": 453164894.0, + "step": 17819 + }, + { + "epoch": 1.9569514605754448, + "grad_norm": 2.0561516284942627, + "learning_rate": 1e-06, + "loss": 0.9566, + "mean_token_accuracy": 0.7018963098526001, + "num_tokens": 453198669.0, + "step": 17820 + }, + { + "epoch": 1.9570612782780583, + "grad_norm": 2.5824100971221924, + "learning_rate": 1e-06, + "loss": 0.7703, + "mean_token_accuracy": 0.7529283761978149, + "num_tokens": 453218181.0, + "step": 17821 + }, + { + "epoch": 1.957171095980672, + "grad_norm": 2.34714674949646, + "learning_rate": 1e-06, + "loss": 0.8731, + "mean_token_accuracy": 0.721581757068634, + "num_tokens": 453240526.0, + "step": 17822 + }, + { + "epoch": 1.9572809136832858, + "grad_norm": 2.5593807697296143, + "learning_rate": 1e-06, + "loss": 0.7606, + "mean_token_accuracy": 0.748031497001648, + "num_tokens": 453259821.0, + "step": 17823 + }, + { + "epoch": 1.9573907313858994, + "grad_norm": 2.4055910110473633, + "learning_rate": 1e-06, + "loss": 0.8113, + "mean_token_accuracy": 0.7446436285972595, + "num_tokens": 453281115.0, + "step": 17824 + }, + { + "epoch": 1.9575005490885131, + "grad_norm": 2.4618639945983887, + "learning_rate": 1e-06, + "loss": 0.9056, + "mean_token_accuracy": 0.7194547057151794, + "num_tokens": 453302023.0, + "step": 17825 + }, + { + "epoch": 1.9576103667911267, + "grad_norm": 2.5225064754486084, + "learning_rate": 1e-06, + "loss": 0.8227, + "mean_token_accuracy": 0.7415939569473267, + "num_tokens": 453322278.0, + "step": 17826 + }, + { + "epoch": 1.9577201844937404, + "grad_norm": 2.1354782581329346, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7360318303108215, + "num_tokens": 453348280.0, + "step": 17827 + }, + { + "epoch": 1.9578300021963542, + "grad_norm": 2.489084005355835, + "learning_rate": 1e-06, + "loss": 0.9551, + "mean_token_accuracy": 0.7002851963043213, + "num_tokens": 453370931.0, + "step": 17828 + }, + { + "epoch": 1.9579398198989677, + "grad_norm": 2.0768964290618896, + "learning_rate": 1e-06, + "loss": 0.9533, + "mean_token_accuracy": 0.7068222165107727, + "num_tokens": 453399408.0, + "step": 17829 + }, + { + "epoch": 1.9580496376015812, + "grad_norm": 2.381131649017334, + "learning_rate": 1e-06, + "loss": 0.8558, + "mean_token_accuracy": 0.7354233264923096, + "num_tokens": 453421997.0, + "step": 17830 + }, + { + "epoch": 1.958159455304195, + "grad_norm": 2.3009390830993652, + "learning_rate": 1e-06, + "loss": 0.9755, + "mean_token_accuracy": 0.7037047147750854, + "num_tokens": 453448104.0, + "step": 17831 + }, + { + "epoch": 1.9582692730068088, + "grad_norm": 2.209240436553955, + "learning_rate": 1e-06, + "loss": 0.8486, + "mean_token_accuracy": 0.7370160818099976, + "num_tokens": 453471627.0, + "step": 17832 + }, + { + "epoch": 1.9583790907094225, + "grad_norm": 1.9269847869873047, + "learning_rate": 1e-06, + "loss": 0.9102, + "mean_token_accuracy": 0.7207003235816956, + "num_tokens": 453505009.0, + "step": 17833 + }, + { + "epoch": 1.958488908412036, + "grad_norm": 2.157116651535034, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7345648407936096, + "num_tokens": 453531326.0, + "step": 17834 + }, + { + "epoch": 1.9585987261146496, + "grad_norm": 2.1117401123046875, + "learning_rate": 1e-06, + "loss": 0.8995, + "mean_token_accuracy": 0.7166672348976135, + "num_tokens": 453558589.0, + "step": 17835 + }, + { + "epoch": 1.9587085438172633, + "grad_norm": 2.341616630554199, + "learning_rate": 1e-06, + "loss": 0.8926, + "mean_token_accuracy": 0.725314736366272, + "num_tokens": 453583076.0, + "step": 17836 + }, + { + "epoch": 1.958818361519877, + "grad_norm": 2.05952787399292, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7130537629127502, + "num_tokens": 453613033.0, + "step": 17837 + }, + { + "epoch": 1.9589281792224906, + "grad_norm": 2.04233717918396, + "learning_rate": 1e-06, + "loss": 0.9132, + "mean_token_accuracy": 0.7130540013313293, + "num_tokens": 453642592.0, + "step": 17838 + }, + { + "epoch": 1.9590379969251042, + "grad_norm": 2.190964698791504, + "learning_rate": 1e-06, + "loss": 0.8567, + "mean_token_accuracy": 0.7253556251525879, + "num_tokens": 453668394.0, + "step": 17839 + }, + { + "epoch": 1.959147814627718, + "grad_norm": 2.083202362060547, + "learning_rate": 1e-06, + "loss": 0.8738, + "mean_token_accuracy": 0.7281404137611389, + "num_tokens": 453694241.0, + "step": 17840 + }, + { + "epoch": 1.9592576323303317, + "grad_norm": 2.120837450027466, + "learning_rate": 1e-06, + "loss": 0.8316, + "mean_token_accuracy": 0.7403080463409424, + "num_tokens": 453721873.0, + "step": 17841 + }, + { + "epoch": 1.9593674500329454, + "grad_norm": 2.1361660957336426, + "learning_rate": 1e-06, + "loss": 0.9443, + "mean_token_accuracy": 0.7101126313209534, + "num_tokens": 453750969.0, + "step": 17842 + }, + { + "epoch": 1.959477267735559, + "grad_norm": 2.2732656002044678, + "learning_rate": 1e-06, + "loss": 0.9113, + "mean_token_accuracy": 0.7164230942726135, + "num_tokens": 453776241.0, + "step": 17843 + }, + { + "epoch": 1.9595870854381725, + "grad_norm": 2.242204189300537, + "learning_rate": 1e-06, + "loss": 0.9273, + "mean_token_accuracy": 0.7102511525154114, + "num_tokens": 453802634.0, + "step": 17844 + }, + { + "epoch": 1.9596969031407863, + "grad_norm": 2.08349347114563, + "learning_rate": 1e-06, + "loss": 0.9423, + "mean_token_accuracy": 0.7091456651687622, + "num_tokens": 453830712.0, + "step": 17845 + }, + { + "epoch": 1.9598067208434, + "grad_norm": 2.3330137729644775, + "learning_rate": 1e-06, + "loss": 0.9252, + "mean_token_accuracy": 0.7082518339157104, + "num_tokens": 453854292.0, + "step": 17846 + }, + { + "epoch": 1.9599165385460138, + "grad_norm": 2.6537630558013916, + "learning_rate": 1e-06, + "loss": 0.8633, + "mean_token_accuracy": 0.7243667840957642, + "num_tokens": 453873359.0, + "step": 17847 + }, + { + "epoch": 1.9600263562486273, + "grad_norm": 2.4060163497924805, + "learning_rate": 1e-06, + "loss": 0.9213, + "mean_token_accuracy": 0.7122968435287476, + "num_tokens": 453896804.0, + "step": 17848 + }, + { + "epoch": 1.9601361739512408, + "grad_norm": 2.3022618293762207, + "learning_rate": 1e-06, + "loss": 0.8539, + "mean_token_accuracy": 0.7291589975357056, + "num_tokens": 453920590.0, + "step": 17849 + }, + { + "epoch": 1.9602459916538546, + "grad_norm": 2.6143133640289307, + "learning_rate": 1e-06, + "loss": 0.8818, + "mean_token_accuracy": 0.7355743646621704, + "num_tokens": 453939999.0, + "step": 17850 + }, + { + "epoch": 1.9603558093564684, + "grad_norm": 2.307178020477295, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.725765585899353, + "num_tokens": 453964295.0, + "step": 17851 + }, + { + "epoch": 1.960465627059082, + "grad_norm": 2.056633710861206, + "learning_rate": 1e-06, + "loss": 1.0378, + "mean_token_accuracy": 0.6866474151611328, + "num_tokens": 453994820.0, + "step": 17852 + }, + { + "epoch": 1.9605754447616954, + "grad_norm": 2.418074369430542, + "learning_rate": 1e-06, + "loss": 0.9141, + "mean_token_accuracy": 0.7113103866577148, + "num_tokens": 454016774.0, + "step": 17853 + }, + { + "epoch": 1.9606852624643092, + "grad_norm": 2.320387601852417, + "learning_rate": 1e-06, + "loss": 0.9197, + "mean_token_accuracy": 0.7121918201446533, + "num_tokens": 454040462.0, + "step": 17854 + }, + { + "epoch": 1.960795080166923, + "grad_norm": 2.2836854457855225, + "learning_rate": 1e-06, + "loss": 0.9179, + "mean_token_accuracy": 0.716629147529602, + "num_tokens": 454065669.0, + "step": 17855 + }, + { + "epoch": 1.9609048978695367, + "grad_norm": 2.237172842025757, + "learning_rate": 1e-06, + "loss": 0.9145, + "mean_token_accuracy": 0.7171021699905396, + "num_tokens": 454090930.0, + "step": 17856 + }, + { + "epoch": 1.9610147155721502, + "grad_norm": 2.008831262588501, + "learning_rate": 1e-06, + "loss": 0.9727, + "mean_token_accuracy": 0.7032854557037354, + "num_tokens": 454122409.0, + "step": 17857 + }, + { + "epoch": 1.9611245332747638, + "grad_norm": 2.203052282333374, + "learning_rate": 1e-06, + "loss": 0.8317, + "mean_token_accuracy": 0.740287184715271, + "num_tokens": 454148738.0, + "step": 17858 + }, + { + "epoch": 1.9612343509773775, + "grad_norm": 2.3535890579223633, + "learning_rate": 1e-06, + "loss": 0.9455, + "mean_token_accuracy": 0.7102665901184082, + "num_tokens": 454174686.0, + "step": 17859 + }, + { + "epoch": 1.9613441686799913, + "grad_norm": 2.5772640705108643, + "learning_rate": 1e-06, + "loss": 0.918, + "mean_token_accuracy": 0.7208476066589355, + "num_tokens": 454197364.0, + "step": 17860 + }, + { + "epoch": 1.961453986382605, + "grad_norm": 2.221060276031494, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7202767729759216, + "num_tokens": 454223343.0, + "step": 17861 + }, + { + "epoch": 1.9615638040852186, + "grad_norm": 2.093841552734375, + "learning_rate": 1e-06, + "loss": 0.9323, + "mean_token_accuracy": 0.7126145958900452, + "num_tokens": 454251660.0, + "step": 17862 + }, + { + "epoch": 1.9616736217878321, + "grad_norm": 2.2587687969207764, + "learning_rate": 1e-06, + "loss": 0.9116, + "mean_token_accuracy": 0.7179499864578247, + "num_tokens": 454277858.0, + "step": 17863 + }, + { + "epoch": 1.9617834394904459, + "grad_norm": 2.1288323402404785, + "learning_rate": 1e-06, + "loss": 0.9438, + "mean_token_accuracy": 0.7042572498321533, + "num_tokens": 454307012.0, + "step": 17864 + }, + { + "epoch": 1.9618932571930596, + "grad_norm": 2.4210832118988037, + "learning_rate": 1e-06, + "loss": 0.9733, + "mean_token_accuracy": 0.7057504057884216, + "num_tokens": 454330858.0, + "step": 17865 + }, + { + "epoch": 1.9620030748956732, + "grad_norm": 2.2882182598114014, + "learning_rate": 1e-06, + "loss": 0.9385, + "mean_token_accuracy": 0.7035298347473145, + "num_tokens": 454357316.0, + "step": 17866 + }, + { + "epoch": 1.9621128925982867, + "grad_norm": 2.1044466495513916, + "learning_rate": 1e-06, + "loss": 0.9221, + "mean_token_accuracy": 0.7206196784973145, + "num_tokens": 454383336.0, + "step": 17867 + }, + { + "epoch": 1.9622227103009005, + "grad_norm": 2.1549365520477295, + "learning_rate": 1e-06, + "loss": 0.9571, + "mean_token_accuracy": 0.7018932104110718, + "num_tokens": 454410345.0, + "step": 17868 + }, + { + "epoch": 1.9623325280035142, + "grad_norm": 2.5418896675109863, + "learning_rate": 1e-06, + "loss": 0.7861, + "mean_token_accuracy": 0.7449599504470825, + "num_tokens": 454430302.0, + "step": 17869 + }, + { + "epoch": 1.962442345706128, + "grad_norm": 1.8911298513412476, + "learning_rate": 1e-06, + "loss": 0.8963, + "mean_token_accuracy": 0.7170716524124146, + "num_tokens": 454463307.0, + "step": 17870 + }, + { + "epoch": 1.9625521634087415, + "grad_norm": 2.3328959941864014, + "learning_rate": 1e-06, + "loss": 0.9135, + "mean_token_accuracy": 0.7243566513061523, + "num_tokens": 454486195.0, + "step": 17871 + }, + { + "epoch": 1.962661981111355, + "grad_norm": 2.1456761360168457, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7295030355453491, + "num_tokens": 454513029.0, + "step": 17872 + }, + { + "epoch": 1.9627717988139688, + "grad_norm": 2.1382198333740234, + "learning_rate": 1e-06, + "loss": 0.8892, + "mean_token_accuracy": 0.7274567484855652, + "num_tokens": 454542198.0, + "step": 17873 + }, + { + "epoch": 1.9628816165165826, + "grad_norm": 2.2314491271972656, + "learning_rate": 1e-06, + "loss": 0.9514, + "mean_token_accuracy": 0.7071568965911865, + "num_tokens": 454568699.0, + "step": 17874 + }, + { + "epoch": 1.962991434219196, + "grad_norm": 2.323261260986328, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7155609726905823, + "num_tokens": 454594479.0, + "step": 17875 + }, + { + "epoch": 1.9631012519218098, + "grad_norm": 2.938652276992798, + "learning_rate": 1e-06, + "loss": 0.869, + "mean_token_accuracy": 0.7266579866409302, + "num_tokens": 454612141.0, + "step": 17876 + }, + { + "epoch": 1.9632110696244234, + "grad_norm": 2.588581085205078, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.7505918741226196, + "num_tokens": 454632855.0, + "step": 17877 + }, + { + "epoch": 1.9633208873270371, + "grad_norm": 2.492818832397461, + "learning_rate": 1e-06, + "loss": 0.973, + "mean_token_accuracy": 0.7057954668998718, + "num_tokens": 454656973.0, + "step": 17878 + }, + { + "epoch": 1.963430705029651, + "grad_norm": 2.6844284534454346, + "learning_rate": 1e-06, + "loss": 0.8455, + "mean_token_accuracy": 0.7300940155982971, + "num_tokens": 454675404.0, + "step": 17879 + }, + { + "epoch": 1.9635405227322644, + "grad_norm": 2.5562641620635986, + "learning_rate": 1e-06, + "loss": 0.8689, + "mean_token_accuracy": 0.7189953327178955, + "num_tokens": 454696872.0, + "step": 17880 + }, + { + "epoch": 1.963650340434878, + "grad_norm": 2.362584352493286, + "learning_rate": 1e-06, + "loss": 0.8431, + "mean_token_accuracy": 0.7362306714057922, + "num_tokens": 454720968.0, + "step": 17881 + }, + { + "epoch": 1.9637601581374917, + "grad_norm": 2.1594550609588623, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7213895320892334, + "num_tokens": 454749588.0, + "step": 17882 + }, + { + "epoch": 1.9638699758401055, + "grad_norm": 2.092224359512329, + "learning_rate": 1e-06, + "loss": 0.8801, + "mean_token_accuracy": 0.7263128757476807, + "num_tokens": 454777311.0, + "step": 17883 + }, + { + "epoch": 1.9639797935427192, + "grad_norm": 2.1726629734039307, + "learning_rate": 1e-06, + "loss": 0.8349, + "mean_token_accuracy": 0.7319959402084351, + "num_tokens": 454802224.0, + "step": 17884 + }, + { + "epoch": 1.9640896112453328, + "grad_norm": 2.156538248062134, + "learning_rate": 1e-06, + "loss": 1.0114, + "mean_token_accuracy": 0.6925694942474365, + "num_tokens": 454830537.0, + "step": 17885 + }, + { + "epoch": 1.9641994289479463, + "grad_norm": 2.5171854496002197, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7249836325645447, + "num_tokens": 454851477.0, + "step": 17886 + }, + { + "epoch": 1.96430924665056, + "grad_norm": 2.5634493827819824, + "learning_rate": 1e-06, + "loss": 0.9535, + "mean_token_accuracy": 0.7078633308410645, + "num_tokens": 454872405.0, + "step": 17887 + }, + { + "epoch": 1.9644190643531738, + "grad_norm": 2.5150837898254395, + "learning_rate": 1e-06, + "loss": 0.7905, + "mean_token_accuracy": 0.7509108781814575, + "num_tokens": 454892510.0, + "step": 17888 + }, + { + "epoch": 1.9645288820557874, + "grad_norm": 2.2310094833374023, + "learning_rate": 1e-06, + "loss": 0.9223, + "mean_token_accuracy": 0.714485764503479, + "num_tokens": 454917178.0, + "step": 17889 + }, + { + "epoch": 1.964638699758401, + "grad_norm": 2.482830047607422, + "learning_rate": 1e-06, + "loss": 0.8923, + "mean_token_accuracy": 0.7274891138076782, + "num_tokens": 454937883.0, + "step": 17890 + }, + { + "epoch": 1.9647485174610146, + "grad_norm": 2.5745902061462402, + "learning_rate": 1e-06, + "loss": 0.8591, + "mean_token_accuracy": 0.7239010334014893, + "num_tokens": 454959276.0, + "step": 17891 + }, + { + "epoch": 1.9648583351636284, + "grad_norm": 2.381247043609619, + "learning_rate": 1e-06, + "loss": 0.8708, + "mean_token_accuracy": 0.7229169607162476, + "num_tokens": 454982492.0, + "step": 17892 + }, + { + "epoch": 1.9649681528662422, + "grad_norm": 2.3531205654144287, + "learning_rate": 1e-06, + "loss": 0.93, + "mean_token_accuracy": 0.708751916885376, + "num_tokens": 455006622.0, + "step": 17893 + }, + { + "epoch": 1.9650779705688557, + "grad_norm": 2.232372999191284, + "learning_rate": 1e-06, + "loss": 0.9127, + "mean_token_accuracy": 0.712138295173645, + "num_tokens": 455030436.0, + "step": 17894 + }, + { + "epoch": 1.9651877882714692, + "grad_norm": 2.202847480773926, + "learning_rate": 1e-06, + "loss": 0.9173, + "mean_token_accuracy": 0.7167459726333618, + "num_tokens": 455055448.0, + "step": 17895 + }, + { + "epoch": 1.965297605974083, + "grad_norm": 2.4607532024383545, + "learning_rate": 1e-06, + "loss": 0.8909, + "mean_token_accuracy": 0.7254649996757507, + "num_tokens": 455078145.0, + "step": 17896 + }, + { + "epoch": 1.9654074236766967, + "grad_norm": 2.3568170070648193, + "learning_rate": 1e-06, + "loss": 0.8237, + "mean_token_accuracy": 0.7370063662528992, + "num_tokens": 455102586.0, + "step": 17897 + }, + { + "epoch": 1.9655172413793105, + "grad_norm": 2.199812650680542, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.7155038714408875, + "num_tokens": 455128377.0, + "step": 17898 + }, + { + "epoch": 1.965627059081924, + "grad_norm": 2.432041645050049, + "learning_rate": 1e-06, + "loss": 0.7907, + "mean_token_accuracy": 0.7465708255767822, + "num_tokens": 455149930.0, + "step": 17899 + }, + { + "epoch": 1.9657368767845376, + "grad_norm": 2.469701051712036, + "learning_rate": 1e-06, + "loss": 0.9404, + "mean_token_accuracy": 0.7063142657279968, + "num_tokens": 455172092.0, + "step": 17900 + }, + { + "epoch": 1.9658466944871513, + "grad_norm": 2.2014999389648438, + "learning_rate": 1e-06, + "loss": 0.9151, + "mean_token_accuracy": 0.7251815795898438, + "num_tokens": 455199204.0, + "step": 17901 + }, + { + "epoch": 1.965956512189765, + "grad_norm": 2.2141289710998535, + "learning_rate": 1e-06, + "loss": 0.8501, + "mean_token_accuracy": 0.735184907913208, + "num_tokens": 455224772.0, + "step": 17902 + }, + { + "epoch": 1.9660663298923786, + "grad_norm": 2.2540605068206787, + "learning_rate": 1e-06, + "loss": 0.8783, + "mean_token_accuracy": 0.7231685519218445, + "num_tokens": 455249847.0, + "step": 17903 + }, + { + "epoch": 1.9661761475949922, + "grad_norm": 2.5721681118011475, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7137366533279419, + "num_tokens": 455271248.0, + "step": 17904 + }, + { + "epoch": 1.966285965297606, + "grad_norm": 1.873905897140503, + "learning_rate": 1e-06, + "loss": 1.0087, + "mean_token_accuracy": 0.6899241209030151, + "num_tokens": 455308511.0, + "step": 17905 + }, + { + "epoch": 1.9663957830002197, + "grad_norm": 2.22178316116333, + "learning_rate": 1e-06, + "loss": 0.7669, + "mean_token_accuracy": 0.7575721740722656, + "num_tokens": 455332697.0, + "step": 17906 + }, + { + "epoch": 1.9665056007028334, + "grad_norm": 2.309044122695923, + "learning_rate": 1e-06, + "loss": 0.8289, + "mean_token_accuracy": 0.7410631775856018, + "num_tokens": 455355731.0, + "step": 17907 + }, + { + "epoch": 1.966615418405447, + "grad_norm": 2.2966480255126953, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7204170823097229, + "num_tokens": 455380446.0, + "step": 17908 + }, + { + "epoch": 1.9667252361080605, + "grad_norm": 2.0716311931610107, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7151240706443787, + "num_tokens": 455408252.0, + "step": 17909 + }, + { + "epoch": 1.9668350538106742, + "grad_norm": 1.943143606185913, + "learning_rate": 1e-06, + "loss": 1.0256, + "mean_token_accuracy": 0.6892772912979126, + "num_tokens": 455443506.0, + "step": 17910 + }, + { + "epoch": 1.966944871513288, + "grad_norm": 2.2035610675811768, + "learning_rate": 1e-06, + "loss": 0.8524, + "mean_token_accuracy": 0.7415499687194824, + "num_tokens": 455469118.0, + "step": 17911 + }, + { + "epoch": 1.9670546892159018, + "grad_norm": 2.2324554920196533, + "learning_rate": 1e-06, + "loss": 0.9193, + "mean_token_accuracy": 0.7152626514434814, + "num_tokens": 455495287.0, + "step": 17912 + }, + { + "epoch": 1.9671645069185153, + "grad_norm": 2.0380473136901855, + "learning_rate": 1e-06, + "loss": 0.7248, + "mean_token_accuracy": 0.7645852565765381, + "num_tokens": 455519868.0, + "step": 17913 + }, + { + "epoch": 1.9672743246211288, + "grad_norm": 2.2387313842773438, + "learning_rate": 1e-06, + "loss": 0.885, + "mean_token_accuracy": 0.735277533531189, + "num_tokens": 455545084.0, + "step": 17914 + }, + { + "epoch": 1.9673841423237426, + "grad_norm": 2.2686853408813477, + "learning_rate": 1e-06, + "loss": 0.9546, + "mean_token_accuracy": 0.7058365345001221, + "num_tokens": 455569662.0, + "step": 17915 + }, + { + "epoch": 1.9674939600263563, + "grad_norm": 2.2065718173980713, + "learning_rate": 1e-06, + "loss": 0.8966, + "mean_token_accuracy": 0.7216238379478455, + "num_tokens": 455595589.0, + "step": 17916 + }, + { + "epoch": 1.9676037777289699, + "grad_norm": 2.3192930221557617, + "learning_rate": 1e-06, + "loss": 0.8582, + "mean_token_accuracy": 0.7258045077323914, + "num_tokens": 455619933.0, + "step": 17917 + }, + { + "epoch": 1.9677135954315834, + "grad_norm": 2.164058208465576, + "learning_rate": 1e-06, + "loss": 0.9142, + "mean_token_accuracy": 0.708572506904602, + "num_tokens": 455647951.0, + "step": 17918 + }, + { + "epoch": 1.9678234131341972, + "grad_norm": 2.262850522994995, + "learning_rate": 1e-06, + "loss": 0.8454, + "mean_token_accuracy": 0.7364177703857422, + "num_tokens": 455670827.0, + "step": 17919 + }, + { + "epoch": 1.967933230836811, + "grad_norm": 2.4872241020202637, + "learning_rate": 1e-06, + "loss": 0.8058, + "mean_token_accuracy": 0.7434784173965454, + "num_tokens": 455692150.0, + "step": 17920 + }, + { + "epoch": 1.9680430485394247, + "grad_norm": 2.0964272022247314, + "learning_rate": 1e-06, + "loss": 0.7393, + "mean_token_accuracy": 0.7623661756515503, + "num_tokens": 455717452.0, + "step": 17921 + }, + { + "epoch": 1.9681528662420382, + "grad_norm": 2.480487585067749, + "learning_rate": 1e-06, + "loss": 0.8379, + "mean_token_accuracy": 0.7338200807571411, + "num_tokens": 455737955.0, + "step": 17922 + }, + { + "epoch": 1.9682626839446518, + "grad_norm": 1.9935789108276367, + "learning_rate": 1e-06, + "loss": 0.917, + "mean_token_accuracy": 0.7170246839523315, + "num_tokens": 455768192.0, + "step": 17923 + }, + { + "epoch": 1.9683725016472655, + "grad_norm": 2.2010676860809326, + "learning_rate": 1e-06, + "loss": 0.9042, + "mean_token_accuracy": 0.7261466979980469, + "num_tokens": 455796269.0, + "step": 17924 + }, + { + "epoch": 1.9684823193498793, + "grad_norm": 2.2862701416015625, + "learning_rate": 1e-06, + "loss": 0.8711, + "mean_token_accuracy": 0.7348761558532715, + "num_tokens": 455818246.0, + "step": 17925 + }, + { + "epoch": 1.9685921370524928, + "grad_norm": 2.299473762512207, + "learning_rate": 1e-06, + "loss": 0.8011, + "mean_token_accuracy": 0.7450401782989502, + "num_tokens": 455841560.0, + "step": 17926 + }, + { + "epoch": 1.9687019547551066, + "grad_norm": 2.1910877227783203, + "learning_rate": 1e-06, + "loss": 1.0336, + "mean_token_accuracy": 0.6873719692230225, + "num_tokens": 455870506.0, + "step": 17927 + }, + { + "epoch": 1.96881177245772, + "grad_norm": 2.269272804260254, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7260262370109558, + "num_tokens": 455895461.0, + "step": 17928 + }, + { + "epoch": 1.9689215901603339, + "grad_norm": 1.957869052886963, + "learning_rate": 1e-06, + "loss": 0.975, + "mean_token_accuracy": 0.7042719721794128, + "num_tokens": 455927516.0, + "step": 17929 + }, + { + "epoch": 1.9690314078629476, + "grad_norm": 2.5782289505004883, + "learning_rate": 1e-06, + "loss": 0.7752, + "mean_token_accuracy": 0.7546170949935913, + "num_tokens": 455947002.0, + "step": 17930 + }, + { + "epoch": 1.9691412255655611, + "grad_norm": 2.2673771381378174, + "learning_rate": 1e-06, + "loss": 0.8954, + "mean_token_accuracy": 0.7210782170295715, + "num_tokens": 455972144.0, + "step": 17931 + }, + { + "epoch": 1.9692510432681747, + "grad_norm": 2.284299612045288, + "learning_rate": 1e-06, + "loss": 0.948, + "mean_token_accuracy": 0.7044590711593628, + "num_tokens": 455998180.0, + "step": 17932 + }, + { + "epoch": 1.9693608609707884, + "grad_norm": 2.780954360961914, + "learning_rate": 1e-06, + "loss": 0.8145, + "mean_token_accuracy": 0.743754506111145, + "num_tokens": 456018359.0, + "step": 17933 + }, + { + "epoch": 1.9694706786734022, + "grad_norm": 2.042165994644165, + "learning_rate": 1e-06, + "loss": 0.9366, + "mean_token_accuracy": 0.7048999667167664, + "num_tokens": 456048143.0, + "step": 17934 + }, + { + "epoch": 1.969580496376016, + "grad_norm": 2.25614070892334, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7092391848564148, + "num_tokens": 456073351.0, + "step": 17935 + }, + { + "epoch": 1.9696903140786295, + "grad_norm": 2.0742666721343994, + "learning_rate": 1e-06, + "loss": 0.9092, + "mean_token_accuracy": 0.7199166417121887, + "num_tokens": 456102148.0, + "step": 17936 + }, + { + "epoch": 1.969800131781243, + "grad_norm": 2.0824811458587646, + "learning_rate": 1e-06, + "loss": 0.893, + "mean_token_accuracy": 0.7215655446052551, + "num_tokens": 456130455.0, + "step": 17937 + }, + { + "epoch": 1.9699099494838568, + "grad_norm": 2.336117744445801, + "learning_rate": 1e-06, + "loss": 0.8705, + "mean_token_accuracy": 0.7222650647163391, + "num_tokens": 456154080.0, + "step": 17938 + }, + { + "epoch": 1.9700197671864705, + "grad_norm": 2.6996142864227295, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7359822988510132, + "num_tokens": 456172569.0, + "step": 17939 + }, + { + "epoch": 1.970129584889084, + "grad_norm": 2.3528823852539062, + "learning_rate": 1e-06, + "loss": 0.8934, + "mean_token_accuracy": 0.7196335792541504, + "num_tokens": 456195677.0, + "step": 17940 + }, + { + "epoch": 1.9702394025916978, + "grad_norm": 2.129729986190796, + "learning_rate": 1e-06, + "loss": 0.8746, + "mean_token_accuracy": 0.7287067174911499, + "num_tokens": 456222707.0, + "step": 17941 + }, + { + "epoch": 1.9703492202943114, + "grad_norm": 2.3351621627807617, + "learning_rate": 1e-06, + "loss": 0.799, + "mean_token_accuracy": 0.7491135597229004, + "num_tokens": 456244267.0, + "step": 17942 + }, + { + "epoch": 1.9704590379969251, + "grad_norm": 2.3293535709381104, + "learning_rate": 1e-06, + "loss": 0.8065, + "mean_token_accuracy": 0.7432928085327148, + "num_tokens": 456266959.0, + "step": 17943 + }, + { + "epoch": 1.9705688556995389, + "grad_norm": 2.350299835205078, + "learning_rate": 1e-06, + "loss": 0.884, + "mean_token_accuracy": 0.7244507670402527, + "num_tokens": 456291558.0, + "step": 17944 + }, + { + "epoch": 1.9706786734021524, + "grad_norm": 2.0927300453186035, + "learning_rate": 1e-06, + "loss": 0.9013, + "mean_token_accuracy": 0.7203767895698547, + "num_tokens": 456321300.0, + "step": 17945 + }, + { + "epoch": 1.970788491104766, + "grad_norm": 2.452263116836548, + "learning_rate": 1e-06, + "loss": 0.9375, + "mean_token_accuracy": 0.7172207832336426, + "num_tokens": 456344150.0, + "step": 17946 + }, + { + "epoch": 1.9708983088073797, + "grad_norm": 2.1237828731536865, + "learning_rate": 1e-06, + "loss": 0.9725, + "mean_token_accuracy": 0.707883358001709, + "num_tokens": 456372668.0, + "step": 17947 + }, + { + "epoch": 1.9710081265099935, + "grad_norm": 2.2410061359405518, + "learning_rate": 1e-06, + "loss": 0.8953, + "mean_token_accuracy": 0.7199034690856934, + "num_tokens": 456398401.0, + "step": 17948 + }, + { + "epoch": 1.9711179442126072, + "grad_norm": 1.9615442752838135, + "learning_rate": 1e-06, + "loss": 0.9602, + "mean_token_accuracy": 0.7029650211334229, + "num_tokens": 456429783.0, + "step": 17949 + }, + { + "epoch": 1.9712277619152208, + "grad_norm": 2.1523005962371826, + "learning_rate": 1e-06, + "loss": 0.9301, + "mean_token_accuracy": 0.7097322940826416, + "num_tokens": 456456464.0, + "step": 17950 + }, + { + "epoch": 1.9713375796178343, + "grad_norm": 2.4218590259552, + "learning_rate": 1e-06, + "loss": 0.8516, + "mean_token_accuracy": 0.7355633974075317, + "num_tokens": 456477514.0, + "step": 17951 + }, + { + "epoch": 1.971447397320448, + "grad_norm": 2.115750551223755, + "learning_rate": 1e-06, + "loss": 0.9435, + "mean_token_accuracy": 0.7137761116027832, + "num_tokens": 456505847.0, + "step": 17952 + }, + { + "epoch": 1.9715572150230618, + "grad_norm": 2.1472036838531494, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7137700319290161, + "num_tokens": 456532666.0, + "step": 17953 + }, + { + "epoch": 1.9716670327256753, + "grad_norm": 2.0019102096557617, + "learning_rate": 1e-06, + "loss": 0.8984, + "mean_token_accuracy": 0.7221875190734863, + "num_tokens": 456562405.0, + "step": 17954 + }, + { + "epoch": 1.9717768504282889, + "grad_norm": 2.2171127796173096, + "learning_rate": 1e-06, + "loss": 0.9282, + "mean_token_accuracy": 0.7235081791877747, + "num_tokens": 456588103.0, + "step": 17955 + }, + { + "epoch": 1.9718866681309026, + "grad_norm": 2.1252388954162598, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7153774499893188, + "num_tokens": 456617494.0, + "step": 17956 + }, + { + "epoch": 1.9719964858335164, + "grad_norm": 2.1447901725769043, + "learning_rate": 1e-06, + "loss": 0.9626, + "mean_token_accuracy": 0.7104243636131287, + "num_tokens": 456642950.0, + "step": 17957 + }, + { + "epoch": 1.9721063035361301, + "grad_norm": 2.1804895401000977, + "learning_rate": 1e-06, + "loss": 0.9747, + "mean_token_accuracy": 0.7058671712875366, + "num_tokens": 456669820.0, + "step": 17958 + }, + { + "epoch": 1.9722161212387437, + "grad_norm": 2.244004011154175, + "learning_rate": 1e-06, + "loss": 0.908, + "mean_token_accuracy": 0.7136390209197998, + "num_tokens": 456695654.0, + "step": 17959 + }, + { + "epoch": 1.9723259389413572, + "grad_norm": 2.2424283027648926, + "learning_rate": 1e-06, + "loss": 1.0133, + "mean_token_accuracy": 0.6910921335220337, + "num_tokens": 456724246.0, + "step": 17960 + }, + { + "epoch": 1.972435756643971, + "grad_norm": 2.196984052658081, + "learning_rate": 1e-06, + "loss": 0.8661, + "mean_token_accuracy": 0.7248541712760925, + "num_tokens": 456751226.0, + "step": 17961 + }, + { + "epoch": 1.9725455743465847, + "grad_norm": 2.053774833679199, + "learning_rate": 1e-06, + "loss": 0.98, + "mean_token_accuracy": 0.6991483569145203, + "num_tokens": 456783566.0, + "step": 17962 + }, + { + "epoch": 1.9726553920491985, + "grad_norm": 2.136337995529175, + "learning_rate": 1e-06, + "loss": 0.8592, + "mean_token_accuracy": 0.7275183200836182, + "num_tokens": 456810342.0, + "step": 17963 + }, + { + "epoch": 1.972765209751812, + "grad_norm": 2.189103603363037, + "learning_rate": 1e-06, + "loss": 0.886, + "mean_token_accuracy": 0.7136284112930298, + "num_tokens": 456834722.0, + "step": 17964 + }, + { + "epoch": 1.9728750274544256, + "grad_norm": 2.235236644744873, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7116561532020569, + "num_tokens": 456860935.0, + "step": 17965 + }, + { + "epoch": 1.9729848451570393, + "grad_norm": 2.0700936317443848, + "learning_rate": 1e-06, + "loss": 0.947, + "mean_token_accuracy": 0.7070173621177673, + "num_tokens": 456889101.0, + "step": 17966 + }, + { + "epoch": 1.973094662859653, + "grad_norm": 2.474731683731079, + "learning_rate": 1e-06, + "loss": 0.9494, + "mean_token_accuracy": 0.7139683961868286, + "num_tokens": 456911000.0, + "step": 17967 + }, + { + "epoch": 1.9732044805622666, + "grad_norm": 1.9515094757080078, + "learning_rate": 1e-06, + "loss": 0.9041, + "mean_token_accuracy": 0.7215560078620911, + "num_tokens": 456942253.0, + "step": 17968 + }, + { + "epoch": 1.9733142982648801, + "grad_norm": 2.329608201980591, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7291700839996338, + "num_tokens": 456966392.0, + "step": 17969 + }, + { + "epoch": 1.973424115967494, + "grad_norm": 2.5097572803497314, + "learning_rate": 1e-06, + "loss": 0.9201, + "mean_token_accuracy": 0.7188416719436646, + "num_tokens": 456987547.0, + "step": 17970 + }, + { + "epoch": 1.9735339336701077, + "grad_norm": 2.1876275539398193, + "learning_rate": 1e-06, + "loss": 0.8361, + "mean_token_accuracy": 0.7394390106201172, + "num_tokens": 457014639.0, + "step": 17971 + }, + { + "epoch": 1.9736437513727214, + "grad_norm": 2.321669816970825, + "learning_rate": 1e-06, + "loss": 0.9159, + "mean_token_accuracy": 0.723143458366394, + "num_tokens": 457038872.0, + "step": 17972 + }, + { + "epoch": 1.973753569075335, + "grad_norm": 2.509676218032837, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.718254566192627, + "num_tokens": 457059067.0, + "step": 17973 + }, + { + "epoch": 1.9738633867779485, + "grad_norm": 1.9739577770233154, + "learning_rate": 1e-06, + "loss": 0.8344, + "mean_token_accuracy": 0.7397969961166382, + "num_tokens": 457087780.0, + "step": 17974 + }, + { + "epoch": 1.9739732044805622, + "grad_norm": 2.3432018756866455, + "learning_rate": 1e-06, + "loss": 0.8047, + "mean_token_accuracy": 0.746309757232666, + "num_tokens": 457110871.0, + "step": 17975 + }, + { + "epoch": 1.974083022183176, + "grad_norm": 2.218905448913574, + "learning_rate": 1e-06, + "loss": 0.9125, + "mean_token_accuracy": 0.7195302248001099, + "num_tokens": 457136041.0, + "step": 17976 + }, + { + "epoch": 1.9741928398857898, + "grad_norm": 2.484739065170288, + "learning_rate": 1e-06, + "loss": 0.939, + "mean_token_accuracy": 0.7103948593139648, + "num_tokens": 457158618.0, + "step": 17977 + }, + { + "epoch": 1.9743026575884033, + "grad_norm": 2.4500927925109863, + "learning_rate": 1e-06, + "loss": 0.824, + "mean_token_accuracy": 0.7399866580963135, + "num_tokens": 457180848.0, + "step": 17978 + }, + { + "epoch": 1.9744124752910168, + "grad_norm": 2.505829334259033, + "learning_rate": 1e-06, + "loss": 0.7796, + "mean_token_accuracy": 0.7493419051170349, + "num_tokens": 457200734.0, + "step": 17979 + }, + { + "epoch": 1.9745222929936306, + "grad_norm": 2.1973772048950195, + "learning_rate": 1e-06, + "loss": 0.9163, + "mean_token_accuracy": 0.7129265069961548, + "num_tokens": 457227343.0, + "step": 17980 + }, + { + "epoch": 1.9746321106962443, + "grad_norm": 2.240791082382202, + "learning_rate": 1e-06, + "loss": 0.9144, + "mean_token_accuracy": 0.7106101512908936, + "num_tokens": 457253223.0, + "step": 17981 + }, + { + "epoch": 1.9747419283988579, + "grad_norm": 2.446786403656006, + "learning_rate": 1e-06, + "loss": 0.8382, + "mean_token_accuracy": 0.7372768521308899, + "num_tokens": 457275292.0, + "step": 17982 + }, + { + "epoch": 1.9748517461014714, + "grad_norm": 2.381072998046875, + "learning_rate": 1e-06, + "loss": 0.8147, + "mean_token_accuracy": 0.7425171136856079, + "num_tokens": 457297134.0, + "step": 17983 + }, + { + "epoch": 1.9749615638040852, + "grad_norm": 2.231097936630249, + "learning_rate": 1e-06, + "loss": 0.9864, + "mean_token_accuracy": 0.693681538105011, + "num_tokens": 457323378.0, + "step": 17984 + }, + { + "epoch": 1.975071381506699, + "grad_norm": 2.1488759517669678, + "learning_rate": 1e-06, + "loss": 0.8575, + "mean_token_accuracy": 0.7265273332595825, + "num_tokens": 457349786.0, + "step": 17985 + }, + { + "epoch": 1.9751811992093127, + "grad_norm": 2.268374443054199, + "learning_rate": 1e-06, + "loss": 0.9176, + "mean_token_accuracy": 0.7135951519012451, + "num_tokens": 457374772.0, + "step": 17986 + }, + { + "epoch": 1.9752910169119262, + "grad_norm": 2.06942081451416, + "learning_rate": 1e-06, + "loss": 0.9606, + "mean_token_accuracy": 0.6984493732452393, + "num_tokens": 457403376.0, + "step": 17987 + }, + { + "epoch": 1.9754008346145397, + "grad_norm": 2.123523712158203, + "learning_rate": 1e-06, + "loss": 0.8632, + "mean_token_accuracy": 0.7246121168136597, + "num_tokens": 457428962.0, + "step": 17988 + }, + { + "epoch": 1.9755106523171535, + "grad_norm": 2.2702250480651855, + "learning_rate": 1e-06, + "loss": 0.797, + "mean_token_accuracy": 0.7451657056808472, + "num_tokens": 457451722.0, + "step": 17989 + }, + { + "epoch": 1.9756204700197673, + "grad_norm": 2.579969644546509, + "learning_rate": 1e-06, + "loss": 0.8332, + "mean_token_accuracy": 0.7318177819252014, + "num_tokens": 457471488.0, + "step": 17990 + }, + { + "epoch": 1.9757302877223808, + "grad_norm": 2.511737823486328, + "learning_rate": 1e-06, + "loss": 0.8695, + "mean_token_accuracy": 0.7198705673217773, + "num_tokens": 457493016.0, + "step": 17991 + }, + { + "epoch": 1.9758401054249946, + "grad_norm": 2.440162420272827, + "learning_rate": 1e-06, + "loss": 0.9011, + "mean_token_accuracy": 0.7276201844215393, + "num_tokens": 457514957.0, + "step": 17992 + }, + { + "epoch": 1.975949923127608, + "grad_norm": 2.5496835708618164, + "learning_rate": 1e-06, + "loss": 0.8487, + "mean_token_accuracy": 0.7347004413604736, + "num_tokens": 457534846.0, + "step": 17993 + }, + { + "epoch": 1.9760597408302218, + "grad_norm": 2.277029037475586, + "learning_rate": 1e-06, + "loss": 0.9417, + "mean_token_accuracy": 0.7054571509361267, + "num_tokens": 457561288.0, + "step": 17994 + }, + { + "epoch": 1.9761695585328356, + "grad_norm": 2.4096357822418213, + "learning_rate": 1e-06, + "loss": 0.9099, + "mean_token_accuracy": 0.7123696804046631, + "num_tokens": 457584216.0, + "step": 17995 + }, + { + "epoch": 1.9762793762354491, + "grad_norm": 2.073383331298828, + "learning_rate": 1e-06, + "loss": 0.8462, + "mean_token_accuracy": 0.7349880933761597, + "num_tokens": 457612242.0, + "step": 17996 + }, + { + "epoch": 1.9763891939380627, + "grad_norm": 2.1636292934417725, + "learning_rate": 1e-06, + "loss": 0.9121, + "mean_token_accuracy": 0.7198727130889893, + "num_tokens": 457638112.0, + "step": 17997 + }, + { + "epoch": 1.9764990116406764, + "grad_norm": 2.0861399173736572, + "learning_rate": 1e-06, + "loss": 0.9346, + "mean_token_accuracy": 0.7046501040458679, + "num_tokens": 457667546.0, + "step": 17998 + }, + { + "epoch": 1.9766088293432902, + "grad_norm": 2.1393284797668457, + "learning_rate": 1e-06, + "loss": 0.835, + "mean_token_accuracy": 0.7325485944747925, + "num_tokens": 457693797.0, + "step": 17999 + }, + { + "epoch": 1.976718647045904, + "grad_norm": 2.4715161323547363, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.721035361289978, + "num_tokens": 457714773.0, + "step": 18000 + }, + { + "epoch": 1.9768284647485175, + "grad_norm": 2.411618232727051, + "learning_rate": 1e-06, + "loss": 0.9509, + "mean_token_accuracy": 0.7061104774475098, + "num_tokens": 457738893.0, + "step": 18001 + }, + { + "epoch": 1.976938282451131, + "grad_norm": 2.299452543258667, + "learning_rate": 1e-06, + "loss": 0.8759, + "mean_token_accuracy": 0.7290232181549072, + "num_tokens": 457762479.0, + "step": 18002 + }, + { + "epoch": 1.9770481001537448, + "grad_norm": 2.302281618118286, + "learning_rate": 1e-06, + "loss": 0.9552, + "mean_token_accuracy": 0.7112836837768555, + "num_tokens": 457785143.0, + "step": 18003 + }, + { + "epoch": 1.9771579178563585, + "grad_norm": 2.2150886058807373, + "learning_rate": 1e-06, + "loss": 0.9808, + "mean_token_accuracy": 0.7035609483718872, + "num_tokens": 457811928.0, + "step": 18004 + }, + { + "epoch": 1.977267735558972, + "grad_norm": 2.286790609359741, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7334655523300171, + "num_tokens": 457838479.0, + "step": 18005 + }, + { + "epoch": 1.9773775532615858, + "grad_norm": 2.6293182373046875, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.731421947479248, + "num_tokens": 457860945.0, + "step": 18006 + }, + { + "epoch": 1.9774873709641994, + "grad_norm": 2.241084337234497, + "learning_rate": 1e-06, + "loss": 0.901, + "mean_token_accuracy": 0.7178921699523926, + "num_tokens": 457887823.0, + "step": 18007 + }, + { + "epoch": 1.977597188666813, + "grad_norm": 2.0316617488861084, + "learning_rate": 1e-06, + "loss": 1.0607, + "mean_token_accuracy": 0.6857684850692749, + "num_tokens": 457918177.0, + "step": 18008 + }, + { + "epoch": 1.9777070063694269, + "grad_norm": 2.1435494422912598, + "learning_rate": 1e-06, + "loss": 1.026, + "mean_token_accuracy": 0.6885420083999634, + "num_tokens": 457948075.0, + "step": 18009 + }, + { + "epoch": 1.9778168240720404, + "grad_norm": 2.3841915130615234, + "learning_rate": 1e-06, + "loss": 0.898, + "mean_token_accuracy": 0.7218904495239258, + "num_tokens": 457971281.0, + "step": 18010 + }, + { + "epoch": 1.977926641774654, + "grad_norm": 2.2347397804260254, + "learning_rate": 1e-06, + "loss": 0.9392, + "mean_token_accuracy": 0.7149468660354614, + "num_tokens": 457996920.0, + "step": 18011 + }, + { + "epoch": 1.9780364594772677, + "grad_norm": 2.394502878189087, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7209901213645935, + "num_tokens": 458019887.0, + "step": 18012 + }, + { + "epoch": 1.9781462771798815, + "grad_norm": 2.249459743499756, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7309691905975342, + "num_tokens": 458042958.0, + "step": 18013 + }, + { + "epoch": 1.9782560948824952, + "grad_norm": 2.3733322620391846, + "learning_rate": 1e-06, + "loss": 0.9267, + "mean_token_accuracy": 0.712009608745575, + "num_tokens": 458064504.0, + "step": 18014 + }, + { + "epoch": 1.9783659125851087, + "grad_norm": 2.286081075668335, + "learning_rate": 1e-06, + "loss": 0.928, + "mean_token_accuracy": 0.7115492820739746, + "num_tokens": 458090538.0, + "step": 18015 + }, + { + "epoch": 1.9784757302877223, + "grad_norm": 2.4978392124176025, + "learning_rate": 1e-06, + "loss": 0.8974, + "mean_token_accuracy": 0.7148321866989136, + "num_tokens": 458111270.0, + "step": 18016 + }, + { + "epoch": 1.978585547990336, + "grad_norm": 2.28043532371521, + "learning_rate": 1e-06, + "loss": 0.95, + "mean_token_accuracy": 0.7075539827346802, + "num_tokens": 458137782.0, + "step": 18017 + }, + { + "epoch": 1.9786953656929498, + "grad_norm": 2.0190136432647705, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7263845205307007, + "num_tokens": 458165751.0, + "step": 18018 + }, + { + "epoch": 1.9788051833955633, + "grad_norm": 2.1461899280548096, + "learning_rate": 1e-06, + "loss": 0.9447, + "mean_token_accuracy": 0.7077571153640747, + "num_tokens": 458193793.0, + "step": 18019 + }, + { + "epoch": 1.9789150010981769, + "grad_norm": 2.2991998195648193, + "learning_rate": 1e-06, + "loss": 0.8291, + "mean_token_accuracy": 0.7460139989852905, + "num_tokens": 458217416.0, + "step": 18020 + }, + { + "epoch": 1.9790248188007906, + "grad_norm": 2.053623676300049, + "learning_rate": 1e-06, + "loss": 0.8185, + "mean_token_accuracy": 0.7360201478004456, + "num_tokens": 458244666.0, + "step": 18021 + }, + { + "epoch": 1.9791346365034044, + "grad_norm": 2.1692917346954346, + "learning_rate": 1e-06, + "loss": 0.9181, + "mean_token_accuracy": 0.7127773761749268, + "num_tokens": 458271871.0, + "step": 18022 + }, + { + "epoch": 1.9792444542060181, + "grad_norm": 2.213791608810425, + "learning_rate": 1e-06, + "loss": 0.8987, + "mean_token_accuracy": 0.721227765083313, + "num_tokens": 458295867.0, + "step": 18023 + }, + { + "epoch": 1.9793542719086317, + "grad_norm": 2.5726280212402344, + "learning_rate": 1e-06, + "loss": 0.9378, + "mean_token_accuracy": 0.7197952270507812, + "num_tokens": 458316538.0, + "step": 18024 + }, + { + "epoch": 1.9794640896112452, + "grad_norm": 2.200434684753418, + "learning_rate": 1e-06, + "loss": 0.8715, + "mean_token_accuracy": 0.721674919128418, + "num_tokens": 458342130.0, + "step": 18025 + }, + { + "epoch": 1.979573907313859, + "grad_norm": 2.2932896614074707, + "learning_rate": 1e-06, + "loss": 0.9878, + "mean_token_accuracy": 0.696907639503479, + "num_tokens": 458367870.0, + "step": 18026 + }, + { + "epoch": 1.9796837250164727, + "grad_norm": 3.828028440475464, + "learning_rate": 1e-06, + "loss": 0.8521, + "mean_token_accuracy": 0.7313703894615173, + "num_tokens": 458390532.0, + "step": 18027 + }, + { + "epoch": 1.9797935427190865, + "grad_norm": 2.124788761138916, + "learning_rate": 1e-06, + "loss": 0.9089, + "mean_token_accuracy": 0.7206894159317017, + "num_tokens": 458418630.0, + "step": 18028 + }, + { + "epoch": 1.9799033604217, + "grad_norm": 2.3991971015930176, + "learning_rate": 1e-06, + "loss": 0.7689, + "mean_token_accuracy": 0.7588408589363098, + "num_tokens": 458440615.0, + "step": 18029 + }, + { + "epoch": 1.9800131781243135, + "grad_norm": 2.1767520904541016, + "learning_rate": 1e-06, + "loss": 0.7686, + "mean_token_accuracy": 0.7513517141342163, + "num_tokens": 458466627.0, + "step": 18030 + }, + { + "epoch": 1.9801229958269273, + "grad_norm": 2.611325740814209, + "learning_rate": 1e-06, + "loss": 0.8504, + "mean_token_accuracy": 0.7374729514122009, + "num_tokens": 458487998.0, + "step": 18031 + }, + { + "epoch": 1.980232813529541, + "grad_norm": 2.234567880630493, + "learning_rate": 1e-06, + "loss": 0.9061, + "mean_token_accuracy": 0.7220579981803894, + "num_tokens": 458513300.0, + "step": 18032 + }, + { + "epoch": 1.9803426312321546, + "grad_norm": 2.3348779678344727, + "learning_rate": 1e-06, + "loss": 0.8744, + "mean_token_accuracy": 0.7266733050346375, + "num_tokens": 458536221.0, + "step": 18033 + }, + { + "epoch": 1.9804524489347681, + "grad_norm": 2.342883348464966, + "learning_rate": 1e-06, + "loss": 0.8537, + "mean_token_accuracy": 0.7371945381164551, + "num_tokens": 458559321.0, + "step": 18034 + }, + { + "epoch": 1.9805622666373819, + "grad_norm": 2.5166072845458984, + "learning_rate": 1e-06, + "loss": 0.957, + "mean_token_accuracy": 0.7043323516845703, + "num_tokens": 458580792.0, + "step": 18035 + }, + { + "epoch": 1.9806720843399956, + "grad_norm": 2.38071870803833, + "learning_rate": 1e-06, + "loss": 0.9399, + "mean_token_accuracy": 0.7046456336975098, + "num_tokens": 458607547.0, + "step": 18036 + }, + { + "epoch": 1.9807819020426094, + "grad_norm": 2.3866748809814453, + "learning_rate": 1e-06, + "loss": 0.9202, + "mean_token_accuracy": 0.715203583240509, + "num_tokens": 458631358.0, + "step": 18037 + }, + { + "epoch": 1.980891719745223, + "grad_norm": 2.2581374645233154, + "learning_rate": 1e-06, + "loss": 0.9786, + "mean_token_accuracy": 0.694420576095581, + "num_tokens": 458659387.0, + "step": 18038 + }, + { + "epoch": 1.9810015374478365, + "grad_norm": 2.3019015789031982, + "learning_rate": 1e-06, + "loss": 0.8919, + "mean_token_accuracy": 0.7201570272445679, + "num_tokens": 458684076.0, + "step": 18039 + }, + { + "epoch": 1.9811113551504502, + "grad_norm": 2.2813830375671387, + "learning_rate": 1e-06, + "loss": 0.8922, + "mean_token_accuracy": 0.7211126089096069, + "num_tokens": 458707747.0, + "step": 18040 + }, + { + "epoch": 1.981221172853064, + "grad_norm": 2.1294875144958496, + "learning_rate": 1e-06, + "loss": 0.8975, + "mean_token_accuracy": 0.7150560617446899, + "num_tokens": 458736471.0, + "step": 18041 + }, + { + "epoch": 1.9813309905556777, + "grad_norm": 2.363210678100586, + "learning_rate": 1e-06, + "loss": 0.7959, + "mean_token_accuracy": 0.7454201579093933, + "num_tokens": 458758649.0, + "step": 18042 + }, + { + "epoch": 1.9814408082582913, + "grad_norm": 2.2887158393859863, + "learning_rate": 1e-06, + "loss": 0.818, + "mean_token_accuracy": 0.7386845350265503, + "num_tokens": 458783408.0, + "step": 18043 + }, + { + "epoch": 1.9815506259609048, + "grad_norm": 2.569169759750366, + "learning_rate": 1e-06, + "loss": 0.9261, + "mean_token_accuracy": 0.7137270569801331, + "num_tokens": 458804480.0, + "step": 18044 + }, + { + "epoch": 1.9816604436635186, + "grad_norm": 2.2981107234954834, + "learning_rate": 1e-06, + "loss": 0.9018, + "mean_token_accuracy": 0.7142879366874695, + "num_tokens": 458829868.0, + "step": 18045 + }, + { + "epoch": 1.9817702613661323, + "grad_norm": 2.131920337677002, + "learning_rate": 1e-06, + "loss": 0.9149, + "mean_token_accuracy": 0.7131441831588745, + "num_tokens": 458856825.0, + "step": 18046 + }, + { + "epoch": 1.9818800790687459, + "grad_norm": 2.1249423027038574, + "learning_rate": 1e-06, + "loss": 0.8599, + "mean_token_accuracy": 0.7292210459709167, + "num_tokens": 458884105.0, + "step": 18047 + }, + { + "epoch": 1.9819898967713594, + "grad_norm": 2.071777820587158, + "learning_rate": 1e-06, + "loss": 0.8825, + "mean_token_accuracy": 0.7302037477493286, + "num_tokens": 458910974.0, + "step": 18048 + }, + { + "epoch": 1.9820997144739732, + "grad_norm": 2.0506184101104736, + "learning_rate": 1e-06, + "loss": 1.003, + "mean_token_accuracy": 0.695703387260437, + "num_tokens": 458941867.0, + "step": 18049 + }, + { + "epoch": 1.982209532176587, + "grad_norm": 2.3991897106170654, + "learning_rate": 1e-06, + "loss": 0.8255, + "mean_token_accuracy": 0.737252414226532, + "num_tokens": 458964765.0, + "step": 18050 + }, + { + "epoch": 1.9823193498792007, + "grad_norm": 1.9978852272033691, + "learning_rate": 1e-06, + "loss": 0.9415, + "mean_token_accuracy": 0.7108440399169922, + "num_tokens": 458995709.0, + "step": 18051 + }, + { + "epoch": 1.9824291675818142, + "grad_norm": 2.427565336227417, + "learning_rate": 1e-06, + "loss": 0.9169, + "mean_token_accuracy": 0.7150275111198425, + "num_tokens": 459018578.0, + "step": 18052 + }, + { + "epoch": 1.9825389852844277, + "grad_norm": 2.3810088634490967, + "learning_rate": 1e-06, + "loss": 0.8571, + "mean_token_accuracy": 0.7300064563751221, + "num_tokens": 459041019.0, + "step": 18053 + }, + { + "epoch": 1.9826488029870415, + "grad_norm": 2.243137836456299, + "learning_rate": 1e-06, + "loss": 0.8389, + "mean_token_accuracy": 0.7351019382476807, + "num_tokens": 459064963.0, + "step": 18054 + }, + { + "epoch": 1.9827586206896552, + "grad_norm": 2.08811092376709, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7082583904266357, + "num_tokens": 459094059.0, + "step": 18055 + }, + { + "epoch": 1.9828684383922688, + "grad_norm": 2.1658935546875, + "learning_rate": 1e-06, + "loss": 0.914, + "mean_token_accuracy": 0.7218809127807617, + "num_tokens": 459120405.0, + "step": 18056 + }, + { + "epoch": 1.9829782560948825, + "grad_norm": 2.323470115661621, + "learning_rate": 1e-06, + "loss": 0.9706, + "mean_token_accuracy": 0.7055416703224182, + "num_tokens": 459145153.0, + "step": 18057 + }, + { + "epoch": 1.983088073797496, + "grad_norm": 2.1963677406311035, + "learning_rate": 1e-06, + "loss": 0.8449, + "mean_token_accuracy": 0.7362508773803711, + "num_tokens": 459170847.0, + "step": 18058 + }, + { + "epoch": 1.9831978915001098, + "grad_norm": 2.239638090133667, + "learning_rate": 1e-06, + "loss": 0.9701, + "mean_token_accuracy": 0.7058930397033691, + "num_tokens": 459198705.0, + "step": 18059 + }, + { + "epoch": 1.9833077092027236, + "grad_norm": 2.438356399536133, + "learning_rate": 1e-06, + "loss": 0.8935, + "mean_token_accuracy": 0.7162613272666931, + "num_tokens": 459220765.0, + "step": 18060 + }, + { + "epoch": 1.9834175269053371, + "grad_norm": 1.942789077758789, + "learning_rate": 1e-06, + "loss": 0.9138, + "mean_token_accuracy": 0.7147284746170044, + "num_tokens": 459254020.0, + "step": 18061 + }, + { + "epoch": 1.9835273446079507, + "grad_norm": 2.0130395889282227, + "learning_rate": 1e-06, + "loss": 0.9183, + "mean_token_accuracy": 0.7213991284370422, + "num_tokens": 459285937.0, + "step": 18062 + }, + { + "epoch": 1.9836371623105644, + "grad_norm": 2.5325732231140137, + "learning_rate": 1e-06, + "loss": 0.9465, + "mean_token_accuracy": 0.7083094120025635, + "num_tokens": 459308523.0, + "step": 18063 + }, + { + "epoch": 1.9837469800131782, + "grad_norm": 2.180008888244629, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7426964044570923, + "num_tokens": 459333941.0, + "step": 18064 + }, + { + "epoch": 1.983856797715792, + "grad_norm": 2.114894390106201, + "learning_rate": 1e-06, + "loss": 0.889, + "mean_token_accuracy": 0.7207364439964294, + "num_tokens": 459360546.0, + "step": 18065 + }, + { + "epoch": 1.9839666154184055, + "grad_norm": 2.1979739665985107, + "learning_rate": 1e-06, + "loss": 0.8554, + "mean_token_accuracy": 0.7261908054351807, + "num_tokens": 459384819.0, + "step": 18066 + }, + { + "epoch": 1.984076433121019, + "grad_norm": 2.2447853088378906, + "learning_rate": 1e-06, + "loss": 0.8771, + "mean_token_accuracy": 0.7233859300613403, + "num_tokens": 459409221.0, + "step": 18067 + }, + { + "epoch": 1.9841862508236328, + "grad_norm": 2.1528360843658447, + "learning_rate": 1e-06, + "loss": 0.7593, + "mean_token_accuracy": 0.7647289037704468, + "num_tokens": 459432313.0, + "step": 18068 + }, + { + "epoch": 1.9842960685262465, + "grad_norm": 2.666738271713257, + "learning_rate": 1e-06, + "loss": 0.8831, + "mean_token_accuracy": 0.7248176336288452, + "num_tokens": 459453371.0, + "step": 18069 + }, + { + "epoch": 1.98440588622886, + "grad_norm": 2.211073875427246, + "learning_rate": 1e-06, + "loss": 0.8341, + "mean_token_accuracy": 0.7383408546447754, + "num_tokens": 459477778.0, + "step": 18070 + }, + { + "epoch": 1.9845157039314738, + "grad_norm": 2.3397176265716553, + "learning_rate": 1e-06, + "loss": 1.0106, + "mean_token_accuracy": 0.7061613202095032, + "num_tokens": 459503353.0, + "step": 18071 + }, + { + "epoch": 1.9846255216340873, + "grad_norm": 1.8602230548858643, + "learning_rate": 1e-06, + "loss": 0.963, + "mean_token_accuracy": 0.7028392553329468, + "num_tokens": 459538538.0, + "step": 18072 + }, + { + "epoch": 1.984735339336701, + "grad_norm": 2.475914239883423, + "learning_rate": 1e-06, + "loss": 0.8903, + "mean_token_accuracy": 0.7215853333473206, + "num_tokens": 459561770.0, + "step": 18073 + }, + { + "epoch": 1.9848451570393149, + "grad_norm": 2.136047840118408, + "learning_rate": 1e-06, + "loss": 0.9033, + "mean_token_accuracy": 0.7224714756011963, + "num_tokens": 459589102.0, + "step": 18074 + }, + { + "epoch": 1.9849549747419284, + "grad_norm": 2.569489002227783, + "learning_rate": 1e-06, + "loss": 0.8432, + "mean_token_accuracy": 0.7329147458076477, + "num_tokens": 459608735.0, + "step": 18075 + }, + { + "epoch": 1.985064792444542, + "grad_norm": 2.8564774990081787, + "learning_rate": 1e-06, + "loss": 0.7691, + "mean_token_accuracy": 0.7611939907073975, + "num_tokens": 459625040.0, + "step": 18076 + }, + { + "epoch": 1.9851746101471557, + "grad_norm": 2.25860333442688, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7315748333930969, + "num_tokens": 459649968.0, + "step": 18077 + }, + { + "epoch": 1.9852844278497694, + "grad_norm": 2.4162681102752686, + "learning_rate": 1e-06, + "loss": 0.8894, + "mean_token_accuracy": 0.7204896807670593, + "num_tokens": 459674503.0, + "step": 18078 + }, + { + "epoch": 1.9853942455523832, + "grad_norm": 2.2076821327209473, + "learning_rate": 1e-06, + "loss": 0.9428, + "mean_token_accuracy": 0.7188631892204285, + "num_tokens": 459701897.0, + "step": 18079 + }, + { + "epoch": 1.9855040632549967, + "grad_norm": 2.296837329864502, + "learning_rate": 1e-06, + "loss": 0.8113, + "mean_token_accuracy": 0.7409792542457581, + "num_tokens": 459725574.0, + "step": 18080 + }, + { + "epoch": 1.9856138809576103, + "grad_norm": 2.314701795578003, + "learning_rate": 1e-06, + "loss": 0.9031, + "mean_token_accuracy": 0.7159518599510193, + "num_tokens": 459753093.0, + "step": 18081 + }, + { + "epoch": 1.985723698660224, + "grad_norm": 2.513894557952881, + "learning_rate": 1e-06, + "loss": 0.9611, + "mean_token_accuracy": 0.7100720405578613, + "num_tokens": 459776374.0, + "step": 18082 + }, + { + "epoch": 1.9858335163628378, + "grad_norm": 2.106137275695801, + "learning_rate": 1e-06, + "loss": 0.9091, + "mean_token_accuracy": 0.7118786573410034, + "num_tokens": 459806635.0, + "step": 18083 + }, + { + "epoch": 1.9859433340654513, + "grad_norm": 2.2340176105499268, + "learning_rate": 1e-06, + "loss": 0.8363, + "mean_token_accuracy": 0.7376236319541931, + "num_tokens": 459830528.0, + "step": 18084 + }, + { + "epoch": 1.9860531517680649, + "grad_norm": 2.083345413208008, + "learning_rate": 1e-06, + "loss": 0.9372, + "mean_token_accuracy": 0.7075834274291992, + "num_tokens": 459859741.0, + "step": 18085 + }, + { + "epoch": 1.9861629694706786, + "grad_norm": 2.2486345767974854, + "learning_rate": 1e-06, + "loss": 0.8557, + "mean_token_accuracy": 0.7362476587295532, + "num_tokens": 459883838.0, + "step": 18086 + }, + { + "epoch": 1.9862727871732924, + "grad_norm": 2.3858933448791504, + "learning_rate": 1e-06, + "loss": 0.8022, + "mean_token_accuracy": 0.7469971179962158, + "num_tokens": 459907321.0, + "step": 18087 + }, + { + "epoch": 1.9863826048759061, + "grad_norm": 2.1556906700134277, + "learning_rate": 1e-06, + "loss": 1.001, + "mean_token_accuracy": 0.6878670454025269, + "num_tokens": 459934060.0, + "step": 18088 + }, + { + "epoch": 1.9864924225785197, + "grad_norm": 2.3695738315582275, + "learning_rate": 1e-06, + "loss": 0.8649, + "mean_token_accuracy": 0.7268520593643188, + "num_tokens": 459956689.0, + "step": 18089 + }, + { + "epoch": 1.9866022402811332, + "grad_norm": 2.46589732170105, + "learning_rate": 1e-06, + "loss": 0.8871, + "mean_token_accuracy": 0.722470760345459, + "num_tokens": 459978432.0, + "step": 18090 + }, + { + "epoch": 1.986712057983747, + "grad_norm": 1.9208513498306274, + "learning_rate": 1e-06, + "loss": 0.9904, + "mean_token_accuracy": 0.6929641962051392, + "num_tokens": 460010607.0, + "step": 18091 + }, + { + "epoch": 1.9868218756863607, + "grad_norm": 2.2498714923858643, + "learning_rate": 1e-06, + "loss": 0.8413, + "mean_token_accuracy": 0.7401641607284546, + "num_tokens": 460035043.0, + "step": 18092 + }, + { + "epoch": 1.9869316933889745, + "grad_norm": 2.210808277130127, + "learning_rate": 1e-06, + "loss": 0.9479, + "mean_token_accuracy": 0.7180745005607605, + "num_tokens": 460061463.0, + "step": 18093 + }, + { + "epoch": 1.987041511091588, + "grad_norm": 2.0001161098480225, + "learning_rate": 1e-06, + "loss": 0.9412, + "mean_token_accuracy": 0.7133620977401733, + "num_tokens": 460092842.0, + "step": 18094 + }, + { + "epoch": 1.9871513287942015, + "grad_norm": 2.28619122505188, + "learning_rate": 1e-06, + "loss": 0.9081, + "mean_token_accuracy": 0.7183863520622253, + "num_tokens": 460117846.0, + "step": 18095 + }, + { + "epoch": 1.9872611464968153, + "grad_norm": 2.346327543258667, + "learning_rate": 1e-06, + "loss": 0.9483, + "mean_token_accuracy": 0.7071304321289062, + "num_tokens": 460140821.0, + "step": 18096 + }, + { + "epoch": 1.987370964199429, + "grad_norm": 2.4258058071136475, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7258347272872925, + "num_tokens": 460164059.0, + "step": 18097 + }, + { + "epoch": 1.9874807819020426, + "grad_norm": 2.3337345123291016, + "learning_rate": 1e-06, + "loss": 0.9532, + "mean_token_accuracy": 0.7059568166732788, + "num_tokens": 460187291.0, + "step": 18098 + }, + { + "epoch": 1.9875905996046561, + "grad_norm": 2.150872230529785, + "learning_rate": 1e-06, + "loss": 0.8699, + "mean_token_accuracy": 0.7200330495834351, + "num_tokens": 460212723.0, + "step": 18099 + }, + { + "epoch": 1.9877004173072699, + "grad_norm": 2.1273624897003174, + "learning_rate": 1e-06, + "loss": 0.9555, + "mean_token_accuracy": 0.713784396648407, + "num_tokens": 460243121.0, + "step": 18100 + }, + { + "epoch": 1.9878102350098836, + "grad_norm": 2.1210386753082275, + "learning_rate": 1e-06, + "loss": 0.8996, + "mean_token_accuracy": 0.719449520111084, + "num_tokens": 460271827.0, + "step": 18101 + }, + { + "epoch": 1.9879200527124974, + "grad_norm": 2.3491857051849365, + "learning_rate": 1e-06, + "loss": 0.7148, + "mean_token_accuracy": 0.7621512413024902, + "num_tokens": 460293550.0, + "step": 18102 + }, + { + "epoch": 1.988029870415111, + "grad_norm": 2.100114107131958, + "learning_rate": 1e-06, + "loss": 0.9408, + "mean_token_accuracy": 0.7179503440856934, + "num_tokens": 460323908.0, + "step": 18103 + }, + { + "epoch": 1.9881396881177245, + "grad_norm": 2.1413512229919434, + "learning_rate": 1e-06, + "loss": 0.8643, + "mean_token_accuracy": 0.7275428175926208, + "num_tokens": 460351115.0, + "step": 18104 + }, + { + "epoch": 1.9882495058203382, + "grad_norm": 2.2315785884857178, + "learning_rate": 1e-06, + "loss": 0.7776, + "mean_token_accuracy": 0.75839763879776, + "num_tokens": 460375754.0, + "step": 18105 + }, + { + "epoch": 1.988359323522952, + "grad_norm": 2.32385516166687, + "learning_rate": 1e-06, + "loss": 0.8522, + "mean_token_accuracy": 0.7337290048599243, + "num_tokens": 460399333.0, + "step": 18106 + }, + { + "epoch": 1.9884691412255655, + "grad_norm": 2.5704257488250732, + "learning_rate": 1e-06, + "loss": 0.9409, + "mean_token_accuracy": 0.7085739970207214, + "num_tokens": 460420711.0, + "step": 18107 + }, + { + "epoch": 1.9885789589281793, + "grad_norm": 2.1812596321105957, + "learning_rate": 1e-06, + "loss": 0.8512, + "mean_token_accuracy": 0.7337933778762817, + "num_tokens": 460445630.0, + "step": 18108 + }, + { + "epoch": 1.9886887766307928, + "grad_norm": 2.503995418548584, + "learning_rate": 1e-06, + "loss": 0.8312, + "mean_token_accuracy": 0.7377763986587524, + "num_tokens": 460464614.0, + "step": 18109 + }, + { + "epoch": 1.9887985943334066, + "grad_norm": 2.250535011291504, + "learning_rate": 1e-06, + "loss": 0.9349, + "mean_token_accuracy": 0.7237454652786255, + "num_tokens": 460495584.0, + "step": 18110 + }, + { + "epoch": 1.9889084120360203, + "grad_norm": 2.426077365875244, + "learning_rate": 1e-06, + "loss": 0.7547, + "mean_token_accuracy": 0.7524438500404358, + "num_tokens": 460515373.0, + "step": 18111 + }, + { + "epoch": 1.9890182297386338, + "grad_norm": 2.138063669204712, + "learning_rate": 1e-06, + "loss": 0.9728, + "mean_token_accuracy": 0.7069892883300781, + "num_tokens": 460541595.0, + "step": 18112 + }, + { + "epoch": 1.9891280474412474, + "grad_norm": 1.9221280813217163, + "learning_rate": 1e-06, + "loss": 0.8739, + "mean_token_accuracy": 0.7238527536392212, + "num_tokens": 460574293.0, + "step": 18113 + }, + { + "epoch": 1.9892378651438611, + "grad_norm": 2.248103380203247, + "learning_rate": 1e-06, + "loss": 0.919, + "mean_token_accuracy": 0.7107664346694946, + "num_tokens": 460597867.0, + "step": 18114 + }, + { + "epoch": 1.989347682846475, + "grad_norm": 2.5325469970703125, + "learning_rate": 1e-06, + "loss": 0.8947, + "mean_token_accuracy": 0.7211663722991943, + "num_tokens": 460618191.0, + "step": 18115 + }, + { + "epoch": 1.9894575005490887, + "grad_norm": 2.4771463871002197, + "learning_rate": 1e-06, + "loss": 0.8655, + "mean_token_accuracy": 0.7266949415206909, + "num_tokens": 460640343.0, + "step": 18116 + }, + { + "epoch": 1.9895673182517022, + "grad_norm": 1.8849564790725708, + "learning_rate": 1e-06, + "loss": 0.9421, + "mean_token_accuracy": 0.7026643753051758, + "num_tokens": 460675199.0, + "step": 18117 + }, + { + "epoch": 1.9896771359543157, + "grad_norm": 2.7100889682769775, + "learning_rate": 1e-06, + "loss": 0.8395, + "mean_token_accuracy": 0.731763482093811, + "num_tokens": 460693260.0, + "step": 18118 + }, + { + "epoch": 1.9897869536569295, + "grad_norm": 2.225019931793213, + "learning_rate": 1e-06, + "loss": 0.8337, + "mean_token_accuracy": 0.7350361347198486, + "num_tokens": 460718461.0, + "step": 18119 + }, + { + "epoch": 1.9898967713595432, + "grad_norm": 2.018690824508667, + "learning_rate": 1e-06, + "loss": 0.994, + "mean_token_accuracy": 0.7011421918869019, + "num_tokens": 460748227.0, + "step": 18120 + }, + { + "epoch": 1.9900065890621568, + "grad_norm": 2.4361512660980225, + "learning_rate": 1e-06, + "loss": 0.8651, + "mean_token_accuracy": 0.7273523211479187, + "num_tokens": 460767875.0, + "step": 18121 + }, + { + "epoch": 1.9901164067647705, + "grad_norm": 2.1639621257781982, + "learning_rate": 1e-06, + "loss": 0.9134, + "mean_token_accuracy": 0.7166217565536499, + "num_tokens": 460794722.0, + "step": 18122 + }, + { + "epoch": 1.990226224467384, + "grad_norm": 2.2062594890594482, + "learning_rate": 1e-06, + "loss": 0.8615, + "mean_token_accuracy": 0.7321516275405884, + "num_tokens": 460820569.0, + "step": 18123 + }, + { + "epoch": 1.9903360421699978, + "grad_norm": 1.9542293548583984, + "learning_rate": 1e-06, + "loss": 0.9, + "mean_token_accuracy": 0.7232729196548462, + "num_tokens": 460851472.0, + "step": 18124 + }, + { + "epoch": 1.9904458598726116, + "grad_norm": 2.392808198928833, + "learning_rate": 1e-06, + "loss": 0.8365, + "mean_token_accuracy": 0.737363874912262, + "num_tokens": 460873474.0, + "step": 18125 + }, + { + "epoch": 1.9905556775752251, + "grad_norm": 2.2627739906311035, + "learning_rate": 1e-06, + "loss": 0.8152, + "mean_token_accuracy": 0.7432202100753784, + "num_tokens": 460896855.0, + "step": 18126 + }, + { + "epoch": 1.9906654952778386, + "grad_norm": 2.2176272869110107, + "learning_rate": 1e-06, + "loss": 0.9502, + "mean_token_accuracy": 0.7144336104393005, + "num_tokens": 460922970.0, + "step": 18127 + }, + { + "epoch": 1.9907753129804524, + "grad_norm": 2.3431718349456787, + "learning_rate": 1e-06, + "loss": 0.9068, + "mean_token_accuracy": 0.7160228490829468, + "num_tokens": 460946805.0, + "step": 18128 + }, + { + "epoch": 1.9908851306830662, + "grad_norm": 2.0977110862731934, + "learning_rate": 1e-06, + "loss": 0.9175, + "mean_token_accuracy": 0.7148458957672119, + "num_tokens": 460976901.0, + "step": 18129 + }, + { + "epoch": 1.99099494838568, + "grad_norm": 2.329617738723755, + "learning_rate": 1e-06, + "loss": 0.9051, + "mean_token_accuracy": 0.7134032249450684, + "num_tokens": 461001664.0, + "step": 18130 + }, + { + "epoch": 1.9911047660882935, + "grad_norm": 2.3847875595092773, + "learning_rate": 1e-06, + "loss": 0.8838, + "mean_token_accuracy": 0.723512589931488, + "num_tokens": 461024314.0, + "step": 18131 + }, + { + "epoch": 1.991214583790907, + "grad_norm": 2.2565815448760986, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7091364860534668, + "num_tokens": 461049287.0, + "step": 18132 + }, + { + "epoch": 1.9913244014935207, + "grad_norm": 2.5139808654785156, + "learning_rate": 1e-06, + "loss": 0.7435, + "mean_token_accuracy": 0.7592086791992188, + "num_tokens": 461068159.0, + "step": 18133 + }, + { + "epoch": 1.9914342191961345, + "grad_norm": 2.1869378089904785, + "learning_rate": 1e-06, + "loss": 0.8069, + "mean_token_accuracy": 0.7399156093597412, + "num_tokens": 461093191.0, + "step": 18134 + }, + { + "epoch": 1.991544036898748, + "grad_norm": 2.255462884902954, + "learning_rate": 1e-06, + "loss": 0.9527, + "mean_token_accuracy": 0.7112617492675781, + "num_tokens": 461118179.0, + "step": 18135 + }, + { + "epoch": 1.9916538546013618, + "grad_norm": 2.5637760162353516, + "learning_rate": 1e-06, + "loss": 0.7786, + "mean_token_accuracy": 0.7510476112365723, + "num_tokens": 461139798.0, + "step": 18136 + }, + { + "epoch": 1.9917636723039753, + "grad_norm": 2.3140745162963867, + "learning_rate": 1e-06, + "loss": 0.9957, + "mean_token_accuracy": 0.6902751922607422, + "num_tokens": 461165070.0, + "step": 18137 + }, + { + "epoch": 1.991873490006589, + "grad_norm": 2.3063740730285645, + "learning_rate": 1e-06, + "loss": 0.836, + "mean_token_accuracy": 0.7394013404846191, + "num_tokens": 461187779.0, + "step": 18138 + }, + { + "epoch": 1.9919833077092028, + "grad_norm": 2.1086208820343018, + "learning_rate": 1e-06, + "loss": 0.9286, + "mean_token_accuracy": 0.7102259397506714, + "num_tokens": 461216670.0, + "step": 18139 + }, + { + "epoch": 1.9920931254118164, + "grad_norm": 2.1135873794555664, + "learning_rate": 1e-06, + "loss": 0.9427, + "mean_token_accuracy": 0.7071530222892761, + "num_tokens": 461246661.0, + "step": 18140 + }, + { + "epoch": 1.99220294311443, + "grad_norm": 1.9321180582046509, + "learning_rate": 1e-06, + "loss": 1.015, + "mean_token_accuracy": 0.6980733871459961, + "num_tokens": 461279790.0, + "step": 18141 + }, + { + "epoch": 1.9923127608170437, + "grad_norm": 2.06655216217041, + "learning_rate": 1e-06, + "loss": 0.9974, + "mean_token_accuracy": 0.6940777897834778, + "num_tokens": 461308882.0, + "step": 18142 + }, + { + "epoch": 1.9924225785196574, + "grad_norm": 2.320376396179199, + "learning_rate": 1e-06, + "loss": 0.8415, + "mean_token_accuracy": 0.7373912334442139, + "num_tokens": 461331852.0, + "step": 18143 + }, + { + "epoch": 1.9925323962222712, + "grad_norm": 2.1883625984191895, + "learning_rate": 1e-06, + "loss": 0.882, + "mean_token_accuracy": 0.733086884021759, + "num_tokens": 461356939.0, + "step": 18144 + }, + { + "epoch": 1.9926422139248847, + "grad_norm": 2.1864259243011475, + "learning_rate": 1e-06, + "loss": 0.8613, + "mean_token_accuracy": 0.729080080986023, + "num_tokens": 461382695.0, + "step": 18145 + }, + { + "epoch": 1.9927520316274983, + "grad_norm": 2.0416579246520996, + "learning_rate": 1e-06, + "loss": 0.9897, + "mean_token_accuracy": 0.702195405960083, + "num_tokens": 461414738.0, + "step": 18146 + }, + { + "epoch": 1.992861849330112, + "grad_norm": 2.5305120944976807, + "learning_rate": 1e-06, + "loss": 0.902, + "mean_token_accuracy": 0.7191334962844849, + "num_tokens": 461435964.0, + "step": 18147 + }, + { + "epoch": 1.9929716670327258, + "grad_norm": 2.2861990928649902, + "learning_rate": 1e-06, + "loss": 0.9276, + "mean_token_accuracy": 0.7066770792007446, + "num_tokens": 461460619.0, + "step": 18148 + }, + { + "epoch": 1.9930814847353393, + "grad_norm": 2.1065640449523926, + "learning_rate": 1e-06, + "loss": 0.9377, + "mean_token_accuracy": 0.7050936818122864, + "num_tokens": 461490308.0, + "step": 18149 + }, + { + "epoch": 1.9931913024379528, + "grad_norm": 2.0344719886779785, + "learning_rate": 1e-06, + "loss": 0.9722, + "mean_token_accuracy": 0.7090696096420288, + "num_tokens": 461520288.0, + "step": 18150 + }, + { + "epoch": 1.9933011201405666, + "grad_norm": 2.1426007747650146, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.7378938794136047, + "num_tokens": 461545085.0, + "step": 18151 + }, + { + "epoch": 1.9934109378431804, + "grad_norm": 1.9794286489486694, + "learning_rate": 1e-06, + "loss": 0.8477, + "mean_token_accuracy": 0.7278403043746948, + "num_tokens": 461572936.0, + "step": 18152 + }, + { + "epoch": 1.993520755545794, + "grad_norm": 1.9351282119750977, + "learning_rate": 1e-06, + "loss": 0.9155, + "mean_token_accuracy": 0.710093080997467, + "num_tokens": 461604650.0, + "step": 18153 + }, + { + "epoch": 1.9936305732484076, + "grad_norm": 2.0433156490325928, + "learning_rate": 1e-06, + "loss": 1.0068, + "mean_token_accuracy": 0.6965460777282715, + "num_tokens": 461634880.0, + "step": 18154 + }, + { + "epoch": 1.9937403909510212, + "grad_norm": 2.401975631713867, + "learning_rate": 1e-06, + "loss": 0.9007, + "mean_token_accuracy": 0.7199317216873169, + "num_tokens": 461658556.0, + "step": 18155 + }, + { + "epoch": 1.993850208653635, + "grad_norm": 2.275022506713867, + "learning_rate": 1e-06, + "loss": 0.7729, + "mean_token_accuracy": 0.7507558465003967, + "num_tokens": 461681253.0, + "step": 18156 + }, + { + "epoch": 1.9939600263562487, + "grad_norm": 2.1983611583709717, + "learning_rate": 1e-06, + "loss": 0.9362, + "mean_token_accuracy": 0.7126595973968506, + "num_tokens": 461710128.0, + "step": 18157 + }, + { + "epoch": 1.9940698440588625, + "grad_norm": 2.531007766723633, + "learning_rate": 1e-06, + "loss": 0.8476, + "mean_token_accuracy": 0.7312017679214478, + "num_tokens": 461730935.0, + "step": 18158 + }, + { + "epoch": 1.994179661761476, + "grad_norm": 2.2243118286132812, + "learning_rate": 1e-06, + "loss": 0.8422, + "mean_token_accuracy": 0.7365831136703491, + "num_tokens": 461755197.0, + "step": 18159 + }, + { + "epoch": 1.9942894794640895, + "grad_norm": 2.139822483062744, + "learning_rate": 1e-06, + "loss": 0.9054, + "mean_token_accuracy": 0.7224118709564209, + "num_tokens": 461781637.0, + "step": 18160 + }, + { + "epoch": 1.9943992971667033, + "grad_norm": 2.4899139404296875, + "learning_rate": 1e-06, + "loss": 1.0009, + "mean_token_accuracy": 0.7051384449005127, + "num_tokens": 461804658.0, + "step": 18161 + }, + { + "epoch": 1.994509114869317, + "grad_norm": 2.4386754035949707, + "learning_rate": 1e-06, + "loss": 0.8427, + "mean_token_accuracy": 0.7319391965866089, + "num_tokens": 461826444.0, + "step": 18162 + }, + { + "epoch": 1.9946189325719306, + "grad_norm": 2.2440130710601807, + "learning_rate": 1e-06, + "loss": 0.8972, + "mean_token_accuracy": 0.7221953868865967, + "num_tokens": 461850853.0, + "step": 18163 + }, + { + "epoch": 1.994728750274544, + "grad_norm": 2.446378707885742, + "learning_rate": 1e-06, + "loss": 0.8921, + "mean_token_accuracy": 0.7321572303771973, + "num_tokens": 461873535.0, + "step": 18164 + }, + { + "epoch": 1.9948385679771579, + "grad_norm": 2.538005828857422, + "learning_rate": 1e-06, + "loss": 0.8277, + "mean_token_accuracy": 0.7406048774719238, + "num_tokens": 461894417.0, + "step": 18165 + }, + { + "epoch": 1.9949483856797716, + "grad_norm": 2.488698959350586, + "learning_rate": 1e-06, + "loss": 0.9251, + "mean_token_accuracy": 0.7124887704849243, + "num_tokens": 461915893.0, + "step": 18166 + }, + { + "epoch": 1.9950582033823854, + "grad_norm": 2.1382415294647217, + "learning_rate": 1e-06, + "loss": 0.9218, + "mean_token_accuracy": 0.7179111242294312, + "num_tokens": 461942756.0, + "step": 18167 + }, + { + "epoch": 1.995168021084999, + "grad_norm": 2.1651482582092285, + "learning_rate": 1e-06, + "loss": 0.9231, + "mean_token_accuracy": 0.7169608473777771, + "num_tokens": 461969773.0, + "step": 18168 + }, + { + "epoch": 1.9952778387876124, + "grad_norm": 2.265407085418701, + "learning_rate": 1e-06, + "loss": 0.9567, + "mean_token_accuracy": 0.7049822807312012, + "num_tokens": 461996536.0, + "step": 18169 + }, + { + "epoch": 1.9953876564902262, + "grad_norm": 1.958056092262268, + "learning_rate": 1e-06, + "loss": 0.9398, + "mean_token_accuracy": 0.701830267906189, + "num_tokens": 462028313.0, + "step": 18170 + }, + { + "epoch": 1.99549747419284, + "grad_norm": 2.208054780960083, + "learning_rate": 1e-06, + "loss": 0.8274, + "mean_token_accuracy": 0.7376765012741089, + "num_tokens": 462054938.0, + "step": 18171 + }, + { + "epoch": 1.9956072918954535, + "grad_norm": 2.4510281085968018, + "learning_rate": 1e-06, + "loss": 0.8465, + "mean_token_accuracy": 0.7340112924575806, + "num_tokens": 462076015.0, + "step": 18172 + }, + { + "epoch": 1.9957171095980673, + "grad_norm": 2.5582053661346436, + "learning_rate": 1e-06, + "loss": 0.8302, + "mean_token_accuracy": 0.7368786931037903, + "num_tokens": 462097332.0, + "step": 18173 + }, + { + "epoch": 1.9958269273006808, + "grad_norm": 2.141967296600342, + "learning_rate": 1e-06, + "loss": 0.9021, + "mean_token_accuracy": 0.7176141738891602, + "num_tokens": 462124809.0, + "step": 18174 + }, + { + "epoch": 1.9959367450032945, + "grad_norm": 2.414776086807251, + "learning_rate": 1e-06, + "loss": 0.903, + "mean_token_accuracy": 0.7174227833747864, + "num_tokens": 462150723.0, + "step": 18175 + }, + { + "epoch": 1.9960465627059083, + "grad_norm": 2.096121311187744, + "learning_rate": 1e-06, + "loss": 0.9332, + "mean_token_accuracy": 0.7145928144454956, + "num_tokens": 462182137.0, + "step": 18176 + }, + { + "epoch": 1.9961563804085218, + "grad_norm": 2.367164134979248, + "learning_rate": 1e-06, + "loss": 0.8483, + "mean_token_accuracy": 0.735273003578186, + "num_tokens": 462205821.0, + "step": 18177 + }, + { + "epoch": 1.9962661981111354, + "grad_norm": 2.2017576694488525, + "learning_rate": 1e-06, + "loss": 0.8795, + "mean_token_accuracy": 0.7219457626342773, + "num_tokens": 462229807.0, + "step": 18178 + }, + { + "epoch": 1.9963760158137491, + "grad_norm": 2.070802927017212, + "learning_rate": 1e-06, + "loss": 1.0032, + "mean_token_accuracy": 0.6911471486091614, + "num_tokens": 462259171.0, + "step": 18179 + }, + { + "epoch": 1.9964858335163629, + "grad_norm": 2.1913933753967285, + "learning_rate": 1e-06, + "loss": 0.9308, + "mean_token_accuracy": 0.7137295007705688, + "num_tokens": 462285064.0, + "step": 18180 + }, + { + "epoch": 1.9965956512189766, + "grad_norm": 2.1174933910369873, + "learning_rate": 1e-06, + "loss": 0.9603, + "mean_token_accuracy": 0.6995425820350647, + "num_tokens": 462312945.0, + "step": 18181 + }, + { + "epoch": 1.9967054689215902, + "grad_norm": 2.0316812992095947, + "learning_rate": 1e-06, + "loss": 0.9352, + "mean_token_accuracy": 0.7085155248641968, + "num_tokens": 462341488.0, + "step": 18182 + }, + { + "epoch": 1.9968152866242037, + "grad_norm": 1.9956326484680176, + "learning_rate": 1e-06, + "loss": 0.8866, + "mean_token_accuracy": 0.730960488319397, + "num_tokens": 462371401.0, + "step": 18183 + }, + { + "epoch": 1.9969251043268175, + "grad_norm": 2.392441987991333, + "learning_rate": 1e-06, + "loss": 0.8907, + "mean_token_accuracy": 0.7187987565994263, + "num_tokens": 462393092.0, + "step": 18184 + }, + { + "epoch": 1.9970349220294312, + "grad_norm": 2.4921560287475586, + "learning_rate": 1e-06, + "loss": 0.9111, + "mean_token_accuracy": 0.7138452529907227, + "num_tokens": 462414757.0, + "step": 18185 + }, + { + "epoch": 1.9971447397320448, + "grad_norm": 2.115654230117798, + "learning_rate": 1e-06, + "loss": 0.9419, + "mean_token_accuracy": 0.7092539072036743, + "num_tokens": 462442442.0, + "step": 18186 + }, + { + "epoch": 1.9972545574346585, + "grad_norm": 2.0132720470428467, + "learning_rate": 1e-06, + "loss": 0.8942, + "mean_token_accuracy": 0.7231090068817139, + "num_tokens": 462474144.0, + "step": 18187 + }, + { + "epoch": 1.997364375137272, + "grad_norm": 2.6285297870635986, + "learning_rate": 1e-06, + "loss": 0.8666, + "mean_token_accuracy": 0.7320579290390015, + "num_tokens": 462494593.0, + "step": 18188 + }, + { + "epoch": 1.9974741928398858, + "grad_norm": 2.397376298904419, + "learning_rate": 1e-06, + "loss": 0.8618, + "mean_token_accuracy": 0.7334912419319153, + "num_tokens": 462518365.0, + "step": 18189 + }, + { + "epoch": 1.9975840105424996, + "grad_norm": 2.236994981765747, + "learning_rate": 1e-06, + "loss": 0.8725, + "mean_token_accuracy": 0.7280233502388, + "num_tokens": 462544615.0, + "step": 18190 + }, + { + "epoch": 1.997693828245113, + "grad_norm": 2.080369472503662, + "learning_rate": 1e-06, + "loss": 0.9745, + "mean_token_accuracy": 0.6987070441246033, + "num_tokens": 462573851.0, + "step": 18191 + }, + { + "epoch": 1.9978036459477266, + "grad_norm": 2.4844324588775635, + "learning_rate": 1e-06, + "loss": 0.9124, + "mean_token_accuracy": 0.7220171093940735, + "num_tokens": 462595020.0, + "step": 18192 + }, + { + "epoch": 1.9979134636503404, + "grad_norm": 2.1344058513641357, + "learning_rate": 1e-06, + "loss": 0.9558, + "mean_token_accuracy": 0.7045880556106567, + "num_tokens": 462623571.0, + "step": 18193 + }, + { + "epoch": 1.9980232813529542, + "grad_norm": 2.101468086242676, + "learning_rate": 1e-06, + "loss": 0.9832, + "mean_token_accuracy": 0.6915481090545654, + "num_tokens": 462653184.0, + "step": 18194 + }, + { + "epoch": 1.998133099055568, + "grad_norm": 2.2139954566955566, + "learning_rate": 1e-06, + "loss": 0.8673, + "mean_token_accuracy": 0.736561119556427, + "num_tokens": 462677974.0, + "step": 18195 + }, + { + "epoch": 1.9982429167581814, + "grad_norm": 2.0381240844726562, + "learning_rate": 1e-06, + "loss": 0.964, + "mean_token_accuracy": 0.7101449966430664, + "num_tokens": 462708036.0, + "step": 18196 + }, + { + "epoch": 1.998352734460795, + "grad_norm": 2.4479150772094727, + "learning_rate": 1e-06, + "loss": 0.936, + "mean_token_accuracy": 0.7099509239196777, + "num_tokens": 462731226.0, + "step": 18197 + }, + { + "epoch": 1.9984625521634087, + "grad_norm": 2.3899645805358887, + "learning_rate": 1e-06, + "loss": 0.7787, + "mean_token_accuracy": 0.7462434768676758, + "num_tokens": 462753789.0, + "step": 18198 + }, + { + "epoch": 1.9985723698660225, + "grad_norm": 2.3326010704040527, + "learning_rate": 1e-06, + "loss": 0.8556, + "mean_token_accuracy": 0.7340218424797058, + "num_tokens": 462777199.0, + "step": 18199 + }, + { + "epoch": 1.998682187568636, + "grad_norm": 2.521914005279541, + "learning_rate": 1e-06, + "loss": 0.924, + "mean_token_accuracy": 0.7136823534965515, + "num_tokens": 462798358.0, + "step": 18200 + }, + { + "epoch": 1.9987920052712496, + "grad_norm": 2.2609236240386963, + "learning_rate": 1e-06, + "loss": 0.8253, + "mean_token_accuracy": 0.7402776479721069, + "num_tokens": 462821760.0, + "step": 18201 + }, + { + "epoch": 1.9989018229738633, + "grad_norm": 2.1970300674438477, + "learning_rate": 1e-06, + "loss": 0.9817, + "mean_token_accuracy": 0.700217604637146, + "num_tokens": 462851442.0, + "step": 18202 + }, + { + "epoch": 1.999011640676477, + "grad_norm": 2.385692596435547, + "learning_rate": 1e-06, + "loss": 0.7556, + "mean_token_accuracy": 0.7563730478286743, + "num_tokens": 462873003.0, + "step": 18203 + }, + { + "epoch": 1.9991214583790908, + "grad_norm": 2.4782180786132812, + "learning_rate": 1e-06, + "loss": 0.7912, + "mean_token_accuracy": 0.7486187219619751, + "num_tokens": 462893722.0, + "step": 18204 + }, + { + "epoch": 1.9992312760817044, + "grad_norm": 2.2747809886932373, + "learning_rate": 1e-06, + "loss": 0.9363, + "mean_token_accuracy": 0.723507285118103, + "num_tokens": 462918057.0, + "step": 18205 + }, + { + "epoch": 1.999341093784318, + "grad_norm": 2.2300760746002197, + "learning_rate": 1e-06, + "loss": 0.9924, + "mean_token_accuracy": 0.7043567895889282, + "num_tokens": 462944982.0, + "step": 18206 + }, + { + "epoch": 1.9994509114869317, + "grad_norm": 2.404202461242676, + "learning_rate": 1e-06, + "loss": 0.8447, + "mean_token_accuracy": 0.7341800332069397, + "num_tokens": 462968223.0, + "step": 18207 + }, + { + "epoch": 1.9995607291895454, + "grad_norm": 2.1882715225219727, + "learning_rate": 1e-06, + "loss": 0.7993, + "mean_token_accuracy": 0.7504910230636597, + "num_tokens": 462992896.0, + "step": 18208 + }, + { + "epoch": 1.9996705468921592, + "grad_norm": 1.8805887699127197, + "learning_rate": 1e-06, + "loss": 0.9098, + "mean_token_accuracy": 0.7149724364280701, + "num_tokens": 463026885.0, + "step": 18209 + }, + { + "epoch": 1.9997803645947727, + "grad_norm": 2.475222110748291, + "learning_rate": 1e-06, + "loss": 0.83, + "mean_token_accuracy": 0.7360852956771851, + "num_tokens": 463048581.0, + "step": 18210 + }, + { + "epoch": 1.9998901822973862, + "grad_norm": 2.2953367233276367, + "learning_rate": 1e-06, + "loss": 0.8855, + "mean_token_accuracy": 0.7237787246704102, + "num_tokens": 463074792.0, + "step": 18211 + }, + { + "epoch": 2.0, + "grad_norm": 2.2120347023010254, + "learning_rate": 1e-06, + "loss": 0.7902, + "mean_token_accuracy": 0.7480226755142212, + "num_tokens": 463098646.0, + "step": 18212 + }, + { + "epoch": 2.0, + "step": 18212, + "total_flos": 2.0853122720147177e+19, + "train_loss": 0.9531466799229548, + "train_runtime": 21268.4838, + "train_samples_per_second": 13.7, + "train_steps_per_second": 0.856 + } + ], + "logging_steps": 1, + "max_steps": 18212, + "num_input_tokens_seen": 0, + "num_train_epochs": 2, + "save_steps": 9106, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 2.0853122720147177e+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..d29006c --- /dev/null +++ b/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:54bbdbd7702b3650af288937c898e357de58723f266e89fc3c39393dc7d45117 +size 13329